EDUCATIONAL ATTAINMENT AND MORALITY: AN EMPIRICAL …
Transcript of EDUCATIONAL ATTAINMENT AND MORALITY: AN EMPIRICAL …
EDUCATIONAL ATTAINMENT AND MORALITY: AN EMPIRICAL STUDY
A Thesis submitted to the Faculty of the
Graduate School of Arts and Sciences of Georgetown University
in partial fulfillment of the requirements for the degree of
Master of Public Policy in Public Policy
By
Richard Xavier Headley-Soto, B.S.B.A.
Washington, DC April 16, 2013
ii
EDUCATIONAL ATTAINMENT AND MORALITY: AN EMPIRICAL STUDY
Richard Xavier Headley-Soto, B.S.B.A.
Thesis Advisor: Peter Hinrichs, Ph.D.
ABSTRACT
I use ordinary least squares regression with year and region of birth fixed effects to
estimate the relationship between educational attainment and morality. Morality is broken into
six categories. I find statistically significant and sizable relationships between educational
attainment and individuals’ perceptions of morality, moral reasoning, moral attitudes towards
personal or private conduct, and moral attitudes towards citizens’ relationship with government. I
find mixed results concerning education’s relationship with moral attitudes towards
responsibility and loyalty. I do not find a statistically significant relationship between
educational attainment and moral attitudes towards economic inequality. I further analyze the
different levels of education and find that college education drives the relationships found for
cumulative educational attainment. Elementary education has no significant relationship with
morality in any category of morality. High school education has a statistically significant
relationship with some indicators of morality, but not others, across the categories of morality.
Postgraduate education exhibits similar outcomes to college education on moral attitudes
towards private action. It also exhibits a statistically significant relationship with moral attitudes
towards government – in some cases similar to the relationship between college education and
these attitudes, and in other cases diametrically opposed to that relationship.
iii
TABLE OF CONTENTS
I. Introduction ..................................................................................................................................................... 1
History of Education and Morality in the United States ................................................................................ 1
Current Policy Issues ...................................................................................................................................... 2
Possible Channels for an Effect ...................................................................................................................... 4
Research Overview ......................................................................................................................................... 5
II. Literature Review .......................................................................................................................................... 6
Social Returns to Education ........................................................................................................................... 7
Economic Returns to Education ................................................................................................................... 11
Psychological Theories ................................................................................................................................. 12
Summary ....................................................................................................................................................... 13
III. Data ............................................................................................................................................................. 14
IV. Methodology .............................................................................................................................................. 16
V. Results ......................................................................................................................................................... 19
Nature of Morality ........................................................................................................................................ 20
Moral Reasoning .......................................................................................................................................... 22
Moral Attitudes towards Private Actions ..................................................................................................... 24
Moral Attitudes towards Responsibility and Loyalty ................................................................................... 26
Moral Attitudes towards Government .......................................................................................................... 28
Moral Attitudes towards Economic Inequality ............................................................................................. 31
VI. Conclusion .................................................................................................................................................. 33
Table 1: Summary Statistics for Variables Included in “Morality” ................................................................. 37
Table 2: Summary Statistics for the Variables of Interest and Control Variables ........................................... 38
iv
Table 3: Estimates of Education’s Impact on Variables Revealing Morality .................................................. 39
Table 4a: Regression Results for “World is Evil” ............................................................................................ 41
Table 4b: Regression Results for “Morality Has Shades of Gray” .................................................................. 42
Table 4c: Regression Results for “Woman has a Right to Abortion” .............................................................. 43
Table 4d: Regression Results for “Personal Responsibility for Reducing Pain” ............................................. 44
Table 4f: Regression Results for “Inequality in US is Too Large” .................................................................. 46
Table 5a: Estimates of Education’s Impact at Various Levels on Perceptions of the Nature of Morality ....... 47
Table 5b: Estimates of Education’s Impact at Various Levels on Moral Reasoning ....................................... 48
Table 5c: Estimates of Education’s Impact at Various Levels on Moral Attitudes Towards Private Actions . 49
Table 5d: Estimates of Education’s Impact at Various Levels on Moral Attitudes Towards Responsibility .. 50
Table 5e: Estimates of Education’s Impact at Various Levels on Moral Attitudes Towards Government ..... 51
Table 5f: Estimates of Education’s Impact at Various Levels on Moral Attitudes Towards Inequality .......... 52
Reference List ................................................................................................................................................... 53
1
I. Introduction
Education and morality have been associated in intellectual and political spheres
since the advent of philosophic thought. Socrates linked the two in claiming that immoral
behavior resulted from ignorance of the good. Aristotle expounded upon this in his
contention that, to be virtuous, one’s emotional responses must be trained to act in
concert with one’s rational responses to a given situation. In his Nicomachean Ethics, he
asserted “the best way to teach morality is to make it a habit in children.”
History of Education and Morality in the United States In the United States, morality and education were intertwined from the colonial
period forward. Harvard University was founded in 1621 with the purpose of creating a
place for clergy to study scripture. And in 1642, the Massachusetts Bay Colony
established compulsory schooling in all towns of 50 or more people, not only for
economic advancement, but so that all children could learn Calvinist teachings.
Politicians and philosophers alike indicated a belief in a strong relationship
between education and morality. Early in his political career, Abraham Lincoln wrote, "I
desire to see a time when education, and by its means, morality, sobriety, enterprise and
industry, shall become much more general than at present." Horace Mann, an influential
Massachusetts politician and considered by many to be the founding father of secular and
public education in the United States, wrote in the mid-19th century that if children could
go to “good schools, the dark host of private vices and public crimes…might [nearly] be
banished from the world.” And John Dewey, a prominent early 20th century philosopher
on education, thought that education’s effect on morality was omnipresent: “The
2
development of character [occurs] through all the agencies, instrumentalities, and
materials of school life.”
Later in the 20th century, the most prominent thinkers in science and religion
would also link education and morality. Said Albert Einstein: “A man's ethical behavior
should be based effectually on sympathy, education, and social ties and needs.” And said
Rev. Martin Luther King, Jr.: “Intelligence plus character – that is the goal of real
education.” However, education’s connection with moral development has not come
without controversy in the political sphere. Particularly, a large portion of the United
States population has felt, and continues to feel, that only religious teachings should
inform morality. Many demonized Mann for his efforts to establish secular education in
what was then a country dominated by religion. In fact, it was not until a century after
Mann’s stint as Massachusetts’ Secretary of Education that the Supreme Court found
religious instruction in public school to violate the Establishment clause of the
Constitution.
In sum, it has long been thought that one of the objectives of education ought to
be the moral development of pupils. Furthermore, it has been assumed that development
does in fact occur over the course of a child’s educational attainment, and is particularly
shaped by the content of the child’s education.
