Final Report on Consumer Preferences for Cloningagecon.okstate.edu/faculty/publications/3221.pdf ·...
Transcript of Final Report on Consumer Preferences for Cloningagecon.okstate.edu/faculty/publications/3221.pdf ·...
Final Report on Consumer Preferences for Cloning
Submitted to:
U.S. Department of Agriculture
Economic Research Service
1800 M Street, NW
Washington, DC 20036-5831
Submitted by: Jayson L. Lusk
Professor and Willard Sparks Endowed Chair
Department of Agricultural Economics
Oklahoma State University
Phone: (405)744-7465
Email: [email protected]
Date:
October 1, 2008
Table of Contents
Content Page Executive Summary i
1. Introduction 1
2. Previous Research on Consumer Acceptance of Cloning 2
3. Study Objectives 3
4. Methods and Procedures 3
4.1 The Survey Samples 3
4.2 Experimental Treatment: Early Information 6
4.3 Experimental Treatment: Length of Information 6
4.4 The Survey and Statistical Methods 8
5. Results from the Randomly Recruited Sample 22
5.1 Awareness of Animal Cloning and Other Animal Breeding Techniques 22
5.2 Willingness to Eat Cloned Meat and Milk 22
5.3 Beliefs about Safety and Acceptability of Cloned Meat and Milk 23
5.4 Beliefs about Federal Government and Cloned Meat and Milk 23
5.5 Why Are People Concerned about Animal Cloning? 24
5.6 Preferences for Cloning Policies 27
5.7 Ground Beef Market Simulator 28
5.7 Milk Market Simulator 29
6. Comparing Results across Three Survey Samples 31
6.1 Sample Characteristics 31
6.2 Food Values 31
6.3 Awareness of Animal Cloning and Other Animal Breeding Techniques 32
6.4 Willingness to Eat Cloned Meat and Milk 32
6.5 Beliefs about Safety and Acceptability of Cloned Meat and Milk 33
6.6 Beliefs about Federal Government and Cloned Meat and Milk 33
6.7 Relative Importance of Competing Objections to Cloning 33
6.8 Relationship between Milk/Meat Purchases and 34
Willingness to Eat Cloned Food
6.9 Preferences for Cloning Policies 35
6.10 Ground Beef Choice-Based Conjoint Estimates 36
6.11 Milk Choice-Based Conjoint Estimates 36
7. Conclusions 37
8. References 39
9. Tables and Figures 41
i
Executive Summary
In the summer of 2008, three surveys were administered to over 6,000 U.S. consumers to gauge
the public‟s attitude toward the use of cloning in meat and milk production. One survey was
administered to a true probability-based random sample of 2,256 people, and two other surveys
were administered to “opt in” panels of consumers for which actual meat and milk purchase
behavior was known. A variety of survey methods and statistical techniques were used to
provide a comprehensive investigation of people‟s preferences for animal cloning. Unless
otherwise noted the results reported in the executive summary are from the probability-based
random sample. Key results from the study are as follows:
As compared to other assisted reproductive technologies, such as artificial insemination
and biotechnology, people have a higher level of awareness of animal cloning.
Providing people a week to digest information about cloning versus providing
information only at the time the survey was taken had virtually no impact on people‟s
attitudes toward cloning.
People provided a one-half page information statement about cloning including the
FDA‟s statement about the safety of cloned meat and milk, are somewhat less concerned
about cloning as compared to people who only received a two sentence definition of
cloning. For example, 32% of people from the true probability-based random sample that
were provided the lengthy information indicated that animal cloning was unacceptable,
whereas 40% respondents in one of the opt-in panels that were only provided a two
sentence definition indicated that animal cloning was unacceptable.
Attitudes toward cloning are neither overwhelmingly positive nor negative. About 31%
are willing to eat meat and drink milk from a cloned animal, about 43% are unwilling,
and 26% are neither willing nor unwilling. About 32% of the public believes animal
cloning is unacceptable, about 34% believes animal cloning is acceptable, and the
remaining 34% is neutral.
There is virtually no difference in people‟s willingness to drink cloned milk and eat
cloned meat. Furthermore, there is virtually no difference in people‟s willingness to eat
milk/meat from cloned animals and milk/meat from the offspring of cloned animals.
Although 64% believe that the meat and milk they buy is safe to eat, only 30% think the
U.S. government is doing everything it can to ensure the safety of food.
Less than 30% expressed trust in information about cloning from U.S. federal agencies.
People were relatively more trusting of information on cloning from University scientists,
the USDA, the FDA, and then the EPA.
Females and those with only a high school education are less supportive of eating
meat/milk from cloned animals than are males and those with a bachelor‟s degree or
higher level of education.
Actual purchase behavior, determined via at-home scans of grocery purchases, is related
to people‟s stated willingness to eat meat/milk from cloned animals obtained from the
opt-in panels. People who tend to primarily buy organic milk are significantly less
willing to drinking milk from cloned animals and are willing to pay more for “clone free”
milk than people who tend to primarily purchase non-organic milk. People who buy few
ii
packages of fresh beef throughout the year are less willing to eat meat from cloned
animals than people who are frequent purchasers of beef.
Of eight potential objections to cloning, the two most prominent are: (i) “cloning is
“unnatural” because it is not a process that occurs in nature” and (ii) “animal cloning will
lead to human cloning.” People are relatively unconcerned about the safety of eating
meat/milk from cloned animals or their offspring.
Although people are willing to pay relatively large premiums for non-cloned meat and
milk when grocery shopping (about $4 per purchase), this does not necessarily translate
into high willingness-to-pay for policies related to cloning. People did not support bans;
results from all three survey samples imply that willingness-to-pay for a ban on animal
cloning is effectively zero. There is mixed support for a policy related to mandatory
tracking of cloned animals; people provided lengthier information on animal cloning are
willing to pay much smaller amounts than people only given a two-sentence definition.
A majority of people in all three samples support a mandatory labeling policy on cloned
meat/milk, even if it causes a food price increase of 30%.
People‟s choices were used to construct a market simulator for fresh ground beef and a
market simulator for unflavored fluid milk. The simulators can be used to forecast
market shares of and willingness-to-pay for ground beef and milk options that differ
according to a variety of user-defined product characteristics including price, fat content,
and use of cloning. The market simulators can be accessed at:
http://agecon.okstate.edu/faculty/publications/3103.xls and
http://agecon.okstate.edu/faculty/publications/3104.xls. Market share simulations
indicate: o When faced with a choice between ground beef from a cloned and a non-cloned
animal, more than 75% of shoppers would choose the non-cloned beef even if
sold at a $1.50 price premium over ground beef from a cloned animal. o When faced with a choice between milk from a cloned and a non-cloned animal,
more than 65% of shoppers would choose the non-cloned milk even if sold at a
$1.50 price premium over milk from a cloned animal. o If faced with a choice of only being able to buy ground beef from a cloned animal
or no ground beef at all, the majority of consumers would choose to buy cloned
beef (assuming the price of ground beef was $3.50/lb or lower). At a price of
$2.00/lb., almost two-thirds of shoppers would buy ground beef from a cloned
animal if it were the only option available. o If faced with a choice of only being able to buy milk from a cloned animal at
$3.00/gallon or no milk at all, only about 47% of consumers would choose to buy
the milk from the cloned animal. o Although consumers are concerned about other product attributes, such as fat
content, price, and use of rBST, use of animal cloning is the most important
attribute to consumers.
1
1. Introduction
Farmers and ranchers have utilized assisted reproductive techniques for decades with little
controversy. For example, commercial embryo transfers have occurred in dairy and beef cattle
since the 1970‟s. Nevertheless, the relatively new reproductive technique of animal cloning has
sparked controversy. Animal cloning is a process by which scientists can copy the genetic or
inherited traits of an animal. Clones are similar to identical twins only born at different times.
Like other assisted reproductive breeding techniques, cloning is appealing to some ranchers and
farmers because it enables them to more quickly breed desirable traits into their herds. Such
improvements have the potential to lower the price and increase the quality of meat and milk
products. Despite these advantages, some consumer groups have expressed concern and even
outrage over the use of meat and milk from cloned animals and their offspring. The controversy
over cloning stems from moral and ethical objections to the technology, concerns about food
safety, and concerns about potential harm to the cloned animals and their surrogate mothers.
In January 2008, after years of detailed study and analysis, the U.S. Food and Drug
Administration (FDA) concluded that, “meat and milk from clones of cattle, swine, and goats,
and the offspring of clones from any species traditionally consumed as food, are as safe to eat as
food from conventionally bred animals.” The FDA‟s science-based risk assessment, which was
peer-reviewed by a group of independent scientific experts in cloning and animal health,
concluded:
1. Cloning poses no unique risks to animal health compared to the risks found with other
reproduction methods including natural mating.
2. The composition of food products from cattle, swine, and goat clones, or the offspring
of any animal clones, is no different from that of conventionally bred animals.
3. Because of the preceding two conclusions, there are no additional risks to people
eating food from cattle, swine, and goat clones or the offspring of any animal clones
traditionally consumed as food.
A copy of FDA‟s report can be found at: http://www.fda.govcvm/cloning.htm.
Despite this news, many consumers appear to be either unaware of the technology or find the
technology unacceptable. As such, important questions exist regarding consumer reaction to and
the market impacts of the presence of cloned meat and milk in the food supply chain. Indeed, in
late 2007 the U.S. Senate passed legislation intended to prohibit the FDA from approving cloned
products until further research is conducted.
These developments suggest the need for an in-depth study of consumer attitudes toward cloning.
For example, at present it is difficult to precisely identify consumers‟ potential objections to the
technology, but such information is needed to forecast how demand for meat and milk from
cloned animals will develop and respond to information and advertising campaigns. Further, it is
unclear whether, to what extent, and under what conditions consumers are willing to accept the
new technology. For example, how much lower priced or higher quality must cloned meat and
milk be to remain competitive with non-cloned meat and milk? Does the public desire stronger
regulation on the practice of animal cloning? Answering such questions with existing market
data is impossible, and as such, a consumer survey is needed to address these issues.
2
2. Previous Research on Consumer Acceptance of Cloning
Some previous opinion polls have been conducted on the issue of animal cloning, and in what
follows, some of the previous findings are briefly summarized. In a 2005 survey, Sosin and
Richards (2005) reported that 29% of consumers surveyed believed that animal cloning is
currently used by farmers and ranchers to breed animals whereas 64% believed that it would be
used in the future. Storey (2006) reported that about 50% of consumers surveyed believed that
use of animal cloning to breed animals for meat and milk production is possible. The survey
found that 73% of consumers had not heard about the FDA issuing a report on the use of animal
cloning to breed animals for meat and milk. A study conducted for the Pew Initiative on Food
and Biotechnology in 2004 found that about 65% of consumers had heard about animal cloning
(The Mellman Group, 2006).
The Pew study found that 29% of consumers indicated that they would purchase meat and milk
from the offspring of cloned animals, but 35% indicated that they would never buy meat and
33% would never buy milk from the offspring of cloned animals. This is similar to the study
conducted in 2006 by the International Food Information Council. IFIC (2006), which found that
about 41% of consumers said they would purchase meat, milk, or eggs from the offspring of
cloned animals. IFIC found that this latter figure increased slightly to 46% in 2007. Storey
(2006) found that approximately 33% of consumers indicated they would continue to purchase
meat and milk products if they came from cloned animals or their offspring.
Although these opinion polls indicate most consumers have heard about animal cloning, they are
somewhat uncomfortable with the technology. Storey (2006) found 32% of consumers felt
animal cloning was morally wrong and 26% were unsure of the safety of meat and milk from
clones and their offspring. The KRC Research study found that consumers found cloning more
acceptable if it improved the overall health of animals, improved nutrition of meat and milk, and
saved rare animal breeds (Sosin and Richards, 2005). The study also found that consumers are
more comfortable with animal cloning when told that it is carefully regulated by the FDA, the
USDA, and the EPA. Consumers are most likely to trust the USDA, FDA, veterinarians, and
physicians when it comes to animal cloning (Sosin and Richards, 2005). Consumers are less
likely to trust groups like the Consumer Federation of America and The Sierra Club when it
comes to animal cloning (Sosin and Richards, 2005).
Most of these previous opinion polls simply asked people to indicate purchase intentions or
attitudes on a five-point scale. A wealth of evidence indicates such data often poorly predicts
actual retail behavior (Morrison, 1979; Morwitz, 1997). Furthermore, such scales do not force
people to make trade-offs between concerns, and as such, it is common for people to rate many
issues as “very important.” This research will utilize “best-worst” or paired comparison
questions to determine the relative degree of concern for cloning and to determine which issues
are most unacceptable to consumers (see Lusk and Briggeman (2008) for a recent use of these
methods in the agricultural economics literature). Furthermore, this research will utilize
contingent valuation and conjoint questions to determine consumers‟ trade-offs between
concerns for cloning, food prices, and food quality in a manner that will permit estimation of
willingness-to-pay, demand elasticities, and market shares.
3
3. Study Objectives
The primary objectives of this study are to:
1) determine consumer awareness of and attitudes toward cloned meat and milk,
2) determine whether consumers‟ preferences for cloned meat and milk are affected by the
length of time they are given to digest unbiased, educational information about the
technology,
3) explore whether consumers‟ preferences for cloned meat and milk are affected by the
length of information given about cloning,
4) determine the effect of survey sample (i.e., opt-in panels with food purchase data vs.
randomly recruited representative sample) on preference for cloning.
5) determine the relationship between people‟s actual purchases of meat and milk and their
stated level of concern for cloning,
6) determine consumer willingness-to-pay for several policies related to animal cloning,
7) determine the trade-offs consumers are willing to make between use of cloning in meat
and milk production and other food attributes such as price and quality, and
8) develop market simulators to forecast market share and willingness-to-pay for ground
beef and milk products that differ according to a number of product characteristics
including use of cloning.
4. Methods and Procedures
4.1 The Survey Samples
A web-based survey was developed and administered by three sub-contractors, each of whom
utilize different recruitment methods and have their own relative advantages and disadvantages.
The first subcontractor, Knowledge Networks (KN) administered the survey to a sample that was
selected using random digit dialing techniques, and as such, represents a true probability sample
based on the general U.S. population. Although KN uses a web-based platform,
representativeness is ensured by providing randomly selected respondents with computers and
on-line access if they do not already have it. Although KN has the advantage of providing a true
probability based sample, they do not have data on consumers‟ actual food purchases. To
remedy this deficiency, identical surveys were administered by two additional sub-contractors,
Nielsen and IRI, which both maintain panels of consumers who record food purchasing history.
The advantage of these panels is that actual food purchases of the consumers are known, but the
disadvantage is that these panels are not true probability-based samples; they are comprised of
opt-in panels with people who volunteer to participate. By combining the results across all three
subcontractors, not only can it be determined whether sampling considerations affect results, but
a better and more in-depth picture of consumer preferences for cloning can be obtained.
It should be emphasized that because only KN uses a true random recruitment method, it is the
only sample which has a firm basis in statistically theory, and as such it is the only sample that
can legitimately be used to make inferences about the U.S. population at large. As such, we give
disproportionate attention to the results from the KN sample.
4
Despite the preceding argument, it should be noted that a random sample of the U.S. population
is not the same thing a random sample of meat eaters or milk drinkers, and as such, it is
worthwhile to study preferences for cloning in the IRI and Nielsen samples and investigate how
such preferences for cloning relate to actual meat and milk purchases.
A key factor that needs to be considered when comparing results across the three survey samples
is that the samples may differ in terms of the characteristics of the people surveyed. As such,
observed differences in concern for cloning across survey samples may be a result of differences
in factors such as race, gender, education, etc. and not a result of opt-in vs. random panels per se.
As such, after each survey was conducted, we created sample-specific, post-stratification weights
to reduce the effects of non-response and non-coverage bias. This was accomplished by
comparing the geographic location and demographic characteristics (age, race, gender, and
education) of each sample to the most recent data from the U.S. Census Bureau, Current
Population Survey. Iterative proportional fitting techniques were used to create weights which,
when applied to the data, force the proportions in each sample to match the proportions in the
population – at least in terms of geographic location, age, race, gender, and education. These
post-stratification weights are used in all results presented in this report.
4.1.1 Probability-Based Sample: Knowledge Networks
The company Knowledge Networks (KN) was contracted to administer a web-based survey to
their panel of respondents. Their panel is the only existing online panel that is representative of
the U.S. population. Probability sampling is used to recruit participants, avoiding well known
problems with “opt in” panels. The Knowledge Networks panel is representative of the U.S.
population, and to ensure representativeness, members of the panel are randomly recruited by
telephone (using both listed and unlisted numbers) and are provided with access to the Internet if
the household does not have ready availability. Thus, the panel is comprised of both Internet and
non-Internet households, all of which are provided the same equipment for participation in
Internet surveys. More information on the panel, recruitment methodology, studies comparing
the Knowledge Network panel to other sampling techniques, and a bibliography of published
academic papers which have employed the Knowledge Network panel can be found at
http://www.knowledgenetworks.com/ganp/.
In June 2008, the survey was sent to 3,222 individuals, 2,256 of whom completed at least a
portion of the questions, implying a response rate of 70%. This sample size implies a sampling
error of about 2.06%. That is, we can be 95% confident that the sampled percentage of people
falling in a particular category is within ± 2.06% of the true percentage of people in the particular
category in population. Table 1 (and table 18) reports the raw (unweighted) and weighted means
for selected socio-economic and demographic variables describing the sample of people
surveyed. As can be seen, the raw (original) sample matched the U.S. population quite well, and
as such, there are only slight differences in the weighted and unweighted means. By
construction, the weighted means for age, gender, race, and education exactly match the U.S.
population means.
5
4.1.2 Opt-In Panel with Scanner Data: Information Resources Inc
Information Resources Inc (IRI) was contracted to provide their so-called AttitudeLink™ panel
survey service, which enabled the linking of our survey with IRI‟s U.S. household panel, a group
of U.S. households who use a handheld scanner to report their bar-coded product purchases.
The IRI panel is designed to be demographically representative at the total U.S. In addition to
the answers to the survey questions, IRI is able to provide, for each household who completed a
survey, data on purchases of milk (organic and non-organic), breakfast meat, and lunch meat.
Importantly, these measures are not based on consumers‟ potentially unreliable memories of past
purchasing behavior, but instead represent actual purchase histories. IRI was unable to provide
data on purchases of fresh meat products because such products often to not have a bar code that
specifies package weight. IRI does not possess information on such “random weight” products.
In late June, early July 2008, the survey was sent to 4,000 individuals, 1,691 of whom completed
the survey questions, implying a response rate of 42.3%. This sample size implies a sampling
error of about 2.38%. That is, we can be 95% confident that the sampled percentage of people
falling in a particular category is within ± 2.38% of the true percentage of people in the particular
category in population. Table 18 reports the raw (unweighted) means for selected socio-
economic and demographic variables describing the sample of people surveyed and compares
them to the U.S. Census data. The IRI sample has a much lower level of education and has a
high share of females as compared to the U.S. population. By construction, after the weights are
applied to the data, then the sample proportions for the various age, gender, race, education, and
location categories exactly match the U.S. population proportions.
4.1.3 Opt-In Panel with Scanner Data: The Nielsen Company
Nielsen was contracted to administer the survey to their so-called Homescan panel. The
Homescan panel is comprised of households who “opt in” or volunteer to participate in the panel.
Panelists agree to scan the barcodes on their purchases and to occasionally complete surveys.
Although panelists “opt in,” Nielsen attempts to select volunteering households such that the
panel is representative of the U.S. population. In addition to answers to the survey questions,
Nielsen is able to provide, for each household who completed a survey, data on units purchased
for milk and beef. Because beef is a random weight item, we only know units purchased and not
the total pounds purchased. Like the IRI sample, this purchase data is not based on consumers‟
memories of past purchasing behavior, but instead represent actual purchase histories.
In late July and early August 2008, the survey was sent to 2,992 individuals, 2,120 of whom
completed the survey questions, implying a response rate of 70.9%. This sample size implies a
sampling error of about 2.13%. That is, we can be 95% confident that the sampled percentage of
people falling in a particular category is within ± 2.13% of the true percentage of people in the
particular category in population. Table 18 reports the raw (unweighted) means for selected
socio-economic and demographic variables describing the sample of people surveyed as
compared to the U.S. Census data. By construction, after the weights are applied to the data,
then the sample proportions for the various age, gender, race, education, and location categories
exactly match the U.S. population proportions.