Current Policy Issues
Increasing educational attainment, at all levels, is a policy goal commonly
espoused by leaders at the federal, state, and local levels. President Obama pledged to
devote considerable resources to early childhood education during his first campaign, and
is currently working with Secretary of Education Arne Duncan to evaluate the effects of
3
spending $10 billion on expanding access to pre-kindergarten to middle-class families.
Universal, or at least greatly expanded, pre-K programs have been implemented in
Oklahoma and Florida, and Andrew Cuomo called for New York to implement and
expand its already authorized universal program in his recent State of the State address to
the New York legislature.
Part of the rationale for additional childhood education has been to improve high
school graduation rates (including reducing the dropout rate) and concomitantly, college
attendance and graduation rates. These goals are largely seen as desirable by policy
makers of all stripes, on the understanding that improved educational attainment
generally results in improved outcomes for individuals in the workforce, and for the
economy as a whole. President Obama mentioned “college” eight times and “education”
fifteen times in his most recent State of the Union, and hardly a single governor has failed
to mention improving access to, and attainment of, higher education when issuing an
annual address to the state legislature. However, rarely do policy makers acknowledge
non-economic effects of increasing educational attainment. If increasing educational
attainment has an effect on individuals’ moral reasoning, attitudes, and perceptions of
morality, then it would be beneficial for policy makers to know about and account for the
effects when crafting policies to improve educational attainment.
Additionally, two political issues fall squarely within the intersection of education
and morality. First, the drive to keep religion out of public schools has left that space
seemingly morally barren for some, and is arguably one of the reasons for an uptick in
enrollments in private educational institutions, as well as home schooling. On the other
hand, the public’s aversion to government-sponsored indoctrination has kept most secular
4
moral teachings out of public school curricula. This has generated outcry from those who
believe that education in civics is a critical part of a well-rounded course of study.
Second, there is persistent criticism that there is “liberal bias” in higher education,
based largely on surveys that find that a substantial majority of college professors self-
identify as liberal. This has induced fear in some commentators that colleges inculcate
liberal values in their students, or in other words, that educational attainment at the
highest levels is responsible for a shift in moral attitudes that is undesirable to some.
Furthermore, polling results have shown Democratic Party candidates have held a
consistent advantage amongst voters with postgraduate education (as well as with voters
with no high school degree), but that Republican Party candidates have held the
advantage amongst voters with at least some college education (but no graduate degree).
It is thus possible that changes in morality driven by educational attainment have political
impacts that shape the partisan divide in government.
Possible Channels for an Effect
Although political furor is often directed at the content of what is taught in school,
there are many more channels – aside from direct transfer of moral information – through
which education might shape moral attitudes. Exposure to new (non-moral) information,
concepts, or modes of thinking might stimulate changes in moral attitudes. Improvements
in students’ logical reasoning skills might similarly spark changes in attitudes. Finally,
but perhaps most prominently, the multitude of social interactions that are generated in
educational environments could have a great effect on the development of moral
reasoning and attitudes.
However, despite the many channels through which education might affect
5
morality, and the common-sense consensus that it does, there are valid reasons to suspect
that educational attainment has no effect on moral attitudes. First, perhaps it truly is only
the content of one’s education, and not the extent of it, that affects morality. Second, it is
possible that attitudes harden at an early age, and that education beyond that age has no
effect on attitudes. Lastly, it is possible that social interactions outside of school, or other
life events, dominate the shaping of one’s attitudes, to the exclusion (or minimization) of
education.
Research Overview
This paper will seek to address whether educational attainment in fact affects
moral attitudes, without regard for the particular curricula or course of study during that
attainment. I will rely upon pooled cross-sectional survey data from the General Social
Survey (GSS) collected between 1972 and 2010. I use the responses to a variety of
questions relating to morality to address whether educational attainment moves moral
attitudes in some subject areas but not others. I will also examine whether there is an
impact on moral reasoning or overall perceptions of morality itself. There has been no
concerted effort in prior research to address this specific question.
My estimation method is ordinary least squares (OLS), controlling for a host of
biographic and demographic variables, variables related to childhood environment and
parental status, and year and region of birth fixed effects. I estimate linear probability
models (LPMs) and find statistically significant relationships between educational
attainment and moral reasoning, as well as between education and perceptions of
morality. I also find statistically significant relationships between educational attainment
and some moral attitudes, particularly attitudes towards personal or private behaviors and
6
attitudes towards individuals’ interactions with government. I find no statistically
significant relationship between educational attainment and moral attitudes towards
economic inequality. Results are inconclusive as to whether educational attainment has a
meaningful relationship with moral attitudes towards responsibility and loyalty.
The remainder of this paper is organized as follows: Section II will discuss related
prior research. Section III will describe the data, and section IV will describe my
empirical methodology. Section V will describe the results, interpret them, and discuss
their implications. Section V will conclude and give suggestions for further research.
II. Literature Review Little empirical investigation has been done on the possible effects of education
on moral or ethical attitudes. Other than a limited study finding a correlation between
education (as well as age and gender) and the responses to a 90-person survey on ethical
behavior in an Iranian company (Ahmadi and Ashrafjahani 2011),1 there appears to be no
empirical analysis directly on point. However, much empirical analysis has been done
with respect to the “social” returns to education. Empirical analysis has also conclusively
established connections between education and crime, as well as education and earnings.
Furthermore, the field of psychology has generated much useful theory when considering
the question at hand.
1 The research question posed focused on business ethics, and specifically the impact of age, gender, education, and the presence of a company code of ethics on managers’ evaluations of the ethicality of employees’ behaviors and the factors leading to unethical behavior. The study found statistically significant correlations between gender and evaluations of the ethicality of employee behavior (women largely believed that employees were more likely to act unethically), as well as between education level and evaluation of the factors leading to unethical behavior (more educated managers believed that the factors posited in the survey as impacting ethical behavior were relevant to actual ethical behavior).
7
Social Returns to Education
An area that has been studied with some vigor in empirical circles is the social, or
non-market, returns to education. Many possible social effects of education have been
analyzed in the empirical literature, with possible but unconfirmed (or muddled)
connections to the relationship between education and moral attitudes. Some studies have
found notable relationships between educational attainment and beliefs or behaviors that
may have moral underpinnings, but have not explicitly made the connection to morality.
Others have examined the connection between educational attainment and explicitly
moral beliefs or behaviors, but have obtained results that do not readily identify whether
a relationship exists.
One of the analyses most relevant to the topic at hand was of GSS data from
1972-2000. Using changes in child labor laws as an instrument for educational
attainment, the study found sizable and statistically significant positive effects on whether
individuals believe anti-religionists, communists, and homosexuals should be allowed
free speech (Dee 2004). Particularly, Dee found that an additional year of attainment
increased the likelihood that an individual would allow free speech to an anti-religionist,
a homosexual, or a communist, by 12.0, 12.3, and 8.0 percentage points, respectively. All
three results were statistically significant at the one-percent level. Belief that a person,
whose views clash with one’s own, should be allowed free speech is certainly reflective
of a certain moral attitude. However, the connection this study might have established
was muddled when it failed to find a statistically significant effect of educational
attainment on whether individuals believe that militarists or racists should be allowed free
8
speech.2 Additionally, the existence of an effect of educational attainment on one attitude
that could be counted as moral cannot be said to establish a conclusive link between
educational attainment and the many attitudes and forms of reasoning that encompass an
individual’s morality.
A number of other analyses of the social return to education touch upon a
potential relationship between educational attainment and morality. Oreopolous and
Salvanes (2011) examined GSS data from 1972-2000) and found significant positive
correlations between educational attainment and an individual’s trust as well as patience,
implicit in negative correlations with whether the individual was ever arrested, ever
smoked, had a child born as a teenager, or agreed that he or she “lived for today.”
Specifically, they found that roughly 30% of respondents with less than a high school
education believed “people can be trusted,” but that the relevant fraction increased to
40% among high school graduates, 45% among individuals with some college education,
and more than 55% for respondents with a college degree or higher. Similarly, more than
50% of respondents without a high school degree said that they “live for today,” had a
child as a teenager, and had smoked at some point in their lives. But these proportions
declined dramatically as education levels increased (to roughly 30%, 10%, and 40% in
the respective categories, for respondents with a college degree or greater). However,
while patience and trust could represent moral attitudes, they might also represent
economic preferences. Oreopolous and Salvanes point out that education might improve
trust because it teaches people how to interact successfully with others, or because it
2 Two further notes about this study are worth mentioning. First, Dee’s OLS estimates suggested statistically significant correlations between educational attainment and allowing all five categories of potentially objectionable speakers the right to speak, although the size of the effects were muted. Second, Dee did not control for the religion the individual was raised in, but rather the individual’s current religion, introducing potential problems with endogeneity.
9
teaches them to fully account for transaction costs (which are lower when people are
trusting). With regard to patience, they suggest that education may teach people to
properly weigh future benefits against current costs and benefits, thus “reducing myopia.”
In both cases, a plausible economic argument weakens the argument that education
makes individuals more patient or trusting exclusively as a matter of moral attitude.
A meta-analysis published by the Federal Reserve Bank of Boston has suggested
possible public returns to education in the categories of charitable giving and self-
reliance, implied by the activity of forgoing government benefits one is entitled to
(Haveman and Wolfe 2002). Particularly, the analysis referenced Hodgkinson and
Wietzman (1988), who found that college graduates donated 50% more of their incomes
and volunteered nearly twice as much as high school graduates. The meta-analysis also
referenced a previous study by the authors, which suggested that more education led to a
decreased probability of receiving disability or welfare benefits to which one would
otherwise be entitled (Haveman and Wolfe 1993). Associations between educational
attainment and these behaviors do not immediately appear to flow through a channel of
economic self-interest, again suggesting a link to moral attitudes. But evidence of
causality is limited, and the studies referenced are dated.
McMahon compiled a number of OLS analyses on aggregate annual data for 78
countries, and found that the percentage of the country’s population enrolled at the
secondary level (lagged) was significantly and positively correlated with democratization
(fair elections, competing parties, civilian control of the military, and decentralized
power) and equality, as suggested by a negative relationship with the country’s Gini
coefficient (1999). But even if these correlations prove reliable and linked to moral
10
attitudes rather than economic self-interest, it would be problematic to extrapolate from
aggregate preferences to individual ones.
Other studies have reviewed links between educational attainment and “social
returns” that are less likely to be indicative of moral reasoning or attitudes. Particularly, it
has been of interest whether increased education increases the likelihood that an
individual will be politically active. Dee’s analysis of High School & Beyond data, using
distance to the closest two-year college and concentration of two-year colleges as
instruments for educational attainment, found statistically significant positive effects on
voter registration and turnout (2004). Dee suggests that one of the mechanisms through
which education might increase civic participation is the shaping of preferences.
“Preferences” could meal moral attitudes regarding what people should do, but it could
also simply refer to what the individual likes to do. Furthermore, Dee postulates that the
effect could be the result of education reducing the cost (to the individual) of civic
participation. It could be added that education might affect individuals’ perception of the
private returns to civic activity.
Additionally, another notable analysis of education’s effect on citizenship
produced mixed results when employing similarly rigorous empirical methods. Milligan,
Moretti and Oreopolous analyzed National Election Studies data (for the US) and
Eurobarometer survey data (for the UK), using compulsory schooling laws as an
instrument for education (2004). The analysis of the US data found a statistically
significant positive effect on whether individuals follow political campaigns and public
affairs, but it found no statistically significant effects on voter registration. Similarly, the
analysis of the UK data found statistically significant positive effects on whether
11
individuals discussed political matters and considered themselves politically active, but
no statistically significant effect on voter registration or on whether individuals follow the
news.
Finally, another empirical study of note considered whether the share of college
graduates in a city increased wages over and above the expected private returns to
education (Moretti 2004). It analyzed National Longitudinal Survey of Youth (NLSY)
and US Census data, using differences in lagged age structure and presence of a land-
grant college as instruments for average educational attainment in a particular city, and
found a significant positive effect. However, while such a finding could possibly be
related to individual preferences for social equality (a moral attitude), it was not
mentioned by Moretti as a potential channel; he instead focused on the possibilities that
educated workers raised the productivity of the uneducated, or that there were “human
capital spillovers.”
Economic Returns to Education It is widely accepted that educational attainment has a causal effect on earnings.
While the magnitudes of estimates vary, subsequent analysis of the various statistical
techniques used to model the relationship has confirmed that sizable and statistically
significant causal effects are present (Card 1999). This has implications for the question
at hand because moral attitudes may be affected by income level, and as such, education
may affect morality obliquely through its effects on earnings.
The effect of education on crime has also been studied thoroughly. Analysis of
US Census data, FBI Uniform Crime Reports, and the NLSY, using changes in state
compulsory education laws as an instrument for educational attainment, found that
12
increased educational attainment had statistically significant negative effects on the
probability of committing a crime (Lochner and Moretti 2004). International studies have
also found significant negative effects (Machin, Marie and Vujic 2011; Hjalmarsson,
Holmlund and Lindquist 2011). However, education’s link to reduced criminal activity is
rarely described as resulting from a fundamental effect on the moral attitudes of students,
but instead is typically viewed as an economic choice. Economic models typically
suggest that education and crime are substitute activities, and that the choice between
them is driven by relative wage rates, opportunity cost, and time preferences.
Nonetheless, as with education’s effect on earnings, moral attitudes may be affected by
one’s participation in criminal activity, and thus affected by educational attainment in an
indirect manner.
Psychological Theories The two most influential thinkers in the field of psychology, with respect to
education’s impact on moral attitudes, are Jean Piaget and Lawrence Kohlberg. A
summary of their work strongly suggests that education has a substantial impact on moral
development (Hersh, Paolitto and Reimer 1990).