6
4.2 Experimental Treatment: Early Information
There are several concerns with asking people about their preferences for a new and
controversial technology for which they are likely unfamiliar, including: (i) people are unlikely
to provide well reasoned or consistent answers, (ii) people‟s responses may largely be driven by
a “shock value” or “gut reaction,” and (iii) people‟s answers may be unduly influenced or easily
swayed by the information presented. Because of these concerns, it is of interest to determine
whether giving people some period of time (one week in our case) between the provision of
information about cloning and the administration of the survey substantively influences behavior.
To investigate this issue, one half of the people in the KN sample were sent an email requesting
that they read a brief information statement about cloning. Participants were directed to a web
page, where they had to click a button to proceed. The “proceed” button was added in an attempt
to ensure the information was actually read by each participant. Then, one week later, the same
subjects were sent another email directing them to complete the survey. In the content of the
survey, the information was provided again. For the other half of the participants in the KN
sample, they only received the information at the time when they took the survey.
By comparing responses across these two treatments in the KN sample, we can identify whether
the early provision of information influenced people‟s attitudes toward cloned meat and milk.
Importantly, this is a test of the effect of the timing of information, not an effect of information
per se. All participants received the exact same information statement regardless of the
treatment to which they were randomly assigned. In the content of the survey, the information
was provided on the third screen; the only questions preceding the information statement were in
regards to people‟s knowledge and awareness of several animal breeding techniques including
cloning and questions aimed at measuring people‟s general food values.
4.3 Experimental Treatment: Length of Information
Because cloning is likely a topic unfamiliar to many survey respondents, people must be
provided some information. In developing an information statement, a difficult balance must be
struck between providing a concise yet informative discussion of the issues. People cannot be
expected to spend a great deal of time reading a wealth of information on the topic. For the KN
sample, we chose to construct an information statement by surveying the literature on cloning
and selecting the key issues (pro and con) to presented to subjects in as neutral a way as possible.
The information statement focused primarily on the information provided by the FDA because it
is a source most people would likely find credible and because it is information that has already
been widely disseminated in the press.
The precise information statement given to subjects in the KN and the Nielsen samples was:
Animal cloning is a process in which scientists can copy the genetic or inherited traits of
an animal. Clones are similar to identical twins only born at different times. Similar to in
vitro fertilization, cloned animals begin in a laboratory, but then are born to surrogate
mothers in the usual way and grow up just like other animals.
7
This reproductive breeding technique is appealing to some ranchers and farmers
because it enables them to create “identical twins” of their best breeding stock – allowing
them to more quickly breed desirable traits into herds. The technique is also appealing to
some consumers because it has the potential to lower the price and increase the quality of
meat and milk.
This reproductive breeding technique is opposed by some people on moral and
ethical grounds. Other people are opposed to animal cloning because, given current
technology, only a small percentage of attempts at cloning are successful and many of the
clones die during all stages of gestation and birth and the procedures may carry risks for
the mother. Although these symptoms are a downside to cloning, they are not necessarily
unique to cloning in comparison to other reproductive techniques.
In January 2008, after years of detailed study and analysis, the U.S. Food and
Drug Administration (FDA) concluded that, “meat and milk from clones of cattle, swine,
and goats, and the offspring of clones from any species traditionally consumed as food,
are as safe to eat as food from conventionally bred animals.” The FDA‟s science-based
risk assessment, which was peer-reviewed by a group of independent scientific experts in
cloning and animal health, concluded:
1. Cloning poses no unique risks to animal health compared to the risks found
with other reproduction methods including natural mating.
2. The composition of food products from cattle, swine, and goat clones, or the
offspring of any animal clones, is no different from that of conventionally
bred animals.
3. Because of the preceding two conclusions, there are no additional risks to
people eating food from cattle, swine, and goat clones or the offspring of any
animal clones traditionally consumed as food.
A copy of FDA‟s report can be found at: http://www.fda.gov/cvm/cloning.htm.
Recall that one-half the participants in the KN sample received this information statement one
week prior to taking the survey, and all KN participants, regardless of treatment, also received
the information statement in the third section of the survey.
Of course, there are some situations in which it would be desirable to utilize an even more
concise information statement. If the sole purpose of this project was to forecasting purchase
behavior, a very brief information statement would be appropriate as it would only give people
that information which they normally have when shopping. However, in the current context,
where inferences are being made about welfare effects of various policies and where elicited
values may be used in cost/benefit analysis or other policy-related analysis, then it is prudent that
people should reasonably informed prior to their attitudes/preferences being elicited. This view
is consistent with the standard recommendations on survey design for contingent valuation
including that of the “blue ribbon” NOAA panel (e.g., Arrow et al., 1993; Boyle, 2004). For this
reason, the true probability-based sample (the KN panel) was given the more complete
information statement shown above.
Recognizing the interest in how such information may affect behavior, we chose to present two
different information statements to the two “opt in” panels, IRI and Nielsen. For the Nielsen
panel we gave respondents, at the time of the survey, the exact information statement shown
8
above – the exact same statement shown to the KN sample. For the IRI sample, we restricted the
information statement to two sentences only providing definitional information. The precise
information statement given to subjects in the IRI sample was:
Animal cloning is a process in which scientists can copy the genetic or inherited traits of
an animal. Cloned animals are similar to identical twins only born at different times.
One question that might arise is whether the approach followed in this research can clearly
identify the effects of (i) early vs. simultaneous provision of information, (ii) long vs. short
information, and (iii) random vs. opt in recruitment method. Under the maintained assumption
that the Nielsen and IRI samples are essentially equivalent, the table below indicates that each of
these effects is indeed identified. Testing for the effect of early vs. simultaneous provision of
information involves a comparison of cell A. to cell B., which holds the recruitment method
constant at “random” and the information length constant at “long.” The effect of long vs. short
information can be tested by comparing cell C. to cell D., which holds the recruitment method
constant at “opt in” and timing of information constant at “simultaneous.” Finally, the effect of
random vs. opt in recruitment method can be tested by comparing cell B. to cell C., which holds
information constant at “long” and “simultaneous.”
Information
Recruitment Method Long,
Early
Long,
Simultaneous
Short,
Simultaneous
Random A. KN
sample 1
B. KN
sample 2
Opt in C. Nielsen
sample
D. IRI
sample
4.4 The Survey and Statistical Methods
As previously indicated, a web-based survey was developed. The survey consisted of nine
sections. Although the nine sections appeared in the same order for all surveys, individual items
within a section were randomly ordered across surveys to prevent an order effect. Except for the
differences in information described above, the surveys administered by all three subcontractors
were identical. What follows is a description of the questions asked in each survey section along
with details on statistical techniques employed to make use of the data.
4.4.1 Food Values
The first section of the survey was designed to measure people‟s general food values. Such
information is helpful in identifying why people find certain food technologies more or less
acceptable. The idea is to identify a set of a values that relate specifically to people‟s food
choices that are perhaps more stable than people‟s preference ranking of a specific set of foods or
food attributes such as cloning. Following Lusk and Briggeman (2008), we selected 11 food
values that are reasonably comprehensive in covering the breadth of issues likely to motivate
consumer food choice. The eleven food values and the associated descriptions are:
9
1. Naturalness (extent to which food is produced without modern technologies)
2. Taste (extent to which consumption of the food is appealing to the senses)
3. Price (the price that is paid for the food)
4. Safety (extent to which consumption of food will not cause illness)
5. Convenience (ease with which food is cooked and/or consumed)
6. Nutrition (amount and type of fat, protein, vitamins, etc.)
7. Tradition (preserving traditional consumption patterns)
8. Origin (where the agricultural commodities were grown)
9. Fairness (the extent to which all parties involved in the production of the food equally
benefit)
10. Appearance (extent to which food looks appealing)
11. Environmental Impact (effect of food production on the environment)
Asking people to rate the importance of each food value on a 5-point or 7-point scale is
problematic because: (i) people are likely to rate several issues as “very important” because no
tradeoffs must be made and (ii) people do not always use measurement scales identically,
causing difficulties in inter-personal and cross-cultural comparisons (Steenkamp and
Baumgartner 1998). As such, we utilized a paired comparison approach, which has been popular
since its introduction by Thurstone‟s (1927) and has been increasing in popularity due to the rise
of “best-worst” scaling techniques, e.g., see Finn and Louviere (1992), Marley and Louviere
(2005), and Flynn et al. (2007). This approach permits the measurement of food values on a
ratio scale by observing people‟s choices of which values are picked as more important.
To implement the approach, people were asked eleven repeated questions of the form, “Is X or Y
more important when you purchase foods?,” where the two food issues X and Y were randomly
chosen from the list of 11 issues shown above. For example, one question might read, “Is
Naturalness (extent to which food is produced without modern technologies) or Taste (extent to
which consumption of the food is appealing to the senses) more important when you purchase
foods?”
In total, there are (11*11-11)/2=55 possible questions that that can be created by pairings of the
issues listed above. Each person randomly received 11 of these pairings and made discrete
choices of which values were deemed more important. When responding to each discrete choice
question, people can be conceptualized as choosing the item that is highest on an underlying
scale of importance. Formally, let αj represent the location of value j on the underlying scale of
importance, and let the true or latent unobserved level of importance for individual i be given by
Iij = αj + εij, where εij is a random error term. The probability that the consumer chooses, say,
item j over item k, as most important is the probability that Iij is greater than Iik. If the εij are
distributed iid type I extreme value, then this probability takes the familiar logit form.
In particular, in each choice set, an individual chose whether issue j or issue k was more
important. The probability that issue j is more important than issue k is:
(1) Prob[issue j is more important than issue k] = kj
j
ee
e
where αj and αk are parameters identifying the relative importance of issue j and issue k, and β is
an overall constant term that corresponds to an order effect (i.e., the propensity to choose the
10
food value presented first in the pairing). In a sample of N individuals making C choices, with
each choice involving a differing pairing of food values, the log-likelihood function is
(2) N
i
C
c
ijcijckj
k
kj
j
ee
ey
ee
eyLogL
1 1
ln)1(ln
where yijc = 1 if issue j is chosen by individual i as most important in choice set c, and where yijc
= 0 if issue k is chosen by individual i as most important in choice set c. In this framework, one
of the 11 parameters must be normalized to zero for identification purposes, and as such we
arbitrarily selected the value “environmental impact” and normalized the parameter to zero such
that the estimated effect of the other issues can be interpreted as the importance of the particular
value relative to the importance of environmental impact.
To ease interpretation and to provide a measurement of the importance of food values on a ratio
scale, the parameters obtained from equation (2) are substituted into the multinomial logit
formula to calculate “shares of preference” or “importance scores” which indicate, of the 11
issues, the percentage of people that would choose issue j as most important as shown in
equation 3.
(3) Importance Score = Share of people believing issue j is most important = J
k
k
j
e
e
1
.
One of the primary objectives of asking these survey questions is to determine how values relate
to preferences for cloning, and as such, we need information on each individual‟s values.
Unfortunately, the logit described in equation (1) assumes that all individuals in the sample place
the same level of importance on each value (i.e., there is no i subscript on αj). To overcome this
weakness of the logit, we also estimated a random parameters logit model (RPL). In particular,
let the importance parameter for individual i and issue j be specified as ijjjij~ , where
j and ζj are the mean and standard deviation of αj in the population, and μi is a random term
normally distributed with mean zero and unit standard deviation. Such a specification implies
that the importance of food value j is assumed to be distributed according to a normal
distribution with mean j and standard deviation ζj. Substituting ijjjij~ into equation
(1) yields a probability statement that depends on the random term in μij. Rather than attempting
to explicitly integrate over these random terms, following Train (2003), the model was estimated
via simulation. In particular, the parameters were estimated by maximizing a simulated log-
likelihood function, evaluated at 100 pseudo-random Halton draws for μij. The random draws
are individual-specific, which takes into consideration the panel nature of the data resulting from
the fact that each person answered 11 repeated choice questions. Train (2003) provides more
information on computational details for the RPL.
As shown by Train (2003) and Huber and Train (2001), once the parameters from the RPL are
estimated, so-called individual-specific estimates can be obtained by using the estimated
parameters as a prior and using each person‟s actual choices to form an individual-specific
posterior estimate. These posterior estimates of food values for each individual can be compared
with each person‟s preferences for cloning to determine if relationships among the variables exist.
11
4.4.2 Awareness of Animal Breeding Techniques
The second section of the survey included five simple questions designed to gauge people‟s
knowledge and awareness of five assisted reproduction technologies that are sometimes used to
breed animals for meat and milk production: artificial insemination, in vitro fertilization,
biotechnology, embryo transfer, and cloning. People were asked, “Overall, how much have you
heard or read about each of the following assisted reproduction technologies that are sometimes
used to breed animals for meat and milk production?” Response categories were: 1 = nothing at
all, 2 = a little, 3 = a moderate amount, 4 = quite a bit, and 5 = a great deal. An exact copy of the
question, as presented in the KN web survey, is shown below.
4.4.3 Information Statement
Information about animal cloning was provided to respondents in the third survey section. As
indicated in section 4.3 above, the KN and Nielsen samples were given an information statement
conveying the key issues (pro and con) on cloning, using information primarily provided by the
FDA. Recall that one-half the participants in the KN sample received this information statement
12
one week prior to taking the survey, and all participants, regardless of treatment, also received
the information statement in the third section of the survey. The IRI sample was simply given a
two-sentence definition of cloning in this section of the survey.
4.4.4 Likert Scale Questions related to Animal Cloning
In section four, respondents were asked to indicate the extent to which they agreed or disagreed
with 20 statements related to animal cloning and government involvement in animal cloning.
Examples of statements appearing in this section included:
“I am willing to eat meat from cloned animals,”
“I am willing to consume milk products from cloned animals,” and
“I trust the U.S. government to properly regulate the use of animal cloning.”
People were asked to respond to each statement on a five-point scale: 1=strongly disagree, 2=
somewhat disagree, 3=neither agree nor disagree, 4= somewhat agree, and 5=strongly agree.
Each of the 20 statements was randomly ordered across surveys.
4.4.5 Preferences for Cloning Policies Measured by Contingent Valuation
Given the controversy surrounding animal cloning, several interest groups have proposed
policies to regulate animal cloning. To gauge the public‟s interest in such policies, three
contingent valuation questions were asked. One question focused on a mandatory tracking
system for cloned animals, another focused on a mandatory labeling system for meat and milk
from cloned animals, and the final question focused on a ban on the practice of animal cloning.
Although people answered three questions, they were asked to assume that the particular policy
in question was the only one on the ballot and were asked to answer each of the three questions
individually assuming only one policy option was under consideration. The three questions were
randomly ordered across surveys.
Following the convention in contingent valuation literature, and the suggestions of the NOAA
panel, the questions were posed as dichotomous choice referendum questions. Rather than using
federal taxes as the payment vehicle, as is typically the practice in contingent valuation, we felt it
would be less objectionable to respondents and more straightforward to use an associated
increase in food prices as the payment vehicle.
The exact phrasing of the three contingent valuation questions was as follows:
1. Suppose the next time you went to vote, there was a referendum on the ballot that would
require the U.S. government to implement a policy that required a tracking system on all
cloned animals. Would you vote in favor of this policy if the policy would increase the
price you would pay for meat and milk products by X% due to the added enforcement
and oversight required by the policy?
13
2. Suppose the next time you went to vote, there was a referendum on the ballot that would
require firms to place a label on all meat and milk products derived from cloned animals
or the offspring of cloned animals. Would you vote in favor of this policy if the policy
would increase the price you would pay for meat and milk products by Y% due to the
added enforcement and oversight required by the policy?
3. Suppose the next time you went to vote, there was a referendum on the ballot that would
ban the practice of animal cloning altogether. Would you vote in favor of this policy if
the policy would increase the price you would pay for meat and milk products by Z% due
to the added enforcement and oversight required by the policy?
Response categories were of the form: 1 = I would vote in favor of <<policy>> and a X%
increase in the price of meat and milk, and 0 = I would vote against <<policy>> and the X%
increase in the price of meat and milk.
An exact screen-shot of one of the questions administered by KN is shown below.
The percentage price increases, X, Y, and Z, were randomly chosen for each individual among
the values of 5%, 10%, 15%, 25%, 50%, 75%, and 100%. The literature on experimental design
suggests the most efficient design (in terms of generating the smallest standard errors on
willingness-to-pay) is one in which the price levels are centered on “true” willingness-to-pay. Of
course, we do not know the “true” willingness-to-pay. Our a priori expectations was that the
average willingness-to-pay values was likely to be in the range of a 10-20% price increase, and
14
as such, we chose to vary several price levels around this range (5%, 10%, 15%, 25%). However,
given the uncertainty in the prior expectation, and the belief that it might take much higher dollar
amounts for people to be indifferent to the policies, we also chose three higher price ranges (50%,
75%, and 100%).
People‟s choices can be analyzed in a random utility framework, where it is assumed that people
choose the option (either vote in favor or against) that maximizes utility. Because people‟s
preferences are not perfectly observed, we make use of the random utility framework
popularized by McFadden (1973) and applied to contingent valuation by Hanemann (1984). In
particular, assuming the errors in the random utility function are logistically distributed, the
probability that a person votes in favor of the policy is given by
(4) Prob[individual i votes in favor of policy j] = ijjj
ijjj
P
P
e
e
1
where, Pij is the percentage food price increase randomly assigned to individual i and policy j.
As shown by Hanemann (1984), λj represents the utility difference between having and not
having the outcome provided by policy j, and δj is the marginal utility of income. Hanemann
(1984) shows that mean willingness-to-pay for policy j is given by -λj/δj. In this simple
framework, this value is the price increase that makes people indifferent to having and not
having the policy, i.e., the price increase that sets the probability statement in (4) to 0.50. The
parameters shown in equation (4) are easily estimated by maximum likelihood estimation using
data on people‟s discrete choices of whether they voted in favor or against each policy at a given
price level.
4.4.6 Relative Importance of Competing Objections to Animal Cloning
In this section of the survey, we once again employed the paired-comparison method to tease out
the motivations behind people‟s underlying concerns about animal cloning. Again, the
advantage of this approach (over, say, responses to simple Likert scale questions) is that people
are forced the indicate their relative degree of concern (i.e., not all issues can be most important),
making inter-personal comparisons is less problematic (i.e., there is only one way to make a
choice), and the measured levels of concern can be easily stated on a ratio scale.
To begin this section of the survey, people were told the following.
Some people are in favor of animal cloning and some people object to the practice. We
are interested in your opinions about a few of the objections that some people have about
animal cloning. For each of the following questions, please indicate which of the two
statements best describes your views toward animal cloning. We recognize that, in some
cases, you may not particularly agree with either statement; however, please choose
which of the two statements best matches your views.
Then, people were asked eight repeated questions of the form, “Which of the following two
statements best describes your views toward animal cloning? X or Y”
The two statements X and Y were randomly selected from the list of 8 issues below:
1. Animal cloning is morally wrong
2. Meat and milk from clones and their offspring is unsafe to eat
15
3. Animal cloning will lead to human cloning
4. Cloning will result in unhealthy farm animals
5. Cloning is “unnatural” because it is not a process that occurs in nature
6. Cloning will reduce genetic diversity to an unacceptable level
7. Cloning results in animals being viewed as “objects‟ to be produced as opposed to being
valuable in and of themselves
8. The scientists and biotechnology companies who developed cloning technology cannot
be trusted to look out for my best interest.
For example, one question might have been, “Which of the following two statements best
describes your views toward animal cloning? „Animal cloning is morally wrong‟ or „Meat and
milk from clones and their offspring is unsafe to eat.‟”
In total, there are (8*8-8)/2=28 possible questions that that can be created representing all
possible pairs of the issues listed above. Each person randomly received 8 of these pairings and
made discrete choices of which statement best described their view toward animal cloning.
To determine the relative importance of each of the 8 issues, a logistic regression was estimated.
The procedures used to estimate the model are exactly the same as that described in section 4.4.1
Food Values, and as such we refrain from repeating the information in here. As with the food
values, we are able to calculate “importance scores” and derive individual-specific estimates of
the importance of each of the cloning issues by estimating a random parameter logit model.
16
4.4.7 Ground Beef Conjoint Questions
In section 7 of the survey, people were asked to answer a series of discrete choice questions
regarding which ground beef option (or none) they would buy when grocery shopping. In
selecting a meat product to choose as the focal product for the conjoint exercise, ground beef was
chosen because: (i) the FDA report on cloning focused on cloning in cattle, swine, and goats (not
poultry), (ii) of these three meat types, per-capita consumption of beef is the highest, and (iii) in
the beef category, ground beef represents the vast majority of products consumed, and this holds
for all income levels, at home, and away from home (Davis and Lin, 2005). Thus, we chose the
most popular species (beef) and the most popular type of beef consumed (ground beef).