Piaget’s theory of cognitive development suggests that intellectual and moral
development occurs in stages as individuals age and perform actions on objects. Actions
lead to consequences that the individual then internalizes and abstracts upon in both
reflective and empirical ways. The abstraction processes coalesce into new knowledge
and insight. With specific regard to moral development, Piaget subscribed to Kantian
ontology, and as such thought morality developed from observations of the world and
from peer interactions; he thought that the drive for autonomy would negate attempts to
13
instill morality directly. Thus, Piaget strongly believed that education could influence
morality, but only through influencing interactions with objects, ideas, and peers.
Morality could not be inculcated directly through dogma.
Kohlberg extended upon Piaget’s theory of cognitive development, creating the
most prominent psychological theory of moral development. The theory holds that moral
reasoning develops in six stages, with two stages at each of three levels: pre-
conventional, conventional, and post-conventional. He suggested that moral development
continues late into an individual’s life span, and although he noted moral regression in his
studies, he thought that regression was not actually occurring. Rather, he attributed
regression to transition between stages, where moral reasoning and behavior could be
misconstrued. Kohlberg believed that individuals cannot skip stages of development, and
thought education a critical component of advancement through stages. Specifically, he
believed education advanced an individual’s conceptualization of the psychology of other
persons, his or her appreciation and reflection on social norms, and his or her ability to
reason logically from multiple perspectives.
Summary There is a plethora of evidence suggesting that educational attainment has an
impact on moral attitudes. However, there is no empirical evidence directly on point. The
empirical evidence that is available in related fields is subject to alternative
interpretations, and psychology can only offer theories, convincing as they might be. This
study will thus seek to contribute to the literature by reliably estimating whether a
relationship exists, and if one is found, by beginning the discussion of what that
relationship might be.
14
III. Data The data used in this analysis come from the General Social Survey (GSS)
cumulative file (1972-2010), a pooled cross-section of data for every year that the survey
has been performed, with a total of 55,087 observations. The survey is conducted and
data compiled by the National Opinion Research Center (NORC) at the University of
Chicago, with the primary purpose of monitoring social attitudes and changes in those
attitudes. The GSS was conducted annually between 1972 and 1993 – with the exceptions
of 1979, 1981, and 1992 – and biannually from 1994 through 2010. In the years it was
conducted annually, the survey averaged 1,547 observations; since it has been conducted
biannually, each survey has averaged 2,855 observations. Each survey was conducted
against an independently drawn sample of US residents, aged 18 years and older, and
living in non-institutional arrangements. Block quota sampling was used to select all
respondents for the 1972-1974 surveys and for half of the respondents to the 1975 and
1976 surveys. Block quota sampling involves segmenting the population into mutually
exclusive subgroups, and then creating the sample by sampling within each block to meet
a predetermined quota of observations from that block. The GSS used blocks stratified by
income and race, and applied quotas based on age, sex, and employment status. Full
probability sampling, featuring entirely random selection of observations, was used to
select the other half of respondents to the 1975 and 1976 surveys, and to select all
respondents for every survey from 1977 forward. From 1972 through 2004, respondents
were limited to English-speaking persons; from 2006 forward, samples have included
Spanish speakers as well.
15
The GSS contains three types of questions: permanent (or “core”) questions that
appear on every survey; rotating questions that appear frequently but not universally;3
and special questions that appear in a single survey.4 The survey organizers have taken
care to retain the exact wording of reappearing questions whenever possible in order to
facilitate time trend studies. Respondents are interviewed by trained social scientists
about their biographic and demographic background, as well as opinions on a multitude
of topics. The median time taken to conduct an interview has historically been roughly 90
minutes. While most questions include response codes for “Don’t Know,” “No Answer,”
and “Not Applicable,” these responses were, in most cases, coded as missing by the
survey organizers in the publicly available data set. However, I have also dropped three
types of observations in order to preserve the statistical integrity of the models estimated.
Education was reported as “unsure” for 122 observations; these were dropped.
Additionally, 476 observations were dropped where age was reported as “89+” years, and
1673 observations were dropped where the respondent either indicated that he or she had
more than 25 siblings or did not provide an answer to the question of how many siblings
he or she had.
There are well-documented issues associated with the use of survey data in
empirical studies, particularly the presence of measurement error in the independent
variables due to respondents’ liability to give imprecise answers (or worse, intentionally
misreport), particularly when the survey is designed poorly or the interviewer is not well
3 Rotating questions appeared for all respondents in two out of every three years from 1972 through 1988. Beginning in 1990, all rotating questions were asked every year, but only of two thirds of respondents. 4 These categories are not necessarily static. Special questions may be asked in multiple years and become rotating or even permanent questions; permanent and rotating questions might be dropped from the survey.
16
trained. The rigor with which the GSS is prepared and executed eases concerns one might
have about survey data generally and measurement error specifically.
IV. Methodology I use the GSS data to measure whether educational attainment has a relationship
with an individual’s moral reasoning and attitudes, as well as his or her conception of the
nature of morality. My models are of the general form:
𝑚𝑜𝑟𝑎𝑙𝑖𝑡𝑦!"# = 𝑒𝑑𝑢𝑐!"#𝛽! + 𝑏𝑖𝑜𝑑𝑒𝑚𝑜!"#! 𝛽! + 𝑝𝑎𝑟𝑒𝑛𝑡𝑠!"#! 𝛽! + 𝑐ℎ𝑖𝑙𝑑ℎ𝑜𝑜𝑑!"#! 𝛽! + 𝛼! + 𝛿! + 𝜖!"#
In these models, 𝑚𝑜𝑟𝑎𝑙𝑖𝑡𝑦!"# represents a response from individual i to one of
various questions that shed light on morality, in survey year t, given that the individual
was raised in region r. I use 27 distinct variables (responses to survey questions) to
measure morality, broken down roughly into six categories: the nature of morality, moral
reasoning, moral attitudes towards interpersonal and private actions, moral attitudes
towards responsibility and loyalty, moral attitudes towards government and its
relationship with citizens, and moral attitudes towards economic inequality.5 Table 1
describes all the variables that were used to measure morality, and includes information
on the sample size and mean for any given variable. Sample sizes vary because, with only
a couple exceptions, the survey questions (variables) included were rotating or special
questions, and thus were not asked universally across surveys and respondents.
5 The classification of the variables into these categories is not meant to be a definitive statement on what they represent. Rather, it is simply a construct used to simplify discussion of the many variables that shed light on morality. Many of the variables speak to more than one of these classifications, while some of them fit into a classification only loosely.
17
The variable of interest in this model, 𝑒𝑑𝑢𝑐!"# (educational attainment), is
measured as the total years of formal schooling completed by the individual, on a scale of
zero to 20. The value is derived from a series of questions the interviewer asks the
respondent about the highest grade he or she has completed. The coefficient of interest
then is 𝛽!, which measures the impact of an additional year of formal school on morality.