To construct the choice scenarios, each ground beef purchase option was described by four
different attributes:
1. Price per Pound
a. $1.99/lb
b. $3.99/lb
2. Percent Lean
a. 80%
b. 90%
3. Percent Saturated Fat
a. 5%
b. 10%
4. Use of Cloning
a. Meat from non-cloned animal
b. Meat from cloned animal
c. Meat from offspring of cloned animal.
The purpose of including several additional attributes other than price and cloning was: (i) to
present realistic choice options to consumers like the ones they would encounter in the
supermarket, and (ii) to determine the importance of cloning relative to these other attributes.
The first three attributes, price, percent lean, and percent saturated fat, were varied at two levels
each. The last attribute (use of cloning) was treated as an alternative-specific attribute, such that
option A was always “meat from non-cloned animal,” option B, was always “meat from cloned
animal,” and option C was always “meat from offspring of cloned animal.” To these three
options, we also added an option D which allowed people to “opt out” or indicate “no purchase.”
To determine which ground beef options to present to respondents, a main effects fractional
factorial design was utilized. In particular, there are three attributes being varied at two levels
each, meaning that there are 23 possible combinations of ground beef options that could be
created for each choice option A, B, and C. Because there are three ground beef options, there
are 23 x 2
3 x 2
3 possible choice sets that could be presented to people. From this full factorial, 12
choices tasks were selected such that the correlations between attributes, both within and across
options, are exactly zero. An example of one of the 12 choices presented to subjects in the KN
sample is shown below.
17
Responses to the choice questions can be analyzed using the random utility framework of
McFadden (1973), where the systematic portion of the utility function is assumed to depend on
the attributes of the choice option. In addition to this systematic portion, the utility function is
assumed to contain a stochastic error term representing the fact that the analyst cannot observe
people‟s preferences with certainty. It is assumed that the consumer chooses the option that
generates the highest utility given available choice options and constraints. More formally, a
random utility function is defined by a deterministic (Vij) and a stochastic ( ij) component:
(5) ijijij VU
where Uij is the ith
consumer‟s utility of choosing option j, Vij is the systematic portion of the
utility function determined by ground beef attributes in alternative j, and ij is a stochastic
element. The probability that a consumer chooses alternative j from a choice set with J total
choice options is
(6) } allfor Prob{ jkVV ikikijij .
If the random errors in equation (5) are independently and identically distributed across the J
alternatives and N individuals with a type I extreme value distribution, McFadden (1973), shows
that the probability of alternative j being chosen is the familiar multinomial logit model (MNL),
18
(7) J
k
V
V
ik
ij
e
ej
1
chosen} is n Prob{optio .
In this application, the consumers‟ utility function for alternative j is assumed to be a linear
function of total amount of fat, type of fat, price, and use of cloning:
(8) Vj = β1 (% lean)j + β2 (% saturated fat)j + β3 (non-clone) + β4(clone) +
β5(clone off spring) j + β6 (price)j
where βk represent marginal utilities of each of the attributes. For identification purposes, the
utility of the “none” option is normalized to zero. Given this normalization, β3, β4, and β5,
represent the utility of having a package of ground beef from a non-cloned, cloned, and offspring
of cloned animal, respectively, relative to not purchasing ground beef at all on the particular
shopping occasion.
Once the parameters in equation (8) are estimated via the maximization of a log-likelihood
function based on the probability statement in equation (7), a variety of useful statistics can be
calculated. For example, market shares can be readily calculated simply by substituting the
estimated parameters back into equation (7) along with the characteristics of the choice options
assumed to be presented to consumers. Given certainty about which option is chosen, simple
willingness-to-pay estimates can be also calculated. For example, willingness-to-pay to have a
particular ground beef choice option relative to “none” is simply: -[β1 (% lean)j + β2 (% saturated
fat)j + β3 (non-clone) + β4(clone) + β5(clone off spring) j]/β6. This is the price of the option that
would make people indifferent to choosing the particular ground beef option and choosing none.
Similarly, the premium someone would be willing to pay for meat from a non-cloned animal
relative to a cloned animal is given by: -[β3 - β4]/β6. This is the price difference would make
people indifferent to choosing a package of ground beef from a non-cloned animal and a package
of ground beef from a cloned animal, assuming all other attributes were the same. Finally, one
can estimate a person‟s willingness-to-pay to keep an option in the choice set. Following Small
and Rosen (1978), this statistic is determined by identifying the maximum expected utility from
a choice set with the option and the maximum expected utility of a choice set without the option,
scaled by the marginal utility of income. Given the assumption about the distribution of the error
term, this statistic is calculated as:
(9) {willingness-to-pay for option k = 1} = J
k
VJ
k
V kk eeβ
-
216
lnln1
.
One drawback to using the MNL outlined above is that it assumes (i) people are homogeneous
and (ii) the independence of irrelevant alternatives. This second assumption implies that the
ratio of market shares between any two options are unaffected by the presence or characteristics
of a third option. To relax these two assumptions, we also estimated a random parameter logit
model.
The random parameter logit can be characterized by letting β represent a vector of all non-price
attributes, such that individual i‟s utility parameters are given by ii , where and ζ
are vectors of the means and standard deviations of β in the population, and μi is a vector of
random terms normally distributed with mean zero and unit standard deviation. Substituting
19
ii into equation (7) yields a probability statement that depends on the random term,
μi:
(10) J
k
V
V
i
ik
ij
e
euj
1
}|chosen is n Prob{optio
Rather than attempting to explicitly integrate over the random terms μi, following Train (2003),
the model was estimated via simulation. The parameters were estimated by maximizing a
simulated log-likelihood function, evaluated at 100 pseudo-random Halton draws for μi. The
random draws are individual-specific, which takes into consideration the panel nature of the data
resulting from the fact that each person answered 12 repeated choice questions. Train (2003)
provides more information on computational details for the RPL. As with the MNL, mean WTP
in the RPL for meat from a non-cloned animal relative to a cloned animal (assuming a constant
price effect) is 643 /][ . The standard deviation of this WTP measure in the population
(again assuming a non-random price effect) is: )()/1( 2
4
2
3
2
6 , where 2
3 and 2
4 are the
variances of β3 and β4, respectively.
4.4.8 Milk Conjoint Questions
Section 8 of the survey was similar to the previous section except, people were asked to answer a
series of discrete choice questions regarding which milk option (or none) they would buy when
grocery shopping. In choosing a dairy product for use in the conjoint questions, plain fluid milk
was chosen because it represents the largest category of dairy purchases. Total consumption of
fluid plain milk was about 116 lbs/person/year in 2005, whereas consumption of cheese
(American, Italian, and miscellaneous) was only 23.8 lbs/person/year, and frozen dairy products
was only 17 lbs/person/year, and yogurt was only about 6 lbs/person/year (USDA-ERS, 2008).
To construct the choice scenarios, each milk option was described by four different attributes:
1. Price per Gallon
a. $2.99/lb
b. $5.99/lb
2. Fat Content
a. Whole
b. 2%
c. 1%
d. Skim
3. Use of rBST
a. No rBST used
b. rBST used
4. Use of Cloning
a. Milk from non-cloned animal
b. Milk from cloned animal
c. Milk from offspring of cloned animal.
20
As in the beef conjoint task, the last attribute (use of cloning) was treated as an alternative-
specific attribute, such that option A was always “milk from non-cloned animal,” option B, was
always “milk from cloned animal,” and option C was always “milk from offspring of cloned
animal.” To these three options, we also added an option D which allowed people to “opt out” or
indicate “no purchase.” To determine which milk options to present to respondents, a main
effects fractional factorial design was utilized. In particular, there are two attributes being varied
at two levels each (price and rBST use) and one attribute varied at four levels (fat content),
meaning that there are 22 x 4 possible combinations of milk options that could be created for
each choice option A, B, and C. Because there are three milk options, there are 22 x 4 x 2
2 x 4 x
22 x 4 possible choice sets that could be presented to people. From this full factorial, 16 choices
tasks were selected such that the correlations between attributes, both within and across options,
are exactly zero. An example of one of the 16 choice tasks presented to subjects in the KN
sample is shown below.
The responses to the choice questions can be analyzed in exactly the same fashion as the beef
conjoint questions as described in section 4.4.7 Ground Beef Conjoint Questions, and as such we
refrain from repeating the information in this section.
21
4.4.9 Demographic Information
The last section of the survey included a few demographic questions. Because KN, IRI, and
Nielsen each maintain a panel of respondents, they are already in possession of information on
each respondent‟s, age, income, education, etc. Thus, there was no need to ask such information.
The last section asked a few questions germane to the present study including questions about
whether the respondent had any ties to farming, questions on people‟s consumption of beef and
milk, and a question regarding whether the respondent was the primary shopper of food in their
household.
22
5. Results from the Randomly Recruited Sample
This section presents the results from the KN sample. KN utilized a random recruitment method
where each individual in the population has an equal chance of inclusion in the sample; a fact
which implies that the results from the KN sample can be used, along with statistical theory, to
make statements about the percentage of the U.S. population having particular attitudes about
animal cloning.
5.1 Awareness of Animal Cloning and Other Animal Breeding Techniques
Table 2 reports people‟s stated awareness of assisted reproduction technologies that are used to
breed animals for meat and milk production. Overall, people indicated that they had heard or
read more about cloning than any of the other techniques. For example, in the sample that did
not receive prior information, only 14% had never heard about animal cloning, whereas 47% had
never heard of embryo transfer, 36% had never heard of biotechnology used to breed animals,
32% had never heard of in vitro fertilization, and 25% had never heard about artificial
insemination. Not surprisingly, providing people information about animal cloning one week
prior to administration of the survey significantly increased awareness of this and several other
reproductive technologies at the time of the survey.
5.2 Willingness to Eat Cloned Meat and Milk
Table 3 reports the extent to which people agreed or disagreed with statements about willingness
to eat and purchase cloned meat and milk. First, we note that providing people a week to digest
information on animal cloning had no affect on stated willingness to eat cloned meat or milk.
Apparently, people who were only provided information on cloning at the time of the survey
were not simply giving answers as result of a “shock value” or “gut reaction,” assuming that they
actually contemplated the information provided. The answers given after a week of time to
contemplate the information were no different than those given “on the spot.” This result may be
due, in part, to the fact that people were already at least somewhat aware of the technology,
especially in comparison to other assisted reproductive technologies (see table 2), and as such,
people may have already formed opinions on the issue prior to the study.
Results in table 3 reveal that there was virtually no difference in willingness to eat meat and
willingness to drink milk from cloned animals. For both meat and milk, about 31% indicated
that they were willing to eat meat/milk from cloned animals, whereas 43% to 44% indicated they
were not. People did not differentiate much between meat/milk from clones and the meat/milk
from the offspring of clones. Regardless of whether the meat/milk was from cloned animals or
their offspring, the percentage of people willing to eat remained about 31% and the percentage
unwilling to eat was about 43% to 44%. A little more than 40% of respondents indicated that
they would likely alter their consumption of meat and milk if they learned that the products came
from cloned animals.
23
Table 3 also reports the result of an “indirect question.” In particular, people were asked to
indicate whether the “average American” was willing to eat meat from cloned animals. People
indicated that they were more willing to eat meat from cloned animals (31%) as compared to the
percentage of people who thought that the “average American” was willing to eat (21%). In
previous research, we have argued that differences in direct and indirect questioning are likely a
result of a type of social desirability bias (see Lusk and Norwood, 2008a,b,c). That is, people
answer direct survey questions in a way to make themselves “look good,” but have no such
motivation when answering questions about what they think others will do. The differences in
direct and indirect questions observed here on cloning are much smaller than the differences we
have observed on questions about organic food and animal welfare. Thus, relative to these other
issues, social desirability bias appears to be of lesser concern for the issue of cloning.
Nevertheless, results do suggest a slight tendency for people to over-state their acceptance of
cloned meat, perhaps out of an attempt to portray themselves as more open to new technologies.
5.3 Beliefs about Safety and Acceptability of Cloned Meat and Milk
People answered a series of agree/disagree questions related to statements about the safety and
acceptability of cloned meat and milk. Table 4 shows that most people (57.5%) were unsure
whether meat currently sold in grocery stores is from cloned animals, suggesting people exhibit a
great deal of uncertainty about the technologies currently being used to breed livestock. About a
quarter of the respondents thought no meat from cloned animals or their offspring was currently
being sold in stores.
Respondents were equally split on the acceptability of animal cloning, with about a third finding
the practice acceptable, a third finding the practice unacceptable, and a third neutral. Only about
21% believed that animal cloning would result in beneficial outcomes for them.
People were equally split on the safety of cloned meat. About 30% agreed that meat from cloned
animals was safe to eat, where as about 29% believed the meat to be unsafe; 41% neither agreed
nor disagreed that meat from cloned animals was safe to eat.
Despite potential concerns about the safety of meat from cloned animals, people expressed
confidence in the safety of meat and milk typically bought in the grocery store. About 64% of
the public believed that, in general, the meat and milk they buy from the grocery stores is safe to
eat. Only 10% disagreed with this statement.
5.4 Beliefs about Federal Government and Cloned Meat and Milk
Although people expressed confidence in the safety of meat and milk (see table 4), somewhat
paradoxically, they expressed little trust or confidence in the federal government to regulate food
safety or cloned meat/milk (see table 5). For example, almost 40% of the public did not believe
the government was doing everything it could to ensure the safety of food products (only 30%
thought they were doing all they could). Only 20% believed that animal cloning is carefully
24
regulated by the U.S. government. Further, only 24% of the public trust the government to
properly regulate the use of animal cloning.
Expressed levels of trust in information about cloning were also relatively low. In order of
decreasing trustworthiness, 32% trust information about cloning from University scientists and
researchers, 29.3% trust information about cloning from the USDA, 28.8% trust information
from the FDA, and only 26.1% of people trust information from the EPA.
These results suggest that the trust the public has in the safety of the general food supply is
apparently not a result of confidence in the government regulating food safety.
5.5 Why Are People Concerned about Animal Cloning?
5.5.1 Relative Importance of Competing Objections to Cloning
People were asked to answer a series of questions in which they were presented with two
statements related to potential objections about animal cloning, and were asked to indicate which
statement best matched their view. Discrete choice models were used to place each issue on an
underlying scale of importance. The models allow us to estimate the probability of a respondent
choosing one issue as more important than another.
The “importance scores” take the form of probabilities, and thus the sum of the estimated
importance scores across all 8 issues must equal 100. If two issues (say issues A and B) are
roughly equivalent in importance to respondents, roughly half the subjects will say Issue A is
more important and half will say Issue B is more important. The importance score calculation
will then assign an identical number (50%) to both issues. Conversely, if Issue A is deemed
more important by 750 individuals, and Issue B deemed more important by 250 people, the
importance score calculation will assign an importance score to Issue A of 75% and an
importance score to issue B of 25%: thus, Issue A is three times as important as Issue B.
Therefore, the importance scores assigned to each issue reflects the percentage of times that issue
was considered to better match people‟s views than other issues. Because these probability
statements are on a ratio scale, they can be compared proportionally. That is, if Issue A‟s
importance score is two times larger than Issue B, then Issue A is twice as important as Issue B.
Table 6 reports the model estimates for the logit and RPL models and the calculated importance
scores for each statement or issue. Likelihood ratio tests indicate that the logit model can be
rejected in favor of the RPL model, and as such, we focus on the results from the RPL
specification, which are shown in figure 1. Results indicate that the two most popular objections
to animal cloning were: (i) “Cloning is „unnatural‟ because it is not a process that occurs in
nature” and (ii) “Animal cloning will lead to human cloning.” These statements provide almost
two times a better match for people‟s views toward cloning than statements such as “Animal
cloning is morally wrong” or “Cloning will reduce genetic diversity to unacceptable levels.”
People were relatively unconcerned that cloning will result in unhealthy farm animals or that
meat and milk from clones and their offspring are unsafe to eat. Indeed, these statements are
about four to five times less likely to match people‟s views toward animal cloning as compared
to concerns about unnaturalness and the potential for human cloning.
25
The results presented in figure 1 provide insight into messages that may persuade people to
become more or less accepting of cloning, but also suggests people are unlikely to be affected
much by persuasion. For example, people seemed to object to cloning primarily because it is not
a process that occurs in nature. That cloning does not occur in nature is, of course, true and no
amount of scientific evidence or persuasion is likely to alter this perception. Concerns about
animal cloning leading to human cloning; however, might be partially mitigated by careful
regulation and oversight. Messages that attempt to alleviate the public‟s concern about the safety
of cloned meat and milk (which was the focus of the report from the FDA) are unlikely to have
much impact because, as shown in table 1, most consumers do not view this issue to be of great
concern in relation to other issues.
5.5.2 Correlations between Willingness-to-Eat and Beliefs about Government Involvement
Table 7 reports simple bi-variate correlations between people‟s agree/disagree responses to the
Likert scale questions. Results reveal high correlations, in the range of 0.5 and higher, between
(i) trust in information about cloning from the USDA, (iii) the belief that the government is
doing all it can to ensure the safety of food products, (iv) perceptions about the safety of cloned
meat, and (v) willingness to eat cloned meat. People who have more trust in the information
from the USDA and believe the government is doing what it can to ensure the safety of food are
also the same people who are more convinced of the safety of meat from cloned animals and are
more willing to eat cloned meat/milk. Another interesting result shown in table 7 is that the
belief that cloned meat is already sold in grocery stores products is positively correlated with
people‟s willingness to eat cloned meat. This potentially points to a type of endowment effect
where people are supportive of what they perceive to be the status quo: if people believe cloned
meat is already being sold, they are more willing to eat: if people believe cloned meat is not
being sold, they are less willing.
5.5.3 Correlations between Competing Concerns for Cloning and Willingness-to-Eat
Although the results presented in Table 6 and Figure 1 provided a picture of the relative level of
concern for several issues related to animal cloning, it is of interest to ask whether people who
express a higher overall level of concern about animal cloning find certain issues to be more or
less problematic than people who express a lower overall level of concern.
Table 8 provides some insight into this issue by reporting bi-variate correlations between
agree/disagree responses to selected Likert scale questions and the individual-specific
“importance scores” for competing cloning concerns derived from the random parameter logit.
Results shown in table 8 indicate that people who are relatively more concerned about the
morality of animal cloning are the same people who are less willing to eat meat from cloned
animals. Thus, although the morality of animal cloning only ranked fourth on the list of
competing concerns, it is an issue highly related to willingness to eat cloned meat. Interestingly,
the issue of most concern to people – that cloning is unnatural because it is not a process that
occurs in nature – is virtually unrelated to people‟s willingness to eat cloned meat. That is,
26
people for whom “unnaturalness” is a relatively big concern are just as likely to express
willingness to eat cloned meat as are people for whom “unnaturalness” is not as big a concern.
5.5.4 Relationships between Socio-Economic, Demographic Factors and Cloning Attitudes
To determine the relationship between socio-economic factors and attitudes toward cloning,
several ordered probit models were estimated (see table 9). The dependent variables are
responses to the agree/disagree Likert-scale questions. Given that the responses to these
questions fall on a 5-point scale, the ordered probit model is the appropriate specification as it
treats the dependent variable as ordinal rather than cardinal. The reported parameter estimates
correspond to the marginal effects on the underlying latent (unobserved) variable, which is the
propensity to agree with each statement.
Table 9 reports the results of the probit regressions. Results reveal that providing respondents
with a week to digest information on animal cloning had no effect on responses to the
agree/disagree Likert scale questions.
Results reveal that females are more likely to agree that animal cloning is unacceptable than
males. Likewise, males are more likely to be willing to eat meat from cloned animals and are
more likely to believe that cloned meat is safe to eat than females. Results also indicate that
people with only a high school diploma were more likely to believe cloning was unsafe and
believe that cloning is unacceptable than people with a bachelor‟s degree or higher. Thus,
education appears to have some relationship to the acceptability of cloning.
Interestingly, and somewhat surprisingly, whether people lived in a household with Internet
access was strongly associated with most of the dependent variables shown in table 9. People in
households without Internet access are more likely to believe that animal cloning is unacceptable
than households with Internet access. Similarly, people in households with Internet access are
more likely to believe meat from clone animals is safe to eat, are more willing to eat cloned meat,
and express greater trust in information from the USDA than non-Internet households. This
finding may be due to the fact that households with Internet access have received more
information about cloning. An alternative explanation is that people with Internet in the
household may be more accepting of technology in general than non-Internet households, and
this general acceptance of technology may spill over into acceptance of cloning technology.
The last few rows in table 9 indicate that people who are the primary shoppers of food in their
household are more likely to disagree with the statement that animal cloning is unacceptable.
Further, people that had children under the age of 12 in their household are less likely to believe
that meat from cloned animals is safe to eat.
5.5.5 Food Values and Their Correlations with Cloning Concerns
Table 10 reports on the results of an analysis carried out to identify people‟s food values, i.e.,
which general issues are most important to people when they purchase food. As shown in figure
27
2, food safety is overwhelmingly the most important food value, being twice as important as the
next most important food value, nutrition. Out of the 11 competing food values, four represent
73% of the overall importance score: safety, nutrition, taste, and price. This latter finding is
consistent with that reported by Lusk and Briggeman (2008).
Of more direct relevance to this study is the relationship between food values and concerns for
cloning. Table 11 reports bi-variate correlations between people‟s agree/disagree responses to
selected Likert scale questions and individual-specific “importance scores” for the food values.
The food value of “naturalness” defined as “the extent to which food is produced without
modern technologies” appears to be most related to trust in the government and willingness to
eat cloned meat and milk. People for whom naturalness was a more important food value were
less likely to believe the government is doing everything it can to ensure the safety of food and
were less willing to eat meat from cloned animals. A similar finding is true for people that
believe origin is a more important general food value. Table 11 shows that people who believe
food safety is a relatively more important general food value are no more or less willing to eat
cloned meat/milk than people who think food safety is a less important food value.
5.6 Preferences for Cloning Policies
People were asked how they would vote on three different federal policies related to animal
cloning. Table 12 reports a simple cross-tabulation reporting the percentage of people who
would vote in favor of each policy at each of the seven price levels randomly assigned to
respondents. Table 13 analyzes this data more formally using logistic regressions predicting the
probability of a “yes” vote as a function of the price increase. Figure 3 illustrates the predicted
probabilities of voting in favor of each policy as a function of the change in food prices.
Results reveal low levels of support for a policy requiring the mandatory tracking of cloned
animals and a policy banning animal cloning. Indeed, table 13 shows that the point-estimate on
mean willingness-to-pay is negative and that 0% is clearly within the 95% confidence intervals.
These results indicate that the public is not willing to pay to enforce a mandatory tracking system
or to ban the practice of animal cloning. One possible explanation for the finding that the point
estimate on the ban is negative (-13%) is that people expect that some benefits may result from
research on animal cloning and they are reluctant to impede scientific discoveries that may result
in some future benefit. Indeed, Hamilton, Sunding, and Zilberman (2003) argue that people are
often unwilling to pay for policies because of a lost option value; when a regulation or ban takes
place it is often difficult to reverse and people lose the option or ability to change their mind as
more information becomes available in the future.
The only policy for which mean willingness-to-pay was significantly greater than zero was a
mandatory labeling policy. Results indicate that people would be willing to pay up to 31%
higher food prices for the ability to know whether meat and milk is from a cloned animal; a
policy which provides the option of choosing which product best suits their preferences given
market prices. One important note about a mandatory labeling policy is that, depending on how
it was implemented, it may require mandatory tracking. Thus, although people were unwilling to
pay for mandatory tracking per se, they might value it as a part of a labeling system.
28
5.7 Ground Beef Market Simulator
People answered a series of questions regarding which option of ground beef they would buy (or
none), where each ground beef option was described by a series of attributes including fat
content, price, and use of cloning. The discrete choice data was analyzed using multinomial and
random parameter logit models (see table 14), which provide estimates of people‟s attribute-
based utility function. Likelihood ratio tests indicate that the RPL provides a significantly better
fit to the data, and as such, results from this specification are discussed.
Table 14 shows that the mean price effect is negative (-1.14), meaning increases in price
decrease utility (i.e., higher priced options are less likely to be chosen than lower priced options,
all else held constant). People enjoy increases in ground beef leanness and decreases in saturated
fat content. Comparing the relative magnitude of the dummy variables related to cloning
indicates that people strongly preferred ground beef from a non-cloned animal to ground beef
from a cloned animal (3.922 vs. 0.112). The last column in table 14 reports significant standard
deviations for all product attributes, meaning there is significant heterogeneity in the population
in terms of people‟s preferences for fat content and cloning. For example, while the mean
marginal utility of leanness was 0.057 (i.e., a 1% increase in leanness increases utility by 0.057
units), the standard deviation was 0.067, meaning 95% of the population has a preference
parameter for leanness between -0.074 and 0.188. That some people actually prefer fattier
ground beef is of course consistent with the notion that some people prefer taste over healthiness.
Following previous literature, the standard deviation on the price coefficient was restricted to
zero because: (i) for economic theory to hold, a parametric distribution that is non-negative
would have to be used for the price coefficient (e.g., the log-normal), which typically generates
willingness-to-pay distributions with very fat, and often unrealistic, tails, (ii) it ensures normally
distributed willingness-to-pay estimates, and (iii) it facilitates model convergence.
Table 15 reports willingness-to-pay estimates derived from the models reported in table 14.
Results reveal large willingness-to-pay premiums for meat from non-cloned animals relative to
meat from clones or offspring of clones. The mean willingness-to-pay for non-clone vs. cloned
ground beef is $4.23 per choice occasion. Stated differently, if ground beef from a non-cloned
animal was priced at a premium of $4.23 to ground beef from a cloned animal, the “average”
person would be indifferent to buying the two products. Despite this large willingness-to-pay
estimate, the standard deviation is also quite large. The estimates in table 14 imply that the
standard deviation of willingness-to-pay for non-clone vs. cloned ground beef is $3.44, meaning
that 95% of the population has a willingness-to-pay premium for non-cloned ground beef
between -$2.52 and $10.98. In fact, these estimates imply that 11% of the people in the
population are unwilling to pay a premium for ground beef from a non-cloned animal over
ground beef from a clone.
Results reveal people are willing to pay a $0.50 for a 10% increase in leanness. The implied
marginal effect (a $0.05 increase for each one percent increase in leanness) is very similar to the
$0.04/lb estimate reported by Parcell and Schroeder (2007), who obtained their estimate using
actual market transactions and hedonic price analysis. That our choice-based conjoint model
provides similar results to that obtained from real market transactions provides some confidence
in the underlying validity of the model.
29
The estimates reported in tables 14 and 15 can also be used to get a feel for the relative
importance of cloning as compared to, say, fat content. The results reveal that, on average, it
would take an 84.5% change in leanness to make people indifferent to a cloned and non-cloned
product. That is, people would be indifferent to purchasing ground beef from a cloned product
that was 95% lean and a non-cloned product that was only 10.5% lean (e.g., 95% - 84.5% =
10.5%), holding prices constant. Of course, it is silly to think that anyone would want a product
that is only 10.5% lean, and the difference in leanness involved (84.5%) is extrapolating well
beyond the differences in leanness that existed in the data; however, the results qualitatively
suggest that the attribute of cloning is much more important to people than the attribute of
leanness. A similar comparison can be made for saturated fat content. On average, it would take
a 24.5% change in saturated fat content to make people indifferent to a cloned and non-cloned
product. That is, people would be indifferent to purchasing ground beef from a cloned product
that had 5% saturated fat and a non-cloned product that had 29.5% saturated fat. Although
saturated fat appears relatively more important than leanness, the attribute of cloning is more
important still.
Given the estimates in table 14, any number of market simulations can be constructed. To
facilitate such analyses, a market simulator was constructed, and can be accessed at:
http://agecon.okstate.edu/faculty/publications/3103.xls . Figure 4 shows an example of the
output that can be generated under one particular simulation. Examples of insights that can be
gleaned from the simulator are: (i) when faced with a choice between ground beef from a cloned
and a non-cloned animal, more than 75% of shoppers are predicted to choose the non-cloned
beef even when sold at a $1.50 price premium over ground beef from a cloned animal, and (ii) if
faced with a choice of only being able to buy ground beef from a cloned animal or no ground
beef at all, the majority of consumers would choose to purchase the cloned ground beef
(assuming the price of ground beef was $3.50/lb or lower).
5.8 Milk Market Simulator
People answered a series of questions regarding which milk option they would buy (or none),
where each option was described by a series of attributes including fat content, price, and use of
cloning. The discrete choice data was analyzed using multinomial and random parameter models
(see table 16), which provide estimates of people‟s attribute-based utility function. Likelihood
ratio tests indicate that the RPL provides a significantly better fit to the data, and as such, results
from this specification are discussed.
Table 16 shows that the mean price effect is negative (-0.734), meaning increases in price
decrease utility (i.e., higher priced options are less likely to be chosen, all else held constant).
On average, people most preferred 2% milk, followed by Skim milk, 1% milk, and then whole
milk. However, the RPL model shows, as expected, significant heterogeneity in preferences for
fat content. For example, while the mean marginal utility for whole milk versus skim was -
0.545, the standard deviation was 2.054, meaning 95% of the population has a preference
parameter for whole milk between -4.57 and 3.48. Clearly, the variation in preferences for this
attribute is of more importance than the mean. Comparing the magnitude of the dummy
30
variables related to cloning indicates that people strongly preferred milk from a non-cloned
animal to milk from a cloned animal (4.333 vs. 1.668). As in the ground beef models, the
standard deviation on the price parameter was restricted to zero.
Table 17 reports willingness-to-pay estimates derived from the models reported in table 16.
Results reveal large willingness-to-pay premiums for milk from non-cloned animals relative to
meat from clones or offspring of clones. The mean willingness-to-pay for non-clone vs. cloned
milk is $3.63 per choice occasion. Despite this large willingness-to-pay estimate, the standard
deviation is also quite large. The estimates in table 16 imply that the standard deviation of
willingness-to-pay for non-clone vs. cloned ground beef is $4.91, meaning that 95% of the
population has a willingness-to-pay value between -$5.99 and $13.25. These estimates imply
that 23% of the people in the population are unwilling to pay a premium for milk from a non-
cloned animal over milk from a clone.
The estimates reported in tables 17 provide a feel for the relative importance of cloning.
Although people are, on average, willing to pay a $1.80 premium for “non rBST” milk, this is
only about half the amount they are willing to pay to avoid milk from a cloned animal. Cloning
is more important to people than the use of rBST.
Given the estimates in table 16, any number of market simulations can be constructed. To
facilitate such analyses, a market simulator was constructed, and can be accessed at:
http://agecon.okstate.edu/faculty/publications/3104.xls. Figures 5 and 6 show examples of
outputs that can be generated under two particular simulation scenarios. Examples of insights
that can be gleaned from the simulator are: (i) when faced with a choice of milk from a cloned
and a non-cloned animal, more than 65% of shoppers are predicted to choose the non-cloned
milk even when sold at a $1.50 price premium over milk from a cloned animal, and (ii) if faced
with a choice of only being able to buy milk from a cloned animal at $3.00/gallon or no milk at
all, only about 47% of consumers would choose to make the purchase.
31
6. Comparing Results across Three Survey Samples
This section compares results from the KN, IRI, and Nielsen samples, and for the IRI and
Nielsen samples, shows the relationship between actual purchases of meat/milk and concerns for
cloning.
6.1 Sample Characteristics
Table 18 shows the original unweighted characteristics of each of the survey samples. None of
the samples perfectly match the U.S. population, but overall the KN random sample appears to
be most similar to the U.S. census data. There were substantially more females and primary
shoppers in the IRI and Nielsen samples than in the KN sample, a result that is not surprising
given the differences in the ways the three panels are constructed and used by the subcontractors.
To facilitate comparisons across survey samples, recall that for each survey sample, post-
stratification weights were created. This means that the remaining results reported from each
sample, KN, IRI, and Nielsen, are each reflective of the U.S. population in terms of geographic
location, age, race, gender, and education. Thus, although differences in concern for cloning
may be observed across samples, such differences are not a result of differences in geographic
location, age, race, gender, and education across the samples.
6.2 Food Values
To begin the comparison of responses from the three survey samples, it is instructive to start by
comparing responses to the food value questions because these questions were asked prior to any
information being provided about cloning, and as such, they allow for a straightforward
comparison across survey samples without having to worry about differences in information.
Table 19 reports the results of logit models estimated to identify people‟s food values, i.e., which
general issues are most important to people when they purchase food. For all three samples,
food safety is overwhelmingly the most important food value, being twice as important as the
next most important food value. The second most important food value was taste for the KN and
Nielsen samples, but the nutrition was the second most important food value for the IRI sample.
The only sample for which the difference in importance assigned to nutrition vs. taste is
quantitatively large is the Nielsen sample, and for the Nielsen sample, taste is 1.4 times more
important than nutrition (by contrast, nutrition is only 1.06 times more important than taste for
the IRI sample). For all three samples, “fairness” is the least important food value. The value of
“naturalness” was slightly more important to the people in the KN sample than for the people in
the IRI and Nielsen samples, and the opposite was true for the value of “appearance.”
A likelihood ratio test indicates that we can reject the null hypothesis that the estimated
parameters are equivalent across all three samples (P < 0.001). This provides evidence that even
after controlling for geographic location, age, race, gender, and education via weighting, there
are significant differences across the KN, IRI, and Nielsen samples – at least in a statistical sense.
32
However, the results in table 19 indicate that the differences across samples are not particularly
large in that they would lead to substantially different conclusions about the food values people
are most concerned about. This is to say that although the differences are statistically significant,
they may not be economically significant.
6.3 Awareness of Animal Cloning and Other Animal Breeding Techniques
Table 20 reports people‟s stated awareness of assisted reproduction technologies that are used to
breed animals for meat and milk production for each of the three survey samples. Other than the
differences in the two KN treatments previously discussed, there are not large differences across
samples. There does appear to be a slight tendency for people in the opt-in panels to indicate a
higher mean level of awareness of all assisted reproductive technologies, but the differences are
not large. The Nielsen sample tended to have a higher fraction of people who had not heard
anything at all about each of the reproductive technologies. For example, 16.4% of the Nielsen
sample had not previously heard or read about cloning, whereas only 10.3% and 14.1% of the
IRI and KN samples, respectively had not previously heard or read about cloning.
6.4 Willingness to Eat Cloned Meat and Milk
Table 21 reports the mean responses (on the 5-point scale) related to the extent to which people
agreed or disagreed with statements about willingness to eat and purchase cloned meat and milk.
Table 22 reports the percentage of respondents in each of the three samples who agreed,
disagreed, and neither agreed nor disagreed with these statements. Because we previously
showed in section 5 that providing people a week to digest information on animal cloning had no
effect on stated willingness to eat cloned meat or milk in the KN sample, we simply pooled the
data across the information treatments in the KN sample.
Results in table 22 reveal that people in the IRI sample, who were only provided the short
information statement, were the least willing to eat cloned meat and milk. These differences are
statistically significant, and suggest that the information from the FDA provided to the KN and
Nielsen samples served to improve acceptance of cloning. Again, although the change was
statistically significant, it is important to emphasize that willingness to eat clone meat only
jumped 10% from the IRI to Nielsen sample and remained below 34% of the total population for
both samples. That is, the information did not generate a massive shift in conclusions about
public acceptance of cloning.
There were not large differences in the KN and Nielsen samples in terms of stated willingness to
eat meat/milk from clones and their offspring, but the Nielsen sample was more likely than the
KN or IRI samples to agree with the statement “If I learned that the meat/milk products I
regularly purchase came from clone animals, I would continue to buy the meat products as
usual.” This provides slight evidence that the Nielsen opt-in panel is more accepting of cloning
than is the general population, as represented by the KN panel.
33
6.5 Beliefs about Safety and Acceptability of Cloned Meat and Milk
Tables 23 and 24 report results related to the series of agree/disagree questions on statements
about the safety and acceptability of cloned meat and milk. Results are similar to that discussed
in the preceding sub-section. The IRI sample was more likely to agree that animal cloning is
unacceptable than the KN and Nielsen samples (40% vs. 32% vs. 34%, respectively). The IRI
sample was least likely to agree that meat from cloned animals was safe to eat, whereas the
Nielsen sample was most likely to agree with this statement. Apparently, the lengthier
information statement served to somewhat reduce fears about the safety of consuming cloned
meat. Interestingly, the last row of table 23 indicates that both the IRI and Nielsen samples
indicated a higher level of agreement, on average, with the statement, “In general, the meat and
milk I buy from the grocery store is safe to eat,” providing some indication that the opt-in panels
are less concerned about food safety as compared to the general public represented by the KN
sample. This finding is consistent with the food values analysis reported on in table 19, where
the “importance score” for food safety was 33.1% for the KN sample, but only 27.8% for the IRI
sample and 30.2% for the Nielsen sample.
6.6 Beliefs about Federal Government and Cloned Meat and Milk
Tables 25 and 26 provide information on people‟s beliefs about the federal government and
cloned food in each of the three survey samples. Providing people lengthier information about
cloning served to increase people‟s beliefs that “animal cloning is carefully regulated by the U.S.
government,” as can be seen by comparing responses across the IRI, Nielsen, and KN samples.
The Nielsen sample was most optimistic that the U.S. government is doing all it can to ensure the
safety of food and can trace cloned animals back to the farm.
The last five rows of tables 25 and 26 indicate little difference across samples in terms the
expressed levels of trust. The ranking of trustworthiness was generally consistent across samples.
People most trusted University scientists and researchers, the USDA, the FDA, and then the EPA.
Regardless of the survey sample, the results suggest that the trust the public has in the safety of
the general food supply is apparently not a result of confidence in the government regulating
food safety; trust in information about cloning was low in all three samples.
6.7 Relative Importance of Competing Objections to Cloning
Table 27 reports an analysis carried out on all three survey samples to determine the relative
importance of competing objections about animal cloning. All three samples rating
“unnaturalness” as the most important objection to cloning, although the IRI sample rated this as
a slightly higher objection than the KN or Nielsen samples. The second most objectionable issue
with cloning, for all three samples, was that “scientists and biotechnology companies who
developed cloning technology cannot be trusted to look out for my best interest.” The least
objectionable issue for all three samples was that meat and milk from clones is unsafe to eat.
This is interesting because the IRI sample was not provided any information about the FDA‟s
statement on the safety of cloned meat and milk products.
34
6.8 Relationship between Milk/Meat Purchases and Willingness to Eat Cloned Food
One of the purported advantages of the opt-in panels is the ability to link actual purchase data to
people‟s survey questions. In this section (and in tables 28 and 29), we explore whether such
actual purchase data are associated with people‟s stated willingness to eat cloned milk and meat,
and investigate whether actual purchases yield similar insights as people‟s stated purchases.
Investigating whether concerns for cloning are higher or lower for people that are heavier
consumers of meat or milk is important because it relates directly to the market impacts of the
new technology. For example, if heavy consumers of meat are significantly less concerned about
cloning, then the market impacts of introducing meat from cloned animals is likely to be much
smaller than if the opposite is true.
6.8.1 Milk
Table 28 reports the results of five ordered probit regressions, where the dependent variable in
each regression is the agree/disagree response on the 5-point scale to the statement “I am willing
to consume milk products from cloned animals.” The regressions include a variety of variables,
which hold constant the effect of age, gender, education, income, etc., when investigating the
effect of milk purchases on willingness to consume cloned milk products.
The last few rows in table 28 show the effect of stated and actual milk purchases on stated
willingness to consume milk from cloned animals. First looking at the effect of stated milk
consumption, there are contradictory results across samples. For the KN sample, people who
never purchase milk or only purchase about once a year are less willing to consume cloned milk
than people who say they purchase milk every day; a similar result is found for the Nielsen
sample. However, for the IRI sample, just the opposite is true - people who say they never
purchase milk or say they only purchase about once a year are more willing to consume cloned
milk than people who say they purchase milk every day.
Turning to the actual purchase data, there is no relationship between annual purchase volume (as
measured in gallons for the IRI sample and units for the Nielsen sample) and willingness to eat
cloned milk. For both samples, we also tested whether there might be a quadratic relationship
between actual milk purchases and willingness-to-consume, but such quadratic terms were not
statistically significant. The IRI data set also contained information for each household on
annual purchases of organic milk (in gallons). Based on this information, a second variable was
created which measures the share of total milk purchases attributable to organic milk. On
average, only 5.4% of total milk purchases were organic, but there was wide variation in the
sample with some people buying no organic milk (i.e., share = 0) and others only buying organic
milk (i.e., share = 1). As shown in table 28, there is a strong and statistically significant
relationship between organic purchases and stated willingness to consume cloned milk. Holding
total milk purchases constant, people who bought more organic milk were much less willing to
drink cloned milk than people who bought less organic milk. Stated differently, those people
who are more interested in organic products (as measured via actual purchase behavior) are also
more concerned about cloning.
35
6.8.2 Meat
Table 29 reports the results of five ordered probit regressions, where the dependent variable in
each regression is the agree/disagree response on the 5-point scale to the statement “I am willing
to eat meat from cloned animals.” The regressions include a variety of variables, which constant
the effect of age, gender, education, income, etc., when investigating the effect of meat purchases
on concerns for cloned milk.