I have also broken down educational attainment into four separate component variables:
elementary education, high school education, college education, and postgraduate
education. Elementary education is defined as attainment within the first eight years of
formal education (on a scale of 0-8), high school as attainment within the next four years
(0-4), college as attainment within the four years after that (0-4), and postgraduate
education as any attainment after college, limited to a maximum of four years (0-4).6 I
will conduct further analysis using these component variables as variables of interest to
see whether or not relationships are consistent across levels of education.
Three vectors of control variables are included in the model in order to minimize
omitted variable bias. Included in 𝑏𝑖𝑜𝑑𝑒𝑚𝑜!"#! are the individual’s age, race, gender, and
an indicator for whether or not the person was born in the United States. Included in
𝑝𝑎𝑟𝑒𝑛𝑡𝑠!"#! are controls for mother’s education, father’s education, whether the
individual’s parents were born in the United States, and whether the individual’s
grandparents were born in the United States. The vector 𝑐ℎ𝑖𝑙𝑑ℎ𝑜𝑜𝑑!"#! represents variables
related to the individual’s childhood environment, particularly the number of siblings the
individual had, the religion he or she was raised in, whether his or her family’s income
was above or below average, and the level of urbanization of the place where the 6 The maximum value of the “postgraduate education” variable follows as a consequence of the coding of the original education variable (from with the component variables were derived), which had a maximum value of 20 years.
18
individual was raised. Table 2 lists all the variables of interest and control variables
included in the models, and provides summary statistics.
In order to strengthen the reliability of the estimates presented, most of the models
estimated include year and region fixed effects, in addition to the control variables
discussed above.7 Specifically, this means that indicator variables for each year the
survey was given will be included, controlling for any shocks or trends in moral attitudes
attributable to the particular period in history in which the survey was conducted.
Indicator variables for the region respondents grew up in will also be included,
controlling for any effects on individuals’ morality attributable to the geographic location
in which he or she was raised. Including these controls further reduces the problem of
omitted variable bias. I also cluster standard errors by the region in which the respondent
grew up.
It must be noted that, for a number of the models estimated, observations were
dropped where the response to the survey question did not indicate alignment with either
state in the binary variable constructed for the purpose of estimating an LPM. That is,
some survey questions asked respondents what their attitude was towards a statement,
and the discrete set of response choices allowed was on a scale, where for example, “1”
indicated “strong agreement” with the statement and “5” indicated “strong
disagreement.” Where the discrete set was an odd number, the middle of the scale
arguably could not be classified as either “agreement” or “disagreement” with the
statement, and therefore could not be used when the original responses were transformed
to binary variables for the purpose of constructing LPMs. This practice necessarily
reduced the statistical power of these models, and arguably could influence the results, as 7 Exceptions are noted as appropriate on the results tables.
19
the construction of binary variables might be said to unduly remove nuance from the
measurement of the attitudes captured by these variables.
There are other valid concerns that can be raised about this approach. The specter
of omitted variable bias looms over this study like most others; because so many
variables are likely to impact an individual’s morality, many variables not accounted for
in these models could be correlated with both educational attainment and moral attitudes.
For example, individuals who experience fewer traumatic events than their peers as
children might both exhibit greater educational attainment as well as different moral
attitudes. Additionally, reverse or simultaneous causality could be a problem for this
study: It is possible that moral attitudes in adolescence and early adulthood impact the
level of educational attainment an individual achieves.
V. Results
Discussion of results is organized by category of morality as identified in the
preceding section, and presented in the order listed in Table 1. Regression results tables
referred to in the discussion are organized as follows: Table 3 reports parameter estimates
for the variable of interest, educational attainment, for each of the variables within
morality, in all categories of morality. Tables 4a through 4f report full regression results
for a selected (representative) variable within each category. Tables 5a through 5f report
parameter estimates for educational attainment by level (elementary, high school, college,
and postgraduate) and organized by category of morality.
20
Nature of Morality The results of a variety of OLS regressions (as reported in Table 3) strongly
suggest a link between educational attainment and individuals’ perceptions of the nature
of morality. Individuals with more formal education were less likely to believe that the
world is fundamentally evil (by two percentage points per year of education), more likely
to believe that human nature is fundamentally good (by 1.2 percentage points per year),
less likely to believe that morality is solely a personal matter (by 0.8 percentage points
per year), and less likely to believe that science breaks down people’s morals (by three
percentage points per year). Parameter estimates were statistically significant at the 1%
level or greater for each regression run within this category, exhibiting remarkable
robustness to inclusion of controls.
That additional education changes the way one perceives the nature of morality,
in the sizable way suggested by the results, makes logical sense. As one is exposed to
new information and knowledge, one’s understanding of the world is reshaped to
accommodate it. Particularly, one would expect (or at least hope) that additional
understanding would lead to positive impressions about the subject matter, possibly
explaining why those with more education see the world and human nature in a more
positive light. In a similar vein, additional understanding acquired through education
reveals the interconnectedness of individuals to the world around them, and thus might
lead them to believe that morality is not an island unto each individual, but the interplay
of judgments made from different individuals’ perspectives. It is probably least surprising
that educational attainment makes people less likely to believe science breaks down
21
morals, since understanding of the scientific method reveals a process that does not
judge, but which only seeks truth.
Examination of the parameter estimates for controls applied to the regressions (as
reported in Table 4a) reveals age, gender, and the urbanicity of one’s childhood
environment to all be statistically significant predictors of whether individuals believe the
world is evil. More interesting, however, is that it appears that the religion one is raised in
is not statistically significant in determining one’s perception of the world.8 If one is
inclined to believe that educational attainment affects how one perceives the world by
adding to individuals’ understanding of it, then these results would as a corollary suggest
that raising a child in religion does not add to that child’s understanding of the world.
While it is tempting to believe that all educational attainment is equally associated
with perceptions of morality, Table 5a suggests that there is a particularly strong
association with college education. Like the results for educational attainment overall,
college education is statistically significant in every regression it is included in, and the
parameter estimates are even more sizable. This comports with the traditional conception
of college as a place where horizons are broadened, both in and out of the classroom. By
contrast, postgraduate education is not a statistically significant predictor of any of the
component variables examined, possibly because most postgraduate education involves a
narrowing of focus. Likewise, nowhere is elementary education a statistically significant
predictor, likely due to its focus on basic skills, and its provincial environment. High
school education, by contrast, may expand understanding in some regards but not others,
leading to differing levels of association and significance amongst the variables analyzed.
8 The parameter estimate of the effect of being raised as a Protestant was weakly significant (p < 10%). There was no statistically significant effect of an individual being raised as a Catholic, as a Jew, or without religion.