The last few rows in table 29 show the effect of stated and actual meat purchases on willingness
to consume meat from cloned animals. For all three samples, there is a tendency for people who
say they purchase meat every day to be more willing to eat cloned meat than people who never
purchase meat. This effect is strongest in the Nielsen sample.
Turning to the actual purchase data, we find that there is no relationship between annual
purchase volume (in lbs) of breakfast and lunch meat purchased and willingness to eat cloned
meat as indicated in the IRI sample. Unfortunately, IRI did have data on sales of organic meat or
data on sales of fresh met. The Nielsen data set contained information on annual fresh beef
purchases (in packages or units). As can be seen in the last row of table 29, increases in actual
purchase of fresh beef are associated with a higher willingness to consume cloned meat.
6.9 Preferences for Cloning Policies
Tables 30, 31, and 32 report average willingness-to-pay (WTP) for mandatory tracking,
mandatory labeling, and banning polices, respectively for each of the three samples.
Table 30 shows that although the KN sample was not willing to pay for a mandatory tracking
policy, the IRI sample was willing to pay 61.7% higher food prices and the Nielsen sample was
willing to pay 32% higher food prices for a mandatory tracking policy. The huge difference in
WTP between the IRI sample and the KN and Nielsen samples suggest that, in this context, there
could be significant economic value associated with information provision.
Table 30 reports mean WTP for a mandatory labeling policy. All three samples reported positive
WTP for this policy. WTP for both the KN and Nielsen samples was about 32% - i.e., people
were willing to pay up to 32% higher food prices to have a mandatory labeling policy on meat
and milk from cloned animals and their offspring. Consistent with the previous results, the IRI
sample was WTP an even higher amount – almost 100% higher prices.
Table 32 shows that WTP for a ban is not significantly different than zero for any of the three
samples. This finding suggests that even though the IRI sample was significantly more opposed
to cloning than the KN or Nielsen samples, they were not willing to prohibit the use of the
technology all together.
36
6.10 Ground Beef Choice-Based Conjoint Estimates
Table 33 shows the multinomial logit estimates fit to people‟s choices of which option of ground
beef they would buy (or none), where each ground beef option was described by a series of
attributes including fat content, price, and use of cloning. A likelihood ratio test indicates that
we can reject the null hypothesis that the estimated parameters are equivalent across all three
samples (P < 0.001). This provides evidence that that even after controlling for geographic
location, age, race, gender, and education by weighting, there are significant differences across
the KN, IRI, and Nielsen samples.
Table 34 translates the multinomial logit estimates into WTP for various beef attributes. WTP
for non-cloned vs. cloned ground beef in the KN and IRI samples is similar, but the Nielsen
sample is WTP $0.56 and $0.91 less than the IRI and Nielsen samples, respectively. It is
interesting that throughout the previous versions of the survey, the IRI sample exhibited higher
stated levels of concern for cloning, but in these choice questions on ground beef, WTP to avoid
cloned ground beef was not significantly higher than the KN sample than in the IRI sample.
Focusing on fat content, results indicate that the KN sample exhibited higher WTP for leanness,
but lower WTP for reductions in saturated fat content than the IRI and Nielsen samples. Results
from the KN sample imply that people are WTP about $0.05 increase for each one percent
increase in leanness, which is quite similar to the $0.04/lb estimate reported by Parcell and
Schroeder (2007), who obtained their estimate using actual market transactions and hedonic
price analysis. That the KN sample provides similar results to that obtained from real market
transactions provides some indication of the validity of the KN sample relative to the other two
samples.
6.11 Milk Choice-Based Conjoint Estimates
Table 35 shows the multinomial logit estimates fit to people‟s choices of which option of milk
they would buy (or none), where each milk option was described by a series of attributes
including fat content, price, and use of cloning. A likelihood ratio test indicates that we can
reject the null hypothesis that the estimated parameters are equivalent across all three samples (P
< 0.001). This provides evidence that that even after controlling for geographic location, age,
race, gender, and education by weighting, there are significant differences across the KN, IRI,
and Nielsen samples.
Table 36 shows the multinomial logit estimates translated into people‟s WTP for various milk
attributes. Unlike the case of beef discussed in the previous sub-section, the results here are
more similar to the rest of the analysis. In particular, WTP for non-cloned vs. cloned milk is
highest in the IRI sample and is lowest in the Nielsen sample.
Because of the strong interaction effect between organic milk purchases and concern for cloning
identified in a previous section, we modified the choice model fit to the IRI sample such that
preferences for cloning were conditioned on people‟s total volume of annual milk purchases (in
gallons) and the share of total purchases resulting from organic milk. The multinomial logit
37
estimates are shown in table 37 and the resulting WTP values are shown in table 38. Results in
table 38 show that WTP for non-cloned vs. cloned milk remains about $4.66 regardless of the
total amount of milk purchased (i.e., WTP is not sensitive to total annual milk purchases).
However, willingness-to-pay was strongly influenced by organic purchasing behavior. For
people who never buy organic milk, WTP for non-cloned milk vs. cloned milk is $4.49 and this
figure jumps to $7.88 for people who only buy organic milk. These findings suggest that
companies wishing to market “cloned free” milk will likely find it profitable to market to the
niche of consumers who are currently purchasing organic milk.
7. Conclusions
This report presented the results of three nationwide surveys on animal cloning conducted with
over 6,000 U.S. consumers.
Given the media attention devoted to the issue of cloning, one might suspect that people are
highly averse to the technology. The survey results, however, suggest a much more mixed
picture, with the public being evenly split on whether animal cloning is acceptable. For example,
in our survey of a random sample of U.S. citizens, we found that only 29% of the public thought
that cloned meat was unsafe to eat. Further, we found that only about 43% were unwilling to eat
meat or drink milk from cloned animals. Our findings are generally similar to those obtained by
previous opinion polls conducted on the issue. For example, our estimate on willingness-to-eat
falls in between the findings from a Pew study indicated that only 35% said they would never
buy meat from a cloned animal, and an IFIC (2007) study reporting that 53% of consumers said
they were either not too likely or not at all likely to purchase meat, milk, or eggs from the cloned
animals.
When asked which particular issues were of most concern related to animal cloning, we found
that concerns about food safety were least prominent. People were most concerned about animal
cloning because, “cloning is „unnatural because it is not a process that occurs in nature,” and
“animal cloning will lead to human cloning.” Only 13% of people believe that the most
problematic issue with animal cloning is that it is morally wrong. Although only a small portion
of the population believes morality to be the most troubling issue with animal cloning, people for
whom this is a primary concern, are much less willing to eat meat from cloned animals than
people who find other issues more problematic.
Comparing results across our three survey samples suggests that information can have a small
but significant influence on people‟s acceptance of cloning. For example, only 20.6% of people
in one sample (who were only given a two sentence definition describing cloning) agreed they
were willing to eat meat from cloned animals, but 34.1% and 30.8% of people in the other two
samples of consumers studied (who were provided a one-half page discussion on cloning)
indicated they were willing to eat meat from cloned animals.
Results also indicate slight differences in responses from “opt in” panels of consumers and true
random samples of consumers. There appeared to be a slight tendency for people in the “opt in”
panel to be more accepting of cloning than people in the random sample. One of the advantages
38
of the “opt in” panels employed in this study is that actual purchase information was available on
each household surveyed. There is a strong and statistically significant relationship between
actual organic purchases of milk and stated willingness to consume cloned milk. Holding total
milk purchases constant, people who bought more organic milk were much less willing to drink
cloned milk than people who bought less organic milk. Stated differently, those people who are
more interested in organic products (as measured via actual purchase behavior) are also more
concerned about cloning and are willing to pay higher amounts to avoid cloned products.
Although total milk consumption was unrelated to concerns for cloning, there is some evidence
that more frequent purchasers of fresh beef are more willing to eat meat from cloned animals as
compared to less frequent purchasers of fresh beef.
Our results revealed that people, on average, are willing to pay significant premiums to avoid
ground beef and milk from cloned animals. For example, the “average” person in the random
sample would be willing to pay a premium of up to $4.23 when purchasing a package of ground
beef to have ground beef come from a non-cloned vs. cloned animal. The willingness-to-pay
premiums to avoid cloning were significantly higher than that to change fat content in ground
beef and milk, and to avoid rBST use in milk. That is, use of cloning is apparently more
important to people than fat content or rBST use. Despite these findings, two important caveats
are in order. First, there is significant heterogeneity in people‟s valuations. For example, results
indicate that ninety-five percent of the population is willing to pay between -$5.99 and $13.25 to
avoid milk from cloned animals. On the one hand, this implies that not everyone is willing to
pay anywhere near the mean premium estimate. Indeed, 23% of the population is unwilling to
pay a premium for milk from a non-cloned animal over milk from a clone. Conversely, there are
some people willing to pay very large premiums for meat and milk from non-cloned animals,
suggesting the potential for viable niche marketing opportunities. For example, roughly 25% of
the population is willing to pay more than $7.00 to have milk from a non-cloned animal rather
than milk from a clone. Second, willingness-to-pay a market premiums to avoid cloned products
does not imply support for several cloning policies. For example, results from the random
sample of U.S. citizens indicates that willingness-to-pay for a mandatory tracking policy on
cloned animals and willingness-to-pay for a ban on animal cloning is effectively zero. The only
policy to receive majority support was a mandatory labeling policy on meat and milk from
cloned animals.
39
8. References
Arrow, Kenneth; Solow, Robert; Portney, Paul; Leamer, Edward E.; Radner, Roy; and Schuman,
Howard. 1993. “Report of the NOAA Panel on Contingent Valuation,” Federal Register,
58(10), January 15 1993:4602-4614.
Boyle, K.J. 2004. “Contingent Valuation in Practice.” In A Primer on Nonmarket Valuation
(P.A. Champ, K.J. Boyle, and T.C. Brown eds). Dordrecht, Netherlands: Kluwer.
Davis, G.G. and B.H. Lin. 2005. “Factors Affecting U.S. Beef Consumption.” U.S. Department
of Agriculture, Economic Research Service, Publication LDP-M-135-02, October, 2005.
Finn, A., and J.J. Louviere. 1992. “Determining the Appropriate Response to Evidence
of Public Concern: The Case of Food Safety.” Journal of Public Policy and Marketing
11:12-25.
Flynn, T.N., J.J. Louviere, T.J. Peters, and J. Coast. 2007. “Best-Worst Scaling: What It
Can Do for Health Care Research and How to Do It.” Journal of Health Economics
26:171-189.
Hamilton, S.F., D.L. Sunding, and D. Zilberman. 2003. “Public Goods and the Value of Product
Quality Regulations: The Case of Food Safety.” Journal of Public Economics
87(2003):799-817.
Hanemann, W.M. 1984. “Welfare Evaluations in Contingent Valuation Experiments with
Discrete Responses.” Journal of Agricultural Economics 66: 332-341.
Huber, J., and K. Train. 2001. “On the Similarity of Classical and Bayesian Estimates of
Individual Mean Partworths.” Marketing Letters 12:259-269.
International Food Information Council. 2006. “Food Biotechnology: A Study of U.S. Consumer
Attitudinal Trends. 2006 REPORT.”
International Food Information Council. 2007. “Food Biotechnology: A Study of U.S. Consumer
Trends. 2007 REPORT.”
Louviere, J.J., D.A. Hensher, and J.D. Swait. 2000. Stated Choice Methods: Analysis and
Application. Cambridge, UK: Cambridge University Press.
Lusk, J.L. and B. Briggeman. 2008. “Food Values.” American Journal of Agricultural
Economics. forthcoming.
Lusk, J.L. and F.B. Norwood. 2008a. “An Inferred Valuation Method.” Land Economics
forthcoming.
Lusk, J.L. and F.B. Norwood. 2008b. “Opinions about Animal Welfare Obtained from Direct
and Indirect Questioning.” Working Paper, Department of Agricultural Economics.
Lusk, J.L. and F.B. Norwood. 2008c. “Bridging the Gap between Laboratory Experiments and
Naturally Occurring Markets.” Working Paper, Department of Agricultural Economics.
Marley, A.A.J., and J.J. Louviere. 2005. “Some Probabilistic Models of Best, Worst, and
Best-Worst Choices.” Journal of Mathematical Psychology 49:464-480.
McFadden, D. 1973. “Conditional Logit Analysis of Qualitative Choice Behavior.”
in Zarembka, Paul (ed.) Frontiers in Econometrics. New York: Academic Press.
Mellman Group. 2006. “Review of Public Opinion Research.” The Pew Initiative on Food and
Biotechnology.
Morrison, D.G. 1979. “Purchase Intentions and Purchase Behavior,” Journal of Marketing,
43:65-74
Morwitz, V.G. 1997. “Why Consumers Don‟t Always Accurately Predict Their Own Future
Behavior.” Marketing Letters. 8:57-70.
40
Parcell, J.L. and T.C. Schroeder. 2007. “Hedonic Retail Beef and Pork Prices.” Journal of
Agricultural and Applied Economics 39:29-46.
Small, K.A. and H.S. Rosen. 1978. “Applied Welfare Economics with Discrete Choice
Models.” Econometrica 49:43-46.
Sosin, J., and M. D. Richards. 2005. “What will consumers do? Understanding consumer
response when meat and milk from cloned animals reach supermarkets. KRC Research.
Storey, M. L. 2006. “Consumers‟ knowledge, attitudes, beliefs, and purchase intent regarding
foods from the offspring of cloned animals.” University of Maryland Center for Food,
Nutrition, and Agriculture Policy. Final Topline Report.
Thurstone, L.L. 1927. “A Law of Comparative Judgment.” Psychological Review 34:273-286.
Train, K.E. 2003. Discrete Choice Methods with Simulation. Cambridge, UK: Cambridge
University Pres.
USDA, ERS. 2008. Loss Adjusted Food Availability Data for Total Dairy Products. Available
online at: http://www.ers.usda.gov/data/foodconsumption/FoodGuideIndex.htm#dairy.
41
Table 1. Characteristics of Survey Respondents in KN sample
Variable Definition Unweighted
mean
Weighted
mean
Age age in years 49.695 46.513
Treatmenta 1 if treatment1; 0 if treatment2 0.498 0.498
Gender 1 if female; 0 if male 0.492 0.517
Income annual household income in $1,000s 63.055 58.699
Internet 1 if household internet access; 0 otherwise 0.697 0.613
NoHS 1 if less than high school; 0 otherwise 0.101 0.138
HS 1 if high school; 0 otherwise 0.316 0.310
SomeCollege 1 if some college; 0 otherwise 0.274 0.279
Bachelors 1 if Bachelor‟s degree or higher; 0 otherwise 0.309 0.273
White 1 if white, non-hispanic; 0 otherwise 0.785 0.696
Black 1 if black, non-hispanic; 0 otherwise 0.069 0.111
Other 1 if other non-hispanic; 0 otherwise 0.039 0.054
Hispanic 1 if hispanic; 0 otherwise 0.076 0.127
Two-races 1 if 2+races, non-hispanic; 0 otherwise 0.031 0.011
Northeast 1 if in Northeast U.S Census Region; 0 otherwise 0.185 0.186
Midwest 1 if in Midwest U.S Census Region; 0 otherwise 0.225 0.222
South 1 if in South U.S Census Region; 0 otherwise 0.365 0.367
West 1 if in West U.S Census Region; 0 otherwise 0.224 0.225
Meat:Never 1 if never purchase meat; 0 otherwise 0.035 0.041
Meat:Yearly 1 if purchase meat a few times a year; 0 otherwise 0.090 0.096
Meat:Monthly 1 if purchase meat about once a month; 0 otherwise 0.278 0.301
Meat:Weekly 1 if purchase meat about once a week; 0 otherwise 0.558 0.521
Meat:Day 1 if purchase meat every day; 0 otherwise 0.034 0.038
Farm 1 if own/work on ranch/farm; 0 otherwise 0.164 0.157
Pshopper 1 if primary shopper for food; 0 otherwise 0.688 0.670
Child 1 if child under age of 12 in household; 0 otherwise 0.237 0.257 aIn treatment 1, respondents received an information statement about cloning one week prior to taking the survey
and received the statement again while taking the survey; in treatment 2, respondents only received the information
statement while taking the survey.
42
Table 2. Knowledge of Assisted Reproduction Technologies That Are Used to Breed Animals
for Meat and Milk Production
Technology Pooled
Treatment 1
Prior
Information
Treatment 2
No Prior
Information
P-
Valuea
Artificial Insemination 2.43
(1.14)b
[22.9%]c
2.50
(1.15)
[21.3%]
2.37
(1.13)
[24.5%]
0.01
In vitro fertilization 2.20
(1.08)
[29.8%]
2.25
(1.09)
[28.1%]
2.16
(1.06)
[31.5%]
0.03
Biotechnology 2.04
(1.05)
[37.1%]
2.05
(1.07)
[38.2%]
2.04
(1.04)
[36.0%]
0.89
Embryo transfer 1.90
(1.05)
[45.2%]
1.94
(1.07)
[43.7%]
1.86
(1.04)
[46.6%]
0.08
Cloning 2.57
(1.03)
[10.6%]
2.70
(1.03)
[7.1%]
2.45
(1.01)
[14.09%]
<0.01
Number of Observations 2,256 1,123 1,133
Note: response to question: “Overall, how much have you heard or read about each of the following assisted
reproduction technologies that are sometimes used to breed animals for meat and milk production?” Response
categories were: 1 = nothing at all, 2 = a little, 3 = a moderate amount, 4 = quite a bit, 5 = a great deal. aP-value from two-sample t-test that means are equivalent across treatments
bNumbers in parentheses ( ) are standard deviations
cNumbers in brackets [ ] are the percentage of respondents indicating 1=nothing at all
43
Table 3. Willingness-to-Eat Cloned Meat and Milk
Statement Pooled Treat
1a
Treat
2a
P-
valueb
Percent
disagreec
Percent
neither
agree nor
disagreed
Percent
agreee
I am willing to eat meat
from cloned animals
2.72
(1.28)f
2.73
(1.30)
2.70
(1.26)
0.57 43.2% 26.0% 30.8%
The average American is
willing to eat meat from
cloned animals
2.76
(0.98)
2.79
(0.98)
2.73
(0.97)
0.15 35.1% 44.2% 20.7%
I am willing to eat meat
from the offspring of cloned
animals
2.72
(1.29)
2.74
(1.32)
2.70
(1.27)
0.43 43.0% 26.1% 30.9%
I am willing to consume
milk products from cloned
animals
2.70
(1.29)
2.70
(1.31)
2.69
(1.27)
0.85 44.4% 24.8% 30.8%
I am willing to consume
milk products from the
offspring of cloned animals
2.73
(1.29)
2.71
(1.31)
2.74
(1.26)
0.53 43.0% 25.7% 31.3%
If I learned that the meat
products I regularly
purchase came from cloned
animals, I would continue to
buy the meat products as
usual
2.78
(1.29)
2.82
(1.30)
2.74
(1.29)
0.13 41.4% 25.7% 32.9%
If I learned that the milk
products I regularly
purchase came from cloned
animals, I would continue to
buy the milk products as
usual
2.78
(1.31)
2.79
(1.33)
2.77
(1.29)
0.65 42.1% 24.6% 33.3%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aIn treatment 1, respondents received an information statement about cloning one week prior to taking the survey
and received the statement again while taking the survey; in treatment 2, respondents only received the information
statement while taking the survey. bP-value from two-sample t-test of the hypothesis that means are equivalent across treatments
cPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
dPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
ePercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
fNumbers in parentheses are standard deviations
44
Table 4. Beliefs about Safety and Acceptability of Cloned Meat and Milk
Statement Pooled Treat
1a
Treat
2a
P-
valueb
Percent
disagreec
Percent
neither
agree nor
disagreed
Percent
agreee
Some of the meat currently
sold in grocery stores is
from cloned animals or their
offspring
2.79
(0.88)f
2.79
(0.87)
2.79
(0.90)
0.99 27.4% 57.5% 15.1%
Animal cloning is
unacceptable
3.03
(1.26)
3.03
(1.29)
3.03
(1.24)
0.94 34.4% 33.7% 31.9%
Animal cloning will result in
beneficial outcomes to me
2.71
(1.09)
2.75
(1.10)
2.66
(1.07)
0.05 35.8% 43.6% 20.6%
The meat from cloned
animals is safe to eat
2.94
(1.12)
2.96
(1.16)
2.93
(1.08)
0.47 29.2% 41.2% 29.6%
In general, the meat and
milk I buy from grocery
stores is safe to eat
3.68
(0.92)
3.72
(0.90)
3.64
(0.95)
0.05 10.3% 26.1% 63.6%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aIn treatment 1, respondents received an information statement about cloning one week prior to taking the survey
and received the statement again while taking the survey; in treatment 2, respondents only received the information
statement while taking the survey bP-value from two-sample t-test of the hypothesis that means are equivalent across treatments
cPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
dPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
ePercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
fNumbers in parentheses are standard deviations
45
Table 5. Perceptions about the Federal Government and Cloned Meat and Milk
Statement Pooled Treat
1a
Treat
2a
P-
valueb
Percent
disagreec
Percent
neither
agree nor
disagreed
Percen
t agreee
The U.S. government is
doing everything it can to
ensure the safety of food
products
2.80
(1.12)f
2.81
(1.11)
2.78
(1.14)
0.46 40.7% 29.7% 29.6%
The U.S. government can
trace the meat from cloned
animals back to the farm on
which the animal lived
2.98
(1.05)f
2.98
(1.06)
2.98
(1.05)
0.87 27.6% 42.8% 29.6%
Animal cloning is carefully
regulated by the U.S.
government
2.71
(1.01)
2.75
(1.01)
2.68
(1.01)
0.14 37.3% 42.7% 20.0%
I trust the U.S. government
to properly regulate the use
of animal cloning
2.60
(1.15)
2.62
(1.17)
2.57
(1.13)
0.30 47.1% 29.0% 24.0%
I trust information about
cloning from the U.S.
Department of Agriculture
(USDA)
2.76
(1.16)
2.79
(1.18)
2.74
(1.13)
0.33 40.0% 30.7% 29.3%
I trust information about
cloning from the U.S. Food
and Drug Administration
(FDA)
2.74
(1.15)
2.75
(1.16)
2.73
(1.14)
0.62 41.3% 29.8% 28.8%
I trust information about
cloning from U.S.
Environmental Protection
Agency (EPA)
2.70
(1.14)
2.73
(1.14)
2.67
(1.14)
0.20 41.7% 32.2% 26.1%
I trust information about
cloning from University
scientists and researchers
2.89
(1.12)
2.92
(1.11)
2.86
(1.12)
0.25 34.5% 33.5% 32.0%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aIn treatment 1, respondents received an information statement about cloning one week prior to taking the survey
and received the statement again while taking the survey; in treatment 2, respondents only received the information
statement while taking the survey bP-value from two-sample t-test of the hypothesis that means are equivalent across treatments
cPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
dPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
ePercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
fNumbers in parentheses are standard deviations
46
Table 6. Relative Importance of Competing Objections to Cloning: Logit and Random
Parameter Logit Estimates fit to Paired Comparison Choices
Variable
Econometric
Estimates
Importance
Scores
Logit RPL
Meana
RPL
St.Dev.b Logit RPL
Intercept (order effect) -0.056
(0.032)c
-0.103*
(0.034)
1.265*
(0.072)
Cloning is “unnatural” because it is not a
process that occurs in nature
0.543*d
(0.050)
0.688*
(0.049)
1.759*
(0.049)
24.5% 23.9%
Animal cloning will lead to human
cloning
-0.073
(0.059)
-0.087
(0.060)
2.630*
(0.047)
13.2% 20.6%
Cloning results in animals being viewed
as “objects‟ to be produced as opposed
to being valuable in and of themselves
-0.060
(0.042)
-0.160*
(0.039)
1.966*
(0.001)
13.4% 14.9%
Animal cloning is morally wrong -0.476*
(0.071)
-0.929*
(0.074)
2.632*
(0.062)
8.8% 13.1%
Cloning will reduce genetic diversity to
an unacceptable level
-0.081
(0.045)
-0.142*
(0.042)
1.269*
(0.057)
13.1% 10.5%
The scientists and biotechnology
companies who developed cloning
technology cannot be trusted to look out
for my best interest
0 0 0
14.2% 7.7%
Cloning will result in unhealthy farm
animals
-0.525*
(0.053)
-0.747*
(0.051)
0.916*
(0.058)
8.4% 5.1%
Meat and milk from clones and their
offspring is unsafe to eat
-1.208*
(0.066)
-1.844*
(0.072)
1.845*
(0.069)
4.3% 4.2%
aThe estimates refer to the estimated mean in the population from the random parameter logit model
bThe estimates refers to the estimated standard deviation in the population from the random parameter logit model
cNumbers in parentheses are standard errors
dOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
Notes: Results based on 17,434 choices made by 2,231 individuals; Log-likelihood function value for logit was -
11052.71 and for random parameter logit was -9826.55; a likelihood ratio test could not reject the hypothesis that
the parameters were the same across the two information treatments
47
Cloning is “unnatural” because it is not a process that occurs in
nature, 23.9%
Animal cloning will lead to human cloning, 20.6%
Cloning results in animals being viewed as “objects’ to be
produced as opposed to being
valuable in and of themselves, 14.9%
Animal cloning is morally wrong, 13.1%
Cloning will reduce genetic diversity to an unacceptable
level, 10.5%
The scientists and biotechnology companies who developed cloning technology
cannot be trusted to look out for my best interest, 7.7%
Cloning will result in unhealthy
farm animals, 5.1%
Meat and milk from clones and their offspring is unsafe to eat,
4.2%
Figure 1. Relative Importance of Competing Objections to Cloning
48
Table 7. Bivariate Correlations between Responses to Statements Related to the Acceptability
of Cloning
Some of
the meat
currently
sold in
grocery
stores is
from
cloned
animals
or their
offspring
The U.S.
governm
ent is
doing
everythin
g it can to
ensure
the safety
of food
products
I trust
informati
on about
cloning
from the
(USDA)
I am
willing to
eat meat
from
cloned
animals
The meat
from
cloned
animals
is safe to
eat
Some of the meat currently sold
in grocery stores is from cloned
animals or their offspring
1.00*a
The U.S. government is doing
everything it can to ensure the
safety of food products
0.16* 1.00
*
I trust information about cloning
from the U.S. Department of
Agriculture (USDA)
0.23* 0.66
* 1.00
*
I am willing to eat meat from
cloned animals
0.29* 0.45
* 0.62
* 1.00
*
The meat from cloned animals is
safe to eat
0.29* 0.51
* 0.65
* 0.79
* 1.00
*
Animal cloning is unacceptable -0.07* -0.22
* -0.35
* -0.55
* -0.48
*
Note: statistics are correlations between responses to questions, “To what extent do you agree or disagree with each
of the following statements?” Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither
agree nor disagree, 4=somewhat agree, and 5=strongly agree. aOne asterisk implies that the parameter is statistically different than zero at the 0.05 level or lower.
49
Table 8. Bivariate Correlations between Responses to Statements Related to the Acceptability
of Cloning
Responses to Agree/Disagree Statements
a
Relative Importance of
Objection to Cloningb
Some of
the meat
currently
sold in
grocery
stores is
from
cloned
animals
or their
offspring
The U.S.
governm
ent is
doing
everythin
g it can
to ensure
the
safety of
food
products
I trust
informati
on about
cloning
from the
USDA
I am
willing
to eat
meat
from
cloned
animals
The meat
from
cloned
animals
is safe to
eat
Animal
cloning
is
unaccept
able
Animal cloning is morally
wrong
-0.07*c
-0.07* -0.13
* -0.24
* -0.21
* 0.21
*
Meat and milk from clones and
their offspring is unsafe to eat
-0.11* -0.12
* -0.12
* -0.16
* -0.16
* 0.10
*
Animal cloning will lead to
human cloning
0.04 0.04 0.06* 0.09
* 0.08
* -0.04
Cloning will result in unhealthy
farm animals
-0.05* -0.13
* -0.13
* -0.13
* -0.13
* 0.03
Cloning is “unnatural” because
it is not a process that occurs in
nature
-0.06* 0.02 -0.01 -0.01 0.00 -0.01
Cloning will reduce genetic
diversity to an unacceptable
level
0.02 -0.03 0.01 0.04 0.02 -0.06*
Cloning results in animals being
viewed as “objects‟ to be
produced as opposed to being
valuable in and of themselves
0.12* 0.12
* 0.16
* 0.20
* 0.18
* -0.14
*
The scientists and
biotechnology companies who
developed cloning technology
cannot be trusted to look out for
my best interest
0.00 -0.14* -0.13
* -0.10
* -0.11
* 0.05
*
aResponses to questions, “To what extent do you agree or disagree with each of the following statements?”
bRelative importance of competing objections to cloning determined by calculating posterior probabilities from the
random parameter logit model fit to paired comparison choice data. cOne asterisk implies the parameter is statistically different than zero at the 0.05 level or lower.
50
Table 9. Relationship between Socio-Economic Characteristics and Cloning Concerns: Ordered
Probit Estimates
Dependent Variablea
Variable
I am
willing to
eat meat
from
cloned
animals
Some of
the meat
currently
sold in the
grocery
stores is
from
cloned
animals or
their
offspring
The U.S.
governme
nt is doing
everything
it can to
ensure the
safety of
food
products
The meat
from
cloned
animals is
safe to eat
Animal
cloning is
unaccepta
ble
I trust
informatio
n about
cloning
from the
(USDA)
Threshold Parameter1 0.618*b
(0.273)c
1.049*
(0.281)
0.433
(0.269)
0.800*
(0.272)
1.201*
(0.271)
0.511
(0.270)
Threshold Parameter2 -0.511*
(0.024)
-0.559*
(0.029)
-0.790*
(0.03)
-0.499*
(0.026)
-0.662*
(0.029)
-0.632*
(0.027)
Threshold Parameter3 -1.23*
(0.033)
-2.237*
(0.044)
-1.581*
(0.037)
-1.653*
(0.039)
-1.578*
(0.038)
-1.462*
(0.036)
Threshold Parameter4 -2.198*
(0.047)
-3.340*
(0.073)
-2.713*
(0.054)
-2.611*
(0.051)
-2.076*
(0.043)
-2.629*
(0.054)
Age 0.004*
(0.002)
0.003
(0.002)
0.003
(0.002)
0.006*
(0.002)
-0.003*
(0.002)
0.000
(0.002)
Treatment 0.018
(0.045)
-0.014
(0.047)
0.017
(0.045)
0.037
(0.045)
0.003
(0.045)
0.034
(0.045)
Gender -0.454*
(0.049)
-0.157*
(0.05)
-0.232*
(0.048)
-0.363*
(0.049)
0.336*
(0.049)
-0.276*
(0.048)
Income 0.002*
(0.001)
-0.001
(0.001)
0.001
(0.001)
0.001*
(0.001)
-0.001
(0.001)
0.001
(0.001)
Internet 0.240*
(0.055)
0.187*
(0.057)
0.200*
(0.054)
0.250*
(0.055)
-0.185*
(0.055)
0.252*
(0.054)
NoHSd 0.023
(0.087)
0.005
(0.09)
0.199*
(0.086)
-0.117
(0.087)
0.045
(0.087)
0.059
(0.086)
HSd -0.116
(0.067)
-0.113
(0.069)
0.097
(0.066)
-0.214*
(0.066)
0.184*
(0.067)
-0.034
(0.066)
SomeColleged 0.037
(0.064)
-0.048
(0.066)
0.083
(0.063)
0.024
(0.064)
0.037
(0.064)
-0.049
(0.064)
Whitee 0.056
(0.218)
-0.108
(0.223)
0.129
(0.214)
-0.009
(0.216)
-0.210
(0.213)
0.112
(0.214)
Blacke -0.220
(0.228)
0.126
(0.232)
0.030
(0.223)
-0.442*
(0.225)
-0.283
(0.223)
-0.079
(0.224)
Othere -0.143
(0.237)
0.133
(0.243)
0.187
(0.233)
-0.362
(0.235)
0.01
(0.233)
-0.006
(0.234)
51
Hispanice -0.15
(0.227)
-0.068
(0.232)
0.152
(0.222)
-0.233
(0.224)
-0.049
(0.222)
0.184
(0.223)
Northeastf -0.079
(0.072)
0.235*
(0.075)
0.03
(0.072)
-0.070
(0.072)
0.120
(0.072)
0.091
(0.072)
Midwestf -0.014
(0.07)
0.091
(0.072)
0.027
(0.069)
0.009
(0.070)
0.122
(0.070)
0.020
(0.070)
Southf -0.143*
(0.063)
0.136*
(0.065)
0.003
(0.062)
-0.123*
(0.063)
0.230*
(0.063)
-0.069
(0.062)
Meat:Neverg -0.304
(0.164)
-0.132
(0.169)
-0.200
(0.163)
-0.038
(0.163)
0.067
(0.165)
-0.038
(0.163)
Meat:Yearlyg -0.227
(0.138)
0.037
(0.144)
0.207
(0.137)
0.132
(0.138)
0.088
(0.14)
0.075
(0.138)
Meat:Monthlyg -0.025
(0.124)
0.005
(0.130)
0.148
(0.124)
0.137
(0.124)
0.223
(0.126)
0.157
(0.124)
Meat:Weeklyg 0.039
(0.121)
0.073
(0.127)
0.181
(0.120)
0.197
(0.121)
0.076
(0.123)
0.220
(0.121)
Farm 0.022
(0.063)
-0.004
(0.065)
0.016
(0.062)
0.059
(0.063)
-0.111
(0.063)
0.073
(0.062)
Pshopper 0.038
(0.054)
0.046
(0.056)
0.089
(0.054)
0.086
(0.054)
-0.132*
(0.055)
0.123
(0.054)
Child -0.072
(0.055)
-0.083
(0.057)
-0.006
(0.055)
-0.108*
(0.055)
0.105
(0.055)
-0.068
(0.055)
Log-Likelihood -3,322.4 -2,613.0 -3,276.8 -3,118.5 -3,341.7 -3,259.7
Number of
Observations
2,216 2,214 2,219 2,217 2,200 2,218
aDependent variable is response to question, “To what extent do you agree or disagree with each of the following
statements?” Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree,
4=somewhat agree, and 5=strongly agree. bOne asterisk indicates that the parameter is statistically different than zero at the 0.05 level or lower.
cNumbers in parentheses are standard errors
dParameter estimate compared to education of Bachelor‟s degree or higher
eParameter estimate compared to ethnicity of 2+ Races, non-hispanic
fParameter estimate compared to residents in West U.S Census Region
gParameter estimate compared to purchasing meat every day
52
Table 10. Food Values: Logit and Random Parameter Logit Estimates fit to Paired Comparison
Choices
Variable
Econometric
Estimates
Importance
Scores
Logit RPL
Meana
RPL
St.Dev.b Logit RPL
Constant 0.149*d
(0.028)c
0.172*
(0.026)
0.489*
(0.019)
Naturalness (extent to which food is
produced without modern
technologies)
0.243*
(0.067)
0.135*
(0.06)
1.892*
(0.055)
4.2% 6.3%
Taste (extent to which consumption
of the food is appealing to the
senses)
1.608*
(0.069)
1.699*
(0.059)
1.085*
(0.048)
16.4% 12.7%
Price (the price that is paid for the
food)
1.094*
(0.062)
1.080*
(0.052)
1.352*
(0.048)
9.8% 8.7%
Safety (extent to which consumption
of food will not cause illness)
2.310*
(0.064)
2.753*
(0.065)
2.043*
(0.070)
33.1% 35.5%
Convenience (ease with which food
is cooked and/or consumed)
0.346*
(0.056)
0.328*
(0.046)
1.565*
(0.051)
4.6% 5.5%
Nutrition (amount and type of fat,
protein, vitamins, etc.)
1.502*
(0.053)
1.679*
(0.048)
1.564*
(0.053)
14.7% 15.8%
Tradition (preserving traditional
consumption patterns)
-0.058*
(0.051)
-0.093*
(0.042)
1.248*
(0.047)
3.1% 2.9%
Origin (where the agricultural
commodities were grown)
0.156*
(0.049)
0.010
(0.041)
1.629*
(0.053)
3.8% 4.4%
Fairness (the extent to which all
parties involved in the production of
the food equally benefit)
-0.091*
(0.046)
-0.250*
(0.04)
1.451*
(0.050)
3.0% 3.1%
Appearance (extent to which food
looks appealing)
0.190*
(0.045)
0.156*
(0.035)
1.174*
(0.045)
4.0% 3.5%
Environmental Impact (effect of food
production on the environment)
0 0 0 3.3% 1.7%
aThe estimates refer to the estimated mean in the population from the random parameter logit model
bThe estimates refers to the estimated standard deviation of the food value in the population from the random
parameter logit model cNumbers in parentheses are standard errors
dOne asterisk implies that the parameter is statistically different than zero at the 0.05 level or lower.
Notes: Results based on 24,665 choices made by 2,255 individuals; Log-likelihood function values for logit and
random parameter logit models are -13,785.58 and -12,890.40, respectively; a likelihood ratio test could not reject
the hypothesis that the parameters were the same across the two information treatments
53
Safety , 35.5%
Nutrition , 15.8%Taste, 12.7%
Price, 8.7%
Naturalness, 6.3%
Convenience, 5.5%
Origin , 4.4%
Appearance , 3.5%
Fairness, 3.1%
Tradition , 2.9% Environmental Impact , 1.7%
Figure 2. Relative Importance of Competing Food Values
54
Table 11. Bivariate Correlations between Responses to Statements Related to the Acceptability
of Cloning
Responses to Agree/Disagree Statements
a
Relative
Importance of
Food Valuesb
Some of
the meat
currently
sold in
grocery
stores is
from
cloned
animals or
their
offspring
The U.S.
governme
nt is doing
everythin
g it can to
ensure the
safety of
food
products
I trust
informatio
n about
cloning
from the
USDA
I am
willing to
eat meat
from
cloned
animals
The meat
from
cloned
animals is
safe to eat
Animal
cloning is
unaccepta
ble
Naturalness 0.01 -0.14*c
-0.09* -0.08
* -0.09
* 0.04
Taste 0.04 0.01 0.09* 0.08
* 0.08
* -0.03
Price 0.00 0.07* 0.07
* 0.05
* 0.06
* -0.05
*
Safety -0.04 0.00 -0.05* -0.03 -0.04 0.05
*
Convenience 0.02 0.04* 0.06
* 0.05
* 0.05
* -0.05
*
Nutrition 0.01 0.02 0.02 -0.01 0.02 -0.05*
Tradition 0.02 0.01 0.00 -0.03 -0.02 0.04*
Origin 0.01 -0.13* -0.08
* -0.10
* -0.10
* -0.02
Fairness 0.03 -0.02 0.01 0.00 0.00 0.02
Appearance 0.08* 0.03 0.03 0.05
* 0.04 -0.04
*
Environmental
Impact 0.05
* -0.03 0.03 0.01 0.02 -0.03
aResponses to questions, “To what extent do you agree or disagree with each of the following statements?”
bRelative importance of competing food values determined by calculating posterior probabilities from the random
parameter logit model fit to paired comparison choice data. cOne asterisk implies that the parameter is statistically different than zero at the 0.05 level or lower.
55
Table 12. Percentage of Respondents Voting in Favor of Three Policies Related to Cloning at
Seven Price Levels
Percent Voting in Favor of . . .
Percent
Increase in
Food Prices
Mandatory
tracking of
cloned animals
Mandatory
labeling of
meat and milk
from cloned
animals
Ban on
practice of
animal cloning
5% 57.1% 65.0% 54.5%
10% 53.8% 63.9% 44.3%
15% 46.1% 52.1% 39.9%
25% 32.4% 46.7% 42.1%
50% 27.2% 38.6% 33.7%
75% 29.5% 31.4% 31.8%
100% 36.5% 33.2% 32.9% Note: The approximate number of observations in each cell is 315
56
Table 13. Probability of Affirmative Vote on Three Cloning Policies: Logit Estimates
Variable
Mandatory
tracking of
cloned animals
Mandatory
labeling of
cloned meat
and milk
Ban on
practice of
animal cloning
Constant (utility of policy vs. no policy) -0.012
(0.067)a
0.468*
(0.066)
-0.103
(0.068)
Percentage Price Increase -0.009*b
(0.001)
-0.015*
(0.001)
-0.008*
(0.001)
Willingness-to-Pay -1.3%
[-15.0%, 12.0%]
32.1%
[22.6%, 43.2%]
-13.2%
[-32.7%, 2.9%]
Log-Likelihood -1,469.5 -1,473.7 -1,466.4
Number of Observations 2,220 2,219 2,214 aNumbers in parentheses are standard errors.
bOne asterisk implies the parameter is statistically different than zero at the 0.05 level or lower
57
25%
30%
35%
40%
45%
50%
55%
60%
65%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Pe
rce
nt
Vo
tin
g In
Fav
or
of
Po
licy
Percent Increase in Food Price
Mandatory tracking of cloned animals
Mandatory labeling of cloned meat and milk
Ban on practice of animal cloning
Figure 3. Predicted Probabilities of Voting In Favor of Three Cloning Policies as a Function of
the Change in Food Price Resulting from the Policy
58
Table 14. Ground Beef Conjoint Models: Multinomial and Random Parameter Logit Estimates
Attribute
Multi-
nomial
Logit
RPL
Meana
RPL
St.Dev.b
Clone vs. None -0.967*d
(0.185)c
0.929*
(0.277)
0.112*
(0.077)
Offspring of Clone vs. None 0.649*
(0.185)
1.200*
(0.279)
0.593*
(0.077)
Non-Clone vs. None 1.357*
(0.185)
5.746*
(0.296)
3.922*
(0.098)
Percent Leanness 0.030*
(0.002)
0.057*
(0.003)
0.067*
(0.002)
Percent Saturated Fat 0.090*
(0.002)
-0.197*
(0.006)
0.056*
(0.011)
Price -0.559*
(0.011)
-1.140*
(0.019)
0
aThe estimates refer to the estimated mean in the population from the random parameter logit model
bThe estimates refers to the estimated standard deviation in the population from the random parameter logit model
cNumbers in parentheses are standard errors
dOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
Notes: Results based on 26,536 choices made by 2,243 individuals; Log-likelihood function values for multinomial
logit and random parameter logit models are -24,381.91 and -14,897.62, respectively.