22
Moral Reasoning Educational attainment also exhibited statistically significant relationships with
variables that shed light on moral reasoning. People with more formal education bent
toward modes of thinking that apply cost-benefit analysis and weighing of multiple
factors to reach the optimal moral outcome: The parameter estimates reported in Table 3
indicate that more education leaves people more likely to view morality in shades of gray
rather than strictly in terms of black or white (by one percentage point per year of formal
education). Those with more formal education were also less likely to believe that sinners
should necessarily be punished (by two percentage points per year), or that the immoral
acts of one individual corrupt society as a whole (by 0.7 percentage points per year).
Results were robust to the inclusion of controls.
If one accepts that education increases individuals’ abstract reasoning abilities and
the number of perspectives that one analyzes situations from, it would follow that the
more educated one is, the more likely he or she would apply more complex modes of
thinking. In other words, it seems plausible that as one becomes more educated, one sees
more facets of a given situation, and therefore must balance more factors in passing
judgment. Hence, he or she will reflect on the evaluation process as one of distinguishing
shades of gray, rather than applying black and white labels. It then follows that the more
educated one is, the less likely he or she would be to indiscriminately punish those that
have done wrong, but instead would consider the totality of the situation to determine
whether incapacitation or rehabilitation is a more appropriate response to the behavior
than retribution or deterrence.
23
Educational attainment did not exhibit a statistically significant relationship with
whether an individual believes that dissenters should be treated leniently. This could be
the result of a limited sample size (and lack of control for year fixed effects, since all
responses were from a single year), but could also reflect that the response to this
question not only reflects reasoning, but an attitude towards nonconformity that does not
fall squarely within any of the categories of morality analyzed here.
Controls applied to the regressions (as reported in Table 4b) were almost
universally not statistically significant predictors of whether individuals believe morality
has shades of gray.9 This is unsurprising, as none of the controls applied appear to be
obvious candidates to affect one’s abstract reasoning abilities.
Analysis of the relationship between educational attainment at different levels and
moral reasoning (as reported in Table 5b) reveals college education as the driver of the
cumulative association between education and moral reasoning. This seems plausible,
given that advanced abstract reasoning and the careful balancing of factors in analysis are
most typically associated with higher education. However, given this mode of analysis,
one would also expect statistically significant parameter estimates for the association
between postgraduate education and the variables in this category, but this level of
education only had a significant relationship with whether individuals believed that a
sinner should be punished, and not with whether the individual believed morality has
shades of gray. Furthermore, no level of education had a statistically significant
relationship with whether individuals believed one immoral person corrupts society,
9 Whether an individual lived on a farm while growing up was the only exception, with a parameter estimate, statistically significant at the 5% level, indicating those individuals were four percentage points less likely to believe morality has shades of gray.
24
making it difficult to interpret the baseline regression that suggests overall educational
attainment has a significant relationship with this variable.
Moral Attitudes towards Private Actions In general, the results listed in Table 3 indicate that individuals with more formal
education demonstrated a greater propensity to accept private behavior without moral
opprobrium. Holding other variables in the model constant, additional formal education
made individuals less likely to believe that premarital or homosexual sex was wrong (by
0.7 and 2.3 percentage points, respectively, per year of education) and more likely to
agree that gay couples should have the right to marry (by 1.4 percentage points per year).
Those with more formal education were also more likely to believe that a woman should
have the right to have an abortion for any reason that she chooses (by two percentage
points per year), and that it is sometimes okay for a grown man to hit another grown man
(by 1.8 percentage points per year). These parameter estimates were all statistically
significant at the 1% level or greater, and were largely robust to the inclusion of controls.
The results here are more than plausible, given the preceding analysis of
educational attainment’s relationship with moral reasoning, as well as with perceptions of
the nature of morality. Educational attainment exposes one to new information and
knowledge, and improves one’s capacity for abstract reasoning, to the effect that
individuals with greater levels of education are prone to a more positive outlook on
human nature, and have a greater tendency to employ complex modes of analysis. Hence,
a person with more education is likely to be more sympathetic to individuals whose
behaviors differ from their own. The more educated one is, the more one might reason
that private behavior by good-natured human beings should be allowed to proceed
25
without moral criticism. Or alternatively, more education could lead one to reason that,
although personally not fond of a particular behavior, maximizing utility requires
maximum accommodation of personal freedom in private behaviors.10
Unlike the previous categories of morality, many of the control variables included
in the regressions were statistically significant predictors of the component morality
variables within this category. Table 4c reports full OLS regression results pertaining to
whether an individual believes that a woman has a right to an abortion for any reason.
Parameter estimates for mother’s education, father’s education, the number of siblings
one has, relative family income as a child, the religion one was raised in, and the
urbanicity of one’s childhood environment were all statistically significant at the 5%
level or greater. This may be because moral attitudes towards abortion, unlike reasoning
or perceptions of morality, may be more prone to impact from preconceived viewpoints
associated with one’s childhood environment. Interestingly, gender was not a statistically
significant predictor of the dependent variable here, indicating that women – holding
other factors in the model constant – are no more likely to believe in their own right to an
abortion than men are.
As was found in the analysis of the relationship between different levels of
education and previous categories of morality, education at the college level is
particularly and consistently related to moral attitudes towards private behaviors (as
reported in Table 5c). Additionally, and unlike the analysis of the previous categories of
morality, postgraduate education is observed to have a strong and statistically significant
10 The only component variable in this category of morality not exhibiting a statistically significant relationship with educational attainment was whether individuals believe that a terminal patient has the right to die. The lack of a relationship could be explained by the fact that utilitarian calculus is more difficult to apply to the freedom to end one’s life; it might suggest that any time a person’s life is on balance more painful than enjoyable, he or she should end it.
26
relationship as well. In fact, both college and postgraduate education parameter estimates
are statistically significant at the 1% level or greater for every component variable within
the category,11 and exhibit relationships larger in magnitude than the baseline estimates
for cumulative educational attainment (as reported in Table 3). A possible explanation for
this phenomenon is that while postgraduate education does not necessarily lead one to
acquire additional analytical skills, it calls upon individuals to apply those analytical
skills in more exacting ways. An individual with a postgraduate education may be no
more likely to apply utilitarian reasoning, but when called upon to do so, does so more
thoroughly and precisely.
Moral Attitudes towards Responsibility and Loyalty The results in this category of morality are decidedly mixed. At first glance it
might appear that, from the results in Table 3, individuals with more formal education
tend to exhibit higher levels of social responsibility.12 Education is a statistically
significant predictor of whether an individual believes a friend has a right to false
testimony (an additional year of education is associated with a 1.2 percentage point
decrease in the likelihood one believes a friend has such a right), and of whether an
individual would tell the police the truth against a friend’s wishes (an additional year is
associated with a 1.3 percentage point increase in the likelihood an individual would tell
the truth). Education was also a statistically significant predictor of whether one agrees
11 There was one exception: The impact of college education was not significant for whether individuals believed that premarital sex was wrong, but was significant for whether individuals believed that a terminal patient has the right to die. 12 The variables in this category are related to “social” responsibility in that the survey questions (to which the variable responses refer) arguably ask the respondent to weigh responsibility to society (in reducing the pain of others, telling police the truth, or acting morally in one’s economic dealings) against responsibility to self, family, and friends (being ambivalent to others’ suffering, helping a friend out of a jam, or taking economic care of oneself and one’s family to the exclusion of moral principles).
27
that there is no morality in the way one makes money (an additional year is associated
with a 2.9 percentage point decrease in the likelihood one agrees with such a statement).
However, a number of caveats must be addressed regarding the results listed in
Table 3. First, year fixed effects were not included for three of the four component
variables within this category, because the relevant survey questions were only asked in
one year. Second, controls for the respondent’s family income at age 16 were not applied,
because responses to the relevant survey question were missing for all individuals who
had valid data on the dependent variables. Third, the results of regressions examining
education at different levels and these variables (as reported in Table 5d) do not align
well with the results of regressions examining cumulative educational attainment and
these variables (Table 3). Particularly, not a single level of education exhibited a
statistically significant relationship with whether individuals believe that a friend has a
right to false testimony or with whether individuals would tell the police the truth against
a friend’s wishes.13 Lastly, although educational attainment exhibited a statistically
significant (and sizable) relationship with whether an individual believes that there is
morality in the way one makes money, both cumulatively and specifically at the high
school and college levels, regressions on this variable did not include controls for place
of birth (for the respondent, his or her parents, or grandparents), because the question was
only asked in the first three years of the GSS (1972-1974), when place of birth questions
had not yet been added to the survey. This is worrisome for two reasons: First, all the
observations are from a limited time period, and thus may not accurately reflect current-
13 Conversely, while there was no statistically significant impact of cumulative educational attainment on whether a individual feels personal responsibility for reducing pain, postgraduate education exhibited a sizable (six percentage points per year) and statistically significant (at the 1% level) impact on this component variable.
28
day attitudes towards this issue. Second, and more poignantly, Table 4d reveals that
controls for whether parents and grandparents were born in the US may be of appreciable
impact on variables in this category.
Furthermore, there is no overriding logical reason to believe that there is, or is
not, an impact of educational attainment on moral attitudes towards responsibility and
loyalty. The relationships observed in previously examined categories were logically
justified by reasoning that education provides individuals with exposure to new
information and knowledge, and with additional analytical skills, leading to a more
positive outlook on the world and increasing the likelihood of applying utilitarian thought
processes. But while utilitarian reasoning typically bends towards private ordering and
personal freedom with regards to behaviors that primarily have an impact only on the
individuals taking part in them, it has less of a conclusive impact on behaviors in which
an individual must weigh their social responsibility (e.g., telling the police the truth or
making money in a scrupulous way) against personal responsibility (e.g., loyalty to a
friend, or providing for oneself and one’s family, whatever the cost).
Moral Attitudes towards Government
The results of OLS regressions, reported in Table 3, on component variables
within this category of morality, suggest a strong link between educational attainment
and moral attitudes towards government. Particularly, individuals with more formal
education were less likely to approve of the death penalty (by one percentage point for
every additional year of education), to agree that cheating the government by lying on
benefits applications was not that reprehensible (by 0.7 percentage points per year), or to
support their country if they believed their country’s actions were wrong (by 4.1
29
percentage points per year). Parameter estimates were generally robust to the inclusion of
controls. 14
That education would alter individuals’ attitudes towards government is quite
plausible. Additional education not only provides new information, knowledge, and
analytical reasoning skills, but it often specifically informs students as to the nature of
their government, how it functions, and what the relationship between government and
governed is supposed to entail. It follows that individuals with more years of education
will develop different viewpoints on what powers government can legitimately wield
(death penalty), how their actions as citizens impact government (cheating on taxes or
government benefits applications), and whether they should support or question their
government’s actions.
Table 4e provides full OLS regression results examining the dependent variable
representing whether individuals would support their country, even if they believed their
country’s actions were wrong. Interestingly, the only independent variables (other than
education) exhibiting statistically significant associations with the dependent variable
were age, and a dummy for whether the individual was not raised with religion. Older
people were more likely to support their country even when they considered its actions to
be wrong, suggesting that as individuals age, their tendency to voice dissent wanes. This
comports with anecdotal evidence that protests and other anti-government activity are
more likely to be spawned and attended by youth. Variables indicating whether the
individual, his parents, or her grandparents were born outside of the US did not exhibit a
significant association, which is notable because it suggests that immigrants to the US
14 Statistically significant parameter estimates for the death penalty variable did not appear until parental status controls were added to the regression.
30
either do not have different attitudes than Americans towards government support
(whether bending towards allegiance or dissent), or they do not retain attitudes brought
from their place of origin.
As was found in the other categories of morality, analysis of educational
attainment at different levels (reported in Table 5e) accords the bulk of the impact of
educational attainment on attitudes towards government to college education in
particular. Parameter estimates for the college education variable were statistically
significant in every regression, and larger in magnitude than the estimates for cumulative
educational attainment. This is believable given that college is the level of education
where students are most likely to learn about the nature of their government, how it
functions, and what the relationship between government and governed is supposed to
entail. Postgraduate education also had a marked impact on whether an individual
approves of the death penalty and whether an individual would support their country, but
not on whether an individual considered cheating (on taxes or government benefits
applications) reprehensible. This suggests, reasonably, that the most advanced levels of
education impact what people believe is the appropriate province of government and
when dissent towards government action is appropriately registered, but not whether they
believe cheating (with respect to government) is bad. Such a divide comports with an
understanding of postgraduate education as focused on advanced theoretical analysis,
rather than practical issues of ethics in one’s relations with government.
A final note regarding Table 5e pertains to the reported (and statistically
significant) relationships between high school education and whether an individual
approves of the death penalty, as well as whether an individual believes that cheating on
31
one’s taxes is reprehensible. These relationships are sizable, and flow in the opposite
direction of the relationships between college education and those dependent variables.
With respect to the death penalty, this may reflect that high school education provides
more information about the extent of government power (with a tacit understanding that
governmental power is legitimate), and less about the appropriate limits on it. With
respect to the issue of cheating on one’s taxes, a ready explanation is not available;
further investigation is warranted.
Moral Attitudes towards Economic Inequality
After balancing the results of all OLS regressions examining this category of
morality, it does not appear as though educational attainment has any meaningful
relationship with individuals’ attitudes towards economic inequality. Table 3 reports that
educational attainment has no statistically significant association with three of the five
component variables within this category. Specifically, changes in years of formal
education appear to have no effect on whether individuals believe income inequality in
the US is too large, or on their attitudes towards the fact that wealthier people receive
better health care and education. In fact, as reported in Table 4f, the only statistically
significant predictors of attitudes towards inequality in the US were gender (males were
12 percentage points less likely to believe that inequality was too large) and whether the
respondent was born in the US (foreign born respondents were 16 percentage points more
likely to believe that income inequality was too large). Not a single level of education (as
reported in Table 5f) had a statistically significant relationship with these three dependent
variables.