59
Table 15. Willingness-to-Pay for Selected Ground Beef Attributes Calculated from Conjoint
Estimates
Willingness-to-Pay ($/choice) for . . . MNL
Mean
RPL
Mean
RPL
St.Dev.
Non-Clone vs. Cloned $4.16 $4.23 $3.44
Non-Clone vs. Offspring of Clone $3.59 $3.99 $3.48
Offspring of Clone vs. Clone $0.57 $0.24 $0.53
Increase in Leanness (90% vs. 80%) $0.54 $0.50 $0.59
Reduction in Saturated Fat (5% vs. 10%) $0.80 $0.86 $0.25
60
Figure 4. Example Output from Ground Beef Market Simulator
61
Table 16. Milk Conjoint Models: Multinomial and Random Parameter Logit Estimates
Variable
Multi-
nomial
Logit
RPL
Meana
RPL
St.Dev.b
Clone vs. None 0.664*d
(0.035)c
1.668*
(0.069)
2.232*
(0.051)
Offspring of Clone vs. None 0.530*
(0.035)
1.591*
(0.070)
2.539*
(0.060)
Non-Clone vs. None 2.240*
(0.034)
4.333*
(0.069)
2.829*
(0.049)
rBST vs. no RBST -0.602*
(0.017)
-1.323*
(0.037)
1.626*
(0.049)
Fat Content: 1% vs. Skim -0.123*
(0.024)
-0.274*
(0.037)
0.550*
(0.057)
Fat Content: 2% vs. Skim 0.261*
(0.023)
0.285*
(0.039)
1.197*
(0.048)
Fat Content: Whole vs. Skim 0.171*
(0.023)
-0.545*
(0.050)
2.054*
(0.051)
Price -0.378*
(0.006)
-0.734*
(0.010)
aThe estimates refer to the estimated mean in the population from the random parameter logit model
bThe estimates refers to the estimated standard deviation in the population from the random parameter logit model
cNumbers in parentheses are standard errors
dOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
Notes: Results based on 35,373 choices made by 2,237 individuals; Log-likelihood function values for multinomial
logit and random parameter logit models are -39,091.28 and -26,187.91, respectively.
62
Table 17. Willingness-to-Pay for Selected Milk Attributes Calculated from Conjoint Estimates
Willingness-to-Pay ($/choice) for . . . MNL
Mean
RPL
Mean
RPL
St.Dev.
Non-Clone vs. Cloned $4.17 $3.63 $4.91
Non-Clone vs. Offspring of Clone $4.53 $3.73 $5.18
Offspring of Clone vs. Clone -$0.35 -$0.10 $4.60
No rBST vs. rBST $1.59 $1.80 $2.21
63
Figure 5. Example Output from Milk Market Simulator
64
Figure 6. Example Output from Milk Market Simulator
65
Table 18. Characteristics of Survey Respondents from Three Survey Samples
Variable Percent in Each Category U.S.
Census
KN
Un-
weighted
IRI
Un-
weighted
Nielsen
Un-
weighted
Age1 18-34 years 30.7% 20.9% 16.7% 15.0%
Age2 35-44 years 19.2% 18.4% 23.2% 18.4%
Age3 45-54 years 19.5% 19.7% 24.4% 26.7%
Age4 55-64 years 14.5% 21.9% 18.7% 25.9%
Age5 65+ years 16.2% 19.1% 17.0% 14.0%
HS High School Degree or Less 46.6% 41.7% 19.5% 33.0%
SomeCollege Some College 27.2% 27.4% 38.6% 33.3%
Bachelors Bachelor‟s degree or higher 26.2% 30.9% 42.0% 33.7%
Female Female 51.6% 49.2% 78.2% 68.3%
White White, non-Hispanic 69.1% 78.5% 76.8% 74.7%
Black Black, non-Hispanic 11.7% 6.9% 11.8% 10.0%
Hispanic Hispanic 13.3% 7.6% 5.6% 10.3%
Other Other race 6.0% 7.0% 5.9% 5.0%
Northeast Northeast U.S Census Region 18.1% 18.5% 20.3% 17.4%
Midwest Midwest U.S Census Region 22.0% 22.5% 23.7% 23.6%
South South U.S Census Region 36.6% 36.5% 35.5% 37.3%
West West U.S Census Region 23.2% 22.4% 20.5% 21.7%
Inc1 Annual HH income less than $25,000 25.3% 17.9% 18.6% 23.9%
Inc2 Annual HH income $25,000 to $99,999 55.6% 65.9% 65.8% 59.0%
Inc3 Annual HH income $100,000 or more 19.1% 16.2% 15.6% 17.2%
Meat:Never never purchase meat
3.5% 1.2% 1.5%
Meat:Yearly purchase meat a few times a year
9.0% 6.9% 6.4%
Meat:Monthly purchase meat about once a month
27.8% 31.0% 29.2%
Meat:Weekly purchase meat about once a week
55.8% 58.5% 60.0%
Meat:Day purchase meat every day
3.4% 2.3% 2.9%
Farm Own/work on ranch/farm
16.4% 10.8% 8.7%
Pshopper Primary shopper for food
68.8% 97.2% 97.9%
Child Child under age of 12 in household
23.7% 29.0% 24.7%
N
2,256 1,691 2,120
66
Table 19. Food Values: Logit Estimates fit to Paired Comparison Choices: Comparison across
Three Survey Samples
Variable
Econometric
Estimates
Importance
Scores
KNa IRI
b Nielsen
c KN
a IRI
b Nielsen
c
Constant 0.149*c
(0.028)d
0.186*
(0.017)
0.094*
(0.015)
Naturalness (extent to which food is
produced without modern
technologies)
0.243*
(0.067)
0.037
(0.051)
0.059
(0.051)
4.2% 3.8% 3.6%
Taste (extent to which consumption
of the food is appealing to the
senses)
1.608*
(0.069)
1.445*
(0.053)
1.668*
(0.057)
16.4% 15.7% 18.0%
Price (the price that is paid for the
food)
1.094*
(0.062)
1.181*
(0.052)
1.205*
(0.055)
9.8% 12.1% 11.3%
Safety (extent to which consumption
of food will not cause illness)
2.310*
(0.064)
2.015*
(0.058)
2.186*
(0.061)
33.1% 27.8% 30.2%
Convenience (ease with which food
is cooked and/or consumed)
0.346*
(0.056)
0.366*
(0.05)
0.566*
(0.054)
4.6% 5.4% 6.0%
Nutrition (amount and type of fat,
protein, vitamins, etc.)
1.502*
(0.053)
1.498*
(0.054)
1.335*
(0.056)
14.7% 16.6% 12.9%
Tradition (preserving traditional
consumption patterns)
-0.058*
(0.051)
-0.018
(0.050)
0.092
(0.053)
3.1% 3.6% 3.7%
Origin (where the agricultural
commodities were grown)
0.156*
(0.049)
-0.068
(0.051)
-0.012
(0.054)
3.8% 3.5% 3.4%
Fairness (the extent to which all
parties involved in the production of
the food equally benefit)
-0.091*
(0.046)
-0.133*
(0.051)
-0.098
(0.054)
3.0% 3.2% 3.1%
Appearance (extent to which food
looks appealing)
0.190*
(0.045)
0.192*
(0.049)
0.266*
(0.054)
4.0% 4.5% 4.4%
Environmental Impact (effect of food
production on the environment)
0 0 0 3.3% 3.7% 3.4%
a Results based on 24,665 choices; Log-likelihood function value = -13,785.58; a likelihood ratio test could not
reject the hypothesis that the parameters were the same across the two information treatments in the KN sample. bResults based on 18,601 choices; Log-likelihood function value = -10,709.4.
cResults based on 23,320 choices; Log-likelihood function value = -13,417.29.
cOne asterisk implies that the parameter is statistically different than zero at the 0.05 level or lower.
dNumbers in parentheses are standard errors.
Note: The log-likelihood function value for a model where parameters are pooled across all three samples is
-37,990.8 ( N=66,586). A likelihood ratio test indicates that we can reject the null hypothesis that parameters are
equal across all three samples (Chi-square value =157.05; df=22; P-value<0.001).
67
Table 20. Knowledge of Assisted Reproduction Technologies That Are Used to Breed Animals
for Meat and Milk Production from Three Survey Samples
Technology
KN
Treatment 1
Prior
Information
KN
Treatment 2
No Prior
Information
IRI Nielsen
Artificial Insemination 2.50
(1.15)
[21.3%]
2.37
(1.13)
[24.5%]
2.58
(1.24)
[22.4%]
2.44
(1.22)
[25.8%]
In vitro fertilization 2.25
(1.09)
[28.1%]
2.16
(1.06)
[31.5%]
2.36
(1.20)
[29.2%]
2.22
(1.19)
[34.0%]
Biotechnology 2.05
(1.07)
[38.2%]
2.04
(1.04)
[36.0%]
2.17
(1.13)
[34.2%]
2.02
(1.09)
[40.3%]
Embryo transfer 1.94
(1.07)
[43.7%]
1.86
(1.04)
[46.6%]
1.95
(1.13)
[46.9%]
1.85
(1.07)
[49.8%]
Cloning 2.70
(1.03)
[7.1%]
2.45
(1.01)
[14.09%]
2.79
(1.14)
[10.3%]
2.53
(1.11)
[16.4%]
Number of Observations 1,123 1,133 1,691 2,120
Note: response to question: “Overall, how much have you heard or read about each of the following assisted
reproduction technologies that are sometimes used to breed animals for meat and milk production?” Response
categories were: 1 = nothing at all, 2 = a little, 3 = a moderate amount, 4 = quite a bit, 5 = a great deal. aP-value from two-sample t-test that means are equivalent across treatments
bNumbers in parentheses ( ) are standard deviations
cNumbers in brackets [ ] are the percentage of respondents indicating 1=nothing at all
68
Table 21. Willingness-to-Eat Cloned Meat and Milk from Three Survey Samples
Statement KN
Pooled IRI
Nielsen
I am willing to eat meat from
cloned animals
2.72
(1.28)a
2.46
(1.24)
2.86
(1.33)
The average American is willing
to eat meat from cloned animals
2.76
(0.98)
2.58
(0.99)
2.72
(0.99)
I am willing to eat meat from the
offspring of cloned animals
2.72
(1.29)
2.48
(1.22)
2.85
(1.32)
I am willing to consume milk
products from cloned animals
2.70
(1.29)
2.47
(1.20)
2.87
(1.32)
I am willing to consume milk
products from the offspring of
cloned animals
2.73
(1.29)
2.56
(1.24)
2.87
(1.33)
If I learned that the meat
products I regularly purchase
came from cloned animals, I
would continue to buy the meat
products as usual
2.78
(1.29)
2.59
(1.30)
2.95
(1.34)
If I learned that the milk
products I regularly purchase
came from cloned animals, I
would continue to buy the milk
products as usual
2.78
(1.31)
2.62
(1.29)
2.93
(1.36)
Note: mean response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aNumbers in parentheses are standard deviations
Note: given the sample sizes and observed variability, the estimated means ratings are expected to be within plus or
minus roughly 0.07 of the true population means with 95% confidence.
69
Table 22. Willingness-to-Eat Cloned Meat and Milk from Three Survey Samples
Statement KN IRI Nielsen
Da
Nb
Ac
D
N A
D
N
A
I am willing to eat meat
from cloned animals
43.2% 26.0% 30.8% 48.1% 31.4% 20.6% 37.9% 27.9% 34.1%
The average American is
willing to eat meat from
cloned animals
35.1% 44.2% 20.7% 40.2% 44.4% 15.4% 39.2% 39.8% 21.1%
I am willing to eat meat
from the offspring of cloned
animals
43.0% 26.1% 30.9% 47.4% 31.4% 21.2% 38.2% 28.2% 33.6%
I am willing to consume
milk products from cloned
animals
44.4% 24.8% 30.8% 48.4% 31.4% 20.2% 38.1% 27.5% 34.4%
I am willing to consume
milk products from the
offspring of cloned animals
43.0% 25.7% 31.3% 45.9% 31.4% 22.8% 38.4% 26.4% 35.2%
If I learned that the meat
products I regularly
purchase came from cloned
animals, I would continue to
buy the meat products as
usual
41.4% 25.7% 32.9% 47.0% 26.3% 26.6% 36.4% 24.6% 39.0%
If I learned that the milk
products I regularly
purchase came from cloned
animals, I would continue to
buy the milk products as
usual
42.1% 24.6% 33.3% 44.2% 29.8% 26.0% 37.2% 24.3% 38.5%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
bPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
cPercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
Note: given the sample sizes, the reported percentages are estimated to be within plus or minus about 2.2% of the
true population percentages with 95% confidence.
70
Table 23. Beliefs about Safety and Acceptability of Cloned Meat and Milk from Three Survey
Samples
Statement KN
Pooled IRI
Nielsen
Some of the meat currently
sold in grocery stores is from
cloned animals or their
offspring
2.79
(0.88)a
2.63
(0.94)
2.89
(0.94)
Animal cloning is unacceptable 3.03
(1.26)
3.34
(1.26)
3.01
(1.28)
Animal cloning will result in
beneficial outcomes to me
2.71
(1.09)
2.51
(1.09)
2.82
(1.14)
The meat from cloned animals
is safe to eat
2.94
(1.12)
2.78
(1.08)
3.10
(1.17)
In general, the meat and milk I
buy from grocery stores is safe
to eat
3.68
(0.92)
3.81
(0.92)
3.91
(0.89)
Note: mean response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aNumbers in parentheses are standard deviations
Note: given the sample sizes and observed variability, the estimated means ratings are expected to be within plus or
minus roughly 0.07 of the true population means with 95% confidence.
71
Table 24. Beliefs about Safety and Acceptability of Cloned Meat and Milk from Three Survey
Samples
Statement KN IRI Nielsen
Da
Nb
Ac
D
N
A
D
N
A
Some of the meat currently
sold in grocery stores is
from cloned animals or their
offspring
27.4% 57.5% 15.1% 32.0% 57.7% 10.4% 22.7% 59.7% 17.6%
Animal cloning is
unacceptable
34.4% 33.7% 31.9% 24.9% 35.1% 40.1% 35.5% 30.5% 34.0%
Animal cloning will result in
beneficial outcomes to me
35.8% 43.6% 20.6% 41.5% 43.4% 15.1% 33.2% 40.8% 26.0%
The meat from cloned
animals is safe to eat
29.2% 41.2% 29.6% 31.6% 47.0% 21.3% 24.8% 39.9% 35.3%
In general, the meat and
milk I buy from grocery
stores is safe to eat
10.3% 26.1% 63.6% 8.4% 22.2% 69.4% 6.9% 17.9% 75.2%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
bPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
cPercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
Note: given the sample sizes, the reported percentages are estimated to be within plus or minus about 2.2% of the
true population percentages with 95% confidence.
72
Table 25. Perceptions about the Federal Government and Cloned Meat and Milk from Three
Survey Samples
Statement KN
Pooled IRI
Nielsen
The U.S. government is doing
everything it can to ensure the
safety of food products
2.80
(1.12)a
2.89
(1.14)
2.96
(1.19)
The U.S. government can trace
the meat from cloned animals
back to the farm on which the
animal lived
2.98
(1.05)
3.02
(1.05)
3.22
(1.07)
Animal cloning is carefully
regulated by the U.S.
government
2.71
(1.01)
2.56
(1.03)
2.82
(1.09)
I trust the U.S. government to
properly regulate the use of
animal cloning
2.60
(1.15)
2.65
(1.19)
2.57
(1.19)
I trust information about cloning
from the U.S. Department of
Agriculture (USDA)
2.76
(1.16)
2.72
(1.15)
2.72
(1.18)
I trust information about cloning
from the U.S. Food and Drug
Administration (FDA)
2.74
(1.15)
2.69
(1.14)
2.70
(1.19)
I trust information about cloning
from U.S. Environmental
Protection Agency (EPA)
2.70
(1.14)
2.73
(1.13)
2.69
(1.18)
I trust information about cloning
from University scientists and
researchers
2.89
(1.12)
2.89
(1.13)
3.00
(1.15)
Note: mean response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aNumbers in parentheses are standard deviations.
Note: given the sample sizes and observed variability, the estimated means ratings are expected to be within plus or
minus roughly 0.07 of the true population means with 95% confidence.
73
Table 26. Perceptions about the Federal Government and Cloned Meat and Milk from Three
Survey Samples
Statement KN IRI Nielsen
Da
Nb
Ac
D
N
A
D
N
A
The U.S. government is
doing everything it can to
ensure the safety of food
products
40.7% 29.7% 29.6% 34.4% 34.9% 30.7% 35.9% 26.4% 37.7%
The U.S. government can
trace the meat from cloned
animals back to the farm on
which the animal lived
27.6% 42.8% 29.6% 24.6% 45.6% 29.8% 20.6% 39.8% 39.6%
Animal cloning is carefully
regulated by the U.S.
government
37.3% 42.7% 20.0% 40.2% 45.2% 14.6% 33.0% 43.4% 23.6%
I trust the U.S. government
to properly regulate the use
of animal cloning
47.1% 29.0% 24.0% 44.6% 31.2% 24.2% 48.5% 27.3% 24.2%
I trust information about
cloning from the U.S.
Department of Agriculture
(USDA)
40.0% 30.7% 29.3% 38.8% 35.0% 26.2% 42.3% 31.0% 26.8%
I trust information about
cloning from the U.S. Food
and Drug Administration
(FDA)
41.3% 29.8% 28.8% 40.8% 34.2% 24.9% 42.2% 31.0% 26.8%
I trust information about
cloning from U.S.
Environmental Protection
Agency (EPA)
41.7% 32.2% 26.1% 37.9% 37.9% 24.2% 42.1% 32.0% 25.9%
I trust information about
cloning from University
scientists and researchers
34.5% 33.5% 32.0% 32.5% 37.7% 29.8% 30.9% 32.7% 36.5%
Note: response to question, “To what extent do you agree or disagree with each of the following statements?”
Response categories were: 1=strongly disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat
agree, and 5=strongly agree. aPercentage of respondents in the pooled sample indicating 1=strongly disagree or 2= somewhat disagree
bPercentage of respondents in the pooled sample indicating 3=neither agree nor disagree
cPercentage of respondents in the pooled sample indicating 4=somewhat agree or 5=strongly agree
Note: given the sample sizes, the reported percentages are estimated to be within plus or minus about 2.2% of the
true population percentages with 95% confidence.
74
Table 27. Relative Importance of Competing Objections to Cloning: Logit Model fit to Paired
Comparison Choices; Comparison of Three Samples
Variable
Econometric
Estimates
Importance
Scores
KNa
IRIb
Nielsenc
KN IRI Nielsen
Intercept (order effect) -0.056
(0.032)d
0.018*
(0.020)
-0.036*
(0.016)
Cloning is “unnatural” because it is not a
process that occurs in nature
0.543*e
(0.050)
0.660*
(0.051)
0.135*
(0.051)
24.5% 27.1% 22.0%
Animal cloning will lead to human
cloning
-0.073
(0.059)
-0.284*
(0.050)
-0.45*
(0.05)
13.2% 10.5% 12.2%
Cloning results in animals being viewed
as “objects‟ to be produced as opposed
to being valuable in and of themselves
-0.060
(0.042)
-0.020*
(0.062)
-0.321*
(0.051)
13.4% 13.7% 13.9%
Animal cloning is morally wrong -0.476*
(0.071)
-0.538*
(0.049)
-1.023*
(0.049)
8.8% 8.2% 6.9%
Cloning will reduce genetic diversity to
an unacceptable level
-0.081
(0.045)
-0.067*
(0.049)
-0.394*
(0.051)
13.1% 13.1% 13.0%
The scientists and biotechnology
companies who developed cloning
technology cannot be trusted to look out
for my best interest
0 0 0
14.2% 14.0% 19.2%
Cloning will result in unhealthy farm
animals
-0.525*
(0.053)
-0.425*
(0.049)
-0.796*
(0.051)
8.4% 9.2% 8.7%
Meat and milk from clones and their
offspring is unsafe to eat
-1.208*
(0.066)
-1.216*
(0.053)
-1.538*
(0.053)
4.3% 4.2% 4.1%
aResults based on 17,434 choices made by 2,231 individuals; Log-likelihood function value = -11,052.71; a
likelihood ratio test could not reject the hypothesis that the parameters were the same across the two information
treatments in the Knowledge Networks sample bResults based on 13,528 choices; Log-likelihood function value = -8,597.77.
cResults based on 16,960 choices; Log-likelihood function value = -10,670.71
dNumbers in parentheses are standard errors
eOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
Note: The log-likelihood function value for a model where parameters are pooled across all three samples is
-30,355.36 ( N=47,922). A likelihood ratio test indicates that we can reject the null hypothesis that parameters are
equal across all three samples (Chi-square value =68.34; df=16; P-value<0.001).
75
Table 28. Relationship between Stated and Actual Milk Purchases, Socio-Economic
Characteristics and Stated Willingness to Drink Cloned Milk: Ordered Probit Estimates for Three
Survey Samplesa
Variable KN
IRI
(stated milk
consumption)
IRI
(actual milk
consumption)
Nielsen
(stated milk
consumption)
Nielsen
(actual milk
consumption)
Intercept 0.927**b
(0.187)c
-0.071
(0.259)
0.859
(0.221)
0.718**
(0.195)
0.583**
(0.137)
Intercept2 -0.515**
(0.024)
-0.52**
(0.029)
-0.531
(0.029)
-0.492**
(0.025)
-0.488**
(0.025)
Intercept3 -1.199**
(0.033)
-1.443**
(0.044)
-1.417
(0.044)
-1.237**
(0.034)
-1.23**
(0.034)
Intercept4 -2.156**
(0.046)
-2.323**
(0.063)
-2.305
(0.064)
-2.032
(0.044)
-2.021**
(0.044)
Male 0.479**
(0.046)
0.126*
(0.06)
0.108
(0.061)
0.518
(0.050)
0.508**
(0.050)
Age1 -0.128
(0.078)
-0.098
(0.099)
-0.088
(0.102)
0.125
(0.082)
0.144
(0.082)
Age2 -0.101
(0.084)
-0.22*
(0.104)
-0.24*
(0.105)
-0.179
(0.083)
-0.165*
(0.082)
Age3 -0.003
(0.08)
-0.109
(0.099)
-0.141
(0.100)
-0.102
(0.078)
-0.098
(0.078)
Age4d
-0.047
(0.078)
0.050
(0.105)
0.038
(0.106)
-0.144
(0.084)
-0.143
(0.084)
Whitee
0.149
(0.096)
-0.008
(0.122)
-0.07
(0.131)
0.149
(0.101)
0.172
(0.101)
Blacke -0.127
(0.117)
-0.583**
(0.145)
-0.551**
(0.154)
-0.378**
(0.119)
-0.378**
(0.119)
Hispanice -0.04
(0.114)
-0.237
(0.141)
-0.362*
(0.149)
-0.080
(0.115)
-0.047
(0.115)
HSf
-0.199**
(0.062)
-0.215**
(0.075)
-0.274**
(0.076)
-0.188**
(0.061)
-0.194**
(0.061)
SomeCollegef 0.021
(0.063)
-0.191*
(0.078)
-0.192*
(0.079)
-0.139*
(0.065)
-0.143*
(0.065)
Inc1g
-0.306**
(0.083)
-0.209
(0.110)
-0.241*
(0.111)
-0.050
(0.082)
-0.077
(0.081)
Inc2g
-0.14*
(0.069)
0.026
(0.095)
-0.084
(0.097)
-0.042
(0.07)
-0.051
(0.07)
Northeasth
-0.086
(0.072)
-0.147
(0.091)
-0.179
(0.094)
-0.008
(0.073)
0.009
(0.073)
Midwesth 0.004
(0.07)
0.082
(0.087)
0.06
(0.088)
0.126
(0.071)
0.123
(0.07)
Southh -0.136*
(0.063)
0.155*
(0.078)
0.151
(0.08)
0.120
(0.063)
0.138*
(0.063)
76
Children -0.08
(0.058)
0.026
(0.073)
0.062
(0.074)
-0.010
(0.061)
0.001
(0.061)
Farm 0.034
(0.063)
0.241*
(0.101)
0.273**
(0.101)
-0.062
(0.084)
-0.060
(0.084)
Milk:Neveri -0.536**
(0.163)
0.561*
(0.239)
-0.693**
(0.189)
Milk:Yearlyi -0.443**
(0.157)
0.689**
(0.226)
-0.202
(0.168)
Milk:Monthlyi -0.116
(0.142)
1.061**
(0.19)
-0.080
(0.144)
Milk:Weeklyi -0.128
(0.135)
0.672**
(0.175)
-0.031
(0.134)
Milk purchases
(gallons/year)j
-0.001
(0.001)
Organic share of
milk purchasesk
-0.885**
(0.174)
Milk purchases
(purchase units/year)l
0.004
(0.002) aDependent variable is response to question, “To what extent do you agree or disagree with each of the following
statement? I am willing to consume milk products from cloned animals.” Response categories were: 1=strongly
disagree, 2= somewhat disagree, 3=neither agree nor disagree, 4=somewhat agree, and 5=strongly agree. bOne and two asterisks indicates statistical significance at the 0.05 and 0.01 levels, respectively
cNumbers in parentheses are standard errors
d Variable defined in table 18; Parameter estimates relative to age 65 and over
eParameter estimate compared to “other races”
fParameter estimate compared to education of Bachelor‟s degree or higher
gVariable defined in table 18; parameter estimate compared to households making more than $100,000/year
hParameter estimate compared to residents in West U.S Census Region
i Dummy variables, with parameter estimates compared to those who purchase milk every day
jAnnual volume (gallons) of milk purchased determined from scanner data
kShare of total annual volume of milk purchased (gallons) resulting from sales of organic milk
lAnnual volume (in units purchased – not controlling for unit size) of milk purchased determined from scanner data
77
Table 29. Relationship between Stated and Actual Meat Purchases, Socio-Economic
Characteristics and Stated Willingness to Eat Cloned Beef: Ordered Probit Estimates for Three
Survey Samplesa
Variable KN
IRI
(stated milk
consumption)
IRI
(actual milk
consumption)
Nielsen
(stated milk
consumption)
Nielsen
(actual milk
consumption)
Intercept 0.764**
(0.170)
0.497
(0.267)
0.527*
(0.217)
0.979**
(0.208)
0.387*
(0.167)
Intercept2 -0.510**
(0.024)
-0.405**
(0.025)
-0.411**
(0.026)
-0.459**
(0.024)
-0.455**
(0.024)
Intercept3 -1.224**
(0.033)
-1.309**
(0.042)
-1.278**
(0.042)
-1.224**
(0.034)
-1.214**
(0.034)
Intercept4 -2.186**
(0.047)
-2.129**
(0.060)
-2.108**
(0.061)
-2.009**
(0.044)
-1.99**
(0.043)
Male 0.434**
(0.046)
0.214**
(0.060)
0.195**
(0.061)
0.562**
(0.051)
0.585**
(0.051)
Age1 -0.083
(0.077)
0.053
(0.099)
0.007
(0.105)
0.126
(0.083)
0.137
(0.083)
Age2 -0.102
(0.084)
-0.194
(0.105)
-0.251*
(0.107)
-0.197*
(0.083)
-0.193*
(0.083)
Age3 0.005
(0.08)
-0.090
(0.100)
-0.091
(0.101)
-0.126
(0.079)
-0.113
(0.078)
Age4d
-0.043
(0.078)
0.033
(0.106)
0.011
(0.107)
-0.165
(0.084)
-0.17*
(0.084)
Whitee
0.174
(0.095)
-0.094
(0.123)
-0.133
(0.132)
0.173
(0.102)
0.186
(0.101)
Blacke -0.147
(0.116)
-0.558**
(0.147)
-0.578**
(0.155)
-0.299*
(0.119)
-0.295*
(0.119)
Hispanice -0.029
(0.112)
-0.341*
(0.143)
-0.380*
(0.150)
-0.022
(0.116)
-0.001
(0.115)
HSf
-0.142*
(0.062)
-0.314**
(0.076)
-0.287**
(0.077)
-0.169**
(0.061)
-0.165**
(0.061)
SomeCollegef 0.002
(0.063)
-0.208**
(0.079)
-0.195*
(0.079)
-0.124
(0.065)
-0.119
(0.065)
Inc1g
-0.327**
(0.083)
-0.016
(0.111)
-0.121
(0.112)
-0.047
(0.082)
-0.081
(0.081)
Inc2g
-0.155*
(0.069)
0.056
(0.095)
-0.025
(0.096)
-0.024
(0.070)
-0.029
(0.070)
Northeasth
-0.085
(0.072)
-0.182*
(0.092)
-0.177
(0.095)
0.004
(0.073)
0.002
(0.073)
Midwesth -0.032
(0.07)
0.123
(0.087)
0.134
(0.088)
0.118
(0.071)
0.128
(0.070)
Southh -0.136*
(0.063)
0.140
(0.078)
0.183*
(0.079)
0.128*
(0.063)
0.131*
(0.063)
78
Children -0.075
(0.057)
0.135
(0.072)
0.088
(0.073)
0.013
(0.060)
0.013
(0.060)
Farm 0.014
(0.063)
0.260*
(0.101)
0.254*
(0.102)
0.066
(0.085)
0.081
(0.085)
Meat:Neveri -0.294
(0.163)
-1.242**
(0.313)
-1.684**
(0.251)
Meat:Yearlyi -0.209
(0.138)
-0.137
(0.199)
-0.595**
(0.155)
Meat:Monthlyi 0.008
(0.123)
-0.106
(0.172)
-0.539**
(0.131)
Meat:Weeklyi 0.075
(0.12)
-0.070
(0.167)
-0.473**
(0.128)
Breakfast meat purchases
(pounds/year)j
0.001
(0.003)
Lunch meat purchases
(pounds/year)k
0.001
(0.003)
Fresh beef purchases
(purchase units/year)l
0.010*
(0.004) aDependent variable is response to question, “To what extent do you agree or disagree with each of the following
statement? I am willing to eat meat from cloned animals.” Response categories were: 1=strongly disagree, 2=
somewhat disagree, 3=neither agree nor disagree, 4=somewhat agree, and 5=strongly agree. bOne and two asterisks indicates statistical significance at the 0.05 and 0.01 levels, respectively
cNumbers in parentheses are standard errors
d Variable defined in table 18; Parameter estimates relative to age 65 and over
eParameter estimate compared to “other races”
fParameter estimate compared to education of Bachelor‟s degree or higher
gVariable defined in table 18; parameter estimate compared to households making more than $100,000/year
hParameter estimate compared to residents in West U.S Census Region
i Dummy variables, with parameter estimates compared to those who purchase meat every day
jAnnual volume (pounds) of breakfast meat purchased determined from scanner data
kAnnual volume (pounds) of lunch meat purchased determined from scanner data
lAnnual volume (in units purchased – not controlling for unit size) of beef purchased determined from scanner data
79
Table 30. Probability of Affirmative Vote on Policy Related to Mandatory Tracking of Cloned
Animals; Comparison across Three Samples
Variable KNa IRI Nielsen
Constant (utility of policy vs. no policy) -0.012
(0.067)b
0.716*
(0.078)
0.362*
(0.073)
Percentage Price Increase -0.009*c
(0.001)
-0.012*
(0.001)
-0.011*
(0.001)
Willingness-to-Pay -1.3%
[-15.6%, 12.9%]
61.7%*
[51.7%, 71.3% ]
32.0%*
[24.0%, 40.1%]
Log-Likelihood -1,469.5 -1,126.3 -1,428.8
Number of Observations 2,220 1,691 2,120 aA likelihood ratio test could not reject the hypothesis that the parameters were the same across the two information
treatments in the KN sample.
bNumbers in parentheses are standard errors.
cOne asterisk implies the parameter is statistically different than zero at the 0.05 level or lower
80
Table 31. Probability of Affirmative Vote on Policy Related to Mandatory Labeling of Cloned
Meat and Milk; Comparison across Three Samples
Variable KNa
IRI Nielsen
Constant (utility of policy vs. no policy) 0.468*c
(0.066)b
0.757*
(0.080)
0.248*
(0.075)
Percentage Price Increase -0.015*
(0.001)
-0.008*
(0.001)
-0.008*
(0.001)
Willingness-to-Pay 32.1%*
[26.1%, 37.9%]
98.5%*
[73.4%, 123.5%]
32.2%*
[20.4%, 44.1%]
Log-Likelihood -1,469.5 -1,117.2 -1,451.3
Number of Observations 2,220 1,691 2,120 aA likelihood ratio test could not reject the hypothesis that the parameters were the same across the two information
treatments in the KN sample.
bNumbers in parentheses are standard errors.
cOne asterisk implies the parameter is statistically different than zero at the 0.05 level or lower
81
Table 32. Probability of Affirmative Vote on Policy Related to Ban on Practice of Animal
Cloning; Comparison across Three Samples
Variable KNa
IRI Nielsen
Constant (utility of policy vs. no policy) -0.103
(0.068)b
0.063
(0.077)
-0.020
(0.077)
Percentage Price Increase -0.008*c
(0.001)
-0.005*
(0.001)
-0.011*
(0.001)
Willingness-to-Pay -13.2%
[-33.7%, 7.4%]
12.4%
[-12.4%, 37.2%]
-1.9%
[-16.1%, 12.4%]
Log-Likelihood -1,466.4 -1160.8 -1,366.6
Number of Observations 2,214 1,691 2,120 aA likelihood ratio test could not reject the hypothesis that the parameters were the same across the two information
treatments in the KN sample.
bNumbers in parentheses are standard errors.
cOne asterisk implies the parameter is statistically different than zero at the 0.05 level or lower
82
Table 33. Ground Beef Conjoint Models: Multinomial Logit Estimates; Comparison across
Three Samples
Attribute KNa
IRIb
Nielsenc
Clone vs. None -0.967*e
(0.185)d
-0.157
(0.225)
0.166
(0.187)
Offspring of Clone vs. None -0.649*
(0.185)
0.057*
(0.225)
0.418*
(0.187)
Non-Clone vs. None 1.357*
(0.185)
2.439*
(0.225)
2.395*
(0.188)
Percent Leanness 0.030*
(0.002)
0.019*
(0.002)
0.023*
(0.002)
Percent Saturated Fat -0.090*
(0.002)
-0.057*
(0.004)
-0.084*
(0.004)
Price -0.559*
(0.011)
-0.681*
(0.014)
-0.683*
(0.011)
aResults based on 26,536 choices made by 2,243 individuals; Log-likelihood function value = -24,381.91.
bResults based on 20,292 choices; Log-likelihood function value = -29,341.8.
cResults based on 25,440 choices; Log-likelihood function value = -30,725.73.
dNumbers in parentheses are standard errors
eOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
Note: The log-likelihood function value for a model where parameters are pooled across all three samples is
-91,917.68 (N=72,268). A likelihood ratio test indicates that we can reject the null hypothesis that parameters are
equal across all three samples (Chi-square value =14,936.47; df=12; P-value<0.001).
83
Table 34. Willingness-to-Pay for Selected Ground Beef Attributes Calculated from Conjoint
Estimates; Comparison across Three Samples
Willingness-to-Pay ($/choice) for . . . KN
IRI
Nielsen
Non-Clone vs. Cloned $4.16 $3.81 $3.25
Non-Clone vs. Offspring of Clone $3.59 $3.50 $2.88
Offspring of Clone vs. Clone $0.57 $0.31 $0.37
Increase in Leanness (90% vs. 80%) $0.54 $0.28 $0.33
Reduction in Saturated Fat (5% vs. 10%) $0.80 $0.42 $0.61
84
Table 35. Milk Conjoint Models: Multinomial Logit Estimates; Comparison across Three
Samples
Attribute KNa
IRIb
Nielsenc
Clone vs. None 0.664*d
(0.035)e
0.399*
(0.041)
0.653*
(0.036)
Offspring of Clone vs. None 0.530*
(0.035)
0.201*
(0.042)
0.561*
(0.036)
Non-Clone vs. None 2.240*
(0.034)
2.287*
(0.041)
2.133*
(0.035)
rBST vs. no RBST -0.602*
(0.017)
-0.463*
(0.021)
-0.489*
(0.018)
Fat Content: 1% vs. Skim -0.123*
(0.024)
0.133*
(0.028)
-0.183*
(0.025)
Fat Content: 2% vs. Skim 0.261*
(0.023)
0.411*
(0.029)
0.290*
(0.024)
Fat Content: Whole vs. Skim -0.171*
(0.023)
-0.141*
(0.028)
-0.035*
(0.024)
Price -0.378*
(0.006)
-0.404*
(0.007)
-0.426*
(0.006)
aResults based on 35,373 choices made by 2,237 individuals; Log-likelihood function value = -39,091.28.
bResults based on 27,056 choices; Log-likelihood function value = -44,087.83.
cResults based on 33,920 choices; Log-likelihood function value = -47,351.13.
dOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
eNumbers in parentheses are standard errors
Note: The log-likelihood function value for a model where parameters are pooled across all three samples is
-145,223.65 (N=96,349). A likelihood ratio test indicates that we can reject the null hypothesis that parameters are
equal across all three samples (Chi-square value =29,386.82; df=16; P-value<0.001).
85
Table 36. Willingness-to-Pay for Selected Milk Attributes Calculated from Conjoint Estimates;
Comparison across Three Samples
Willingness-to-Pay ($/choice) for . . . KN
IRI
Nielsen
Non-Clone vs. Cloned $4.17 $4.68 $3.47
Non-Clone vs. Offspring of Clone $4.53 $5.17 $3.69
Offspring of Clone vs. Clone -$0.35 -$0.49 -$0.22
No rBST vs. rBST $1.59 $1.15 $1.15
86
Table 37. Effect of Actual Milk Purchase Behavior on Milk Conjoint Model; IRI sample
Attribute MNL
Estimate
Price -0.424*a
(0.007)b
Fat Content: 1% vs. Skim 0.153*
(0.029)
Fat Content: 2% vs. Skim 0.461*
(0.029)
Fat Content: Whole vs. Skim -0.087*
(0.029)
rBST vs. no RBST -0.489*
(0.021)
Non-Clone vs. None 2.500*
(0.045)
Non-Clone vs. None x
Annual milk purchased
-0.005*
(0.001)
Non-Clone vs. None x
Organic share
-0.057
(0.078)
Clone vs. None 0.594*
(0.049)
Clone vs. None x
Annual milk purchased
-0.005*
(0.001)
Clone vs. None x
Organic share
-1.492*
(0.194)
Offspring of Clone vs. None 0.427*
(0.049)
Offspring of Clone vs. None x
Annual milk purchased
-0.005*
(0.001)
Offspring of Clone vs. None x
Organic share
-0.903*
(0.174) aOne asterisk represents parameter is statistically different than zero at the 0.05 level or lower
bNumbers in parentheses are standard errors
Notes: Results based on 26,432 choices; Log-likelihood function value = -41,508.52.
87
Table 38. Relationship between Actual Milk Purchases and Willingness-to-Pay for Non-Cloned
Milk; IRI sample
Milk Consumption Behavior
WTP for non-
cloned vs. cloned
milk ($/choice)
Total annual milk purchasesa
1 gallon/year $4.67
25 gallons/year $4.66
50 gallons/year $4.66
Percent of total annual milk purchases that are organic
b
0% organic $4.49
50% organic $6.19
100% organic $7.88 Note: results derived by interaction actual transaction data with survey-based conjoint estimates. The mean total
annual milk purchased was 25.94 gallons/year (high = 291 gallons/year; low = 0 gallons/year). The mean percentage
of total annual milk purchases resulting from organic sales was 5.4% (high=100%; low=0%) aAssuming share of organic milk purchased is 5%
bAssuming total annual milk purchased is 25 gallons/year