32
However, educational attainment did have a statistically significant (at the 1%
level) and sizable relationship with the other two component variables within the
category: Each year of education was associated with a 1.7 percentage point reduction in
likelihood that a respondent believed that the government should reduce inequality, and
with a 1.4 percentage point reduction in likelihood that a respondent believed that the
government should help the poor. Furthermore, the parameter estimates were robust to
the inclusion of controls. Variables isolating different levels of educational attainment
also exhibited statistically significant associations with these dependent variables, at the
high school, college, and postgraduate levels. But the different levels were not consistent
in the direction of their associations: While parameter estimates for high school and
college education had the same negative sign as – and were an order of magnitude larger
than – the parameter estimates for cumulative educational attainment, the parameter
estimates for postgraduate education had a positive sign. This means that while high
school and college education are associated with a reduced likelihood that an individual
will believe that the government should reduce inequality or help the poor, postgraduate
education is associated with an increased likelihood of believing that the government
should intervene.
In light of the mixed results, it would be easy to say that the analysis as to this
category of morality was inconclusive. But reflecting on the discussion of education’s
relationship with other categories of morality (particularly moral attitudes towards private
actions and moral attitudes towards government), it is more reasonable to conclude that
the variables with which educational attainment had a relationship were more indicative
of attitudes towards government and its appropriate role, rather than towards a state of
33
inequality. This comports with the consistent relationships between educational
attainment and variables within that category (moral attitudes towards government).
Furthermore, the phenomenon of different levels of education exhibiting associations
with the dependent variable in opposite directions, as reported here, was also reported for
the variable examining attitudes toward cheating on taxes, included in that category.
Finally, education’s relationship with attitudes towards government proceeds rationally
from the observation that education informs individuals about the appropriate role and
powers of government. The same cannot be said for education and the propriety of a
given level of economic inequality. In sum, recategorizing the dependent variables
associated with government action towards inequality as reflective of attitudes towards
government, rather than attitudes towards inequality, reinforces the conclusion reached
earlier – that educational attainment has a meaningful relationship with attitudes towards
government. It also allows one to state with less equivocality that educational attainment
does not have a meaningful relationship with attitudes towards inequality, as suggested
by the lack of a statistically significant relationship between educational attainment and
the remaining three dependent variables in this category.
VI. Conclusion Education has long been presumed to be associated with moral development. A
variety of issues in the development, implementation, and consequences of education
policy would benefit from knowledge of whether such an association in fact exists, and
what it implies. There are a multitude of channels through which education might affect
morality: direct transfer of moral information; exposure to new (non-moral) information,
34
concepts, or modes of thinking; improvements in logical reasoning abilities; by
generation of varied social interactions. Conversely, there are reasons to believe that
simple educational attainment might not affect moral attitudes: It is possible that only the
content of one’s education (and not the extent of it) determines moral attitudes; attitudes
may harden at a young age and be unaffected by continued education; non-school social
interactions or other life events may dominate the shaping of one’s morality.
To address this question, I used General Social Survey data covering the period
from 1972 through 2010 to estimate the relationship between educational attainment and
morality. I find that educational attainment has a statistically significant and sizable
relationship with perceptions of the nature of morality, with moral reasoning, with moral
attitudes towards private action, and with moral attitudes towards government.
Educational attainment does not appear to have any meaningful relationship with moral
attitudes towards inequality, and the results are too mixed to interpret whether there is a
relationship with moral attitudes towards responsibility. Additionally, I find that college
education is the driver of most statistically significant relationships between education
and the moral categories where relationships were noted. Elementary education, by
contrast, never appears to have a relationship with morality. Postgraduate education was a
statistically significant predictor of moral attitudes towards private actions; it also was
associated with moral attitudes towards government, but in some cases in a different
direction than college education was. Statistically significant relationships between high
school education and morality were too few and inconsistently observed to draw
conclusions from.
35
The policy implications of these results are varied. On the one hand, the results
seem to indicate that policy makers should not be overly concerned about any potential
effects on morality when crafting early childhood education policies that increase
attainment. The evidence that suggests that elementary school education has no effect on
morality would seem to logically extend to even earlier formal education: In general, it
appears that education’s relationship with morality is weaker with more rudimentary and
provincial education. On the other hand, policies designed to improve college attendance
and graduation rates should be designed and pursued with potential shifts in morality in
mind. Furthermore, if moral development is a desired end of education, policy makers
and experts might examine why high school education seems to fail in impacting
individuals’ moral attitudes. On a more cynical note, professional politicians will want to
take notice of the way educational attainment shapes moral attitudes, and thus political
ideology.
There are a multitude of ways this study could be improved upon, which would
allow one to estimate with more certainty whether there is a causal effect of educational
attainment on morality. In particular, by using the non-public use version of the GSS
data, geographic fixed effects could be added at a much more granular level, and perhaps
a valid instrument could be applied (as in Dee’s 2004 study).
Further research into the effects of college education is particularly warranted. It
would be valuable to know if the observed relationship between morality and college
attainment is in fact driven by attainment, or whether matriculation or graduation has a
disproportionate impact on the relationship. It would also be helpful to know whether the
relationship varies with the type of institution attended (e.g., comparing the impact of
36
attending a religious institution to the impact of attending a secular one). Lastly, it would
be useful, if possible, to design a procedure to disentangle the impact of attainment in
general from the impact of instruction in specific subject matter.
37
TABLE 1: SUMMARY STATISTICS FOR VARIABLES INCLUDED IN “MORALITY”
Variable N
Percentage Responding
“Yes” or “Agree” Nature of Morality
World is Evil 11,569 24.6% Human Nature is Good 8,814 77.4% Morality is a Personal Matter 8,141 75.4% Science Breaks Down Morals 1,408 32.8%
Moral Reasoning Morality has Shades of Gray 8,180 83.7% Sinners Should be Punished 7,869 47.4% One Immoral Person Corrupts Society 8,200 62.1% Dissenters Should be Treated Leniently 1,201 79.0%
Moral Attitudes towards Private Actions Premarital Sex is Wrong 31,864 37.2% Homosexual Sex is Wrong 31,059 71.6% Gays Have Right to Marry 6,022 38.2% Woman Has Right to Abortion 30,173 40.4% Terminal Patient has Right to Die 26,479 67.6% Sometimes Okay to Hit Another 18,126 66.3%
Moral Attitudes towards Responsibility Personal Responsibility for Reducing Pain 1,390 62.3% Friend Has Right to False Testimony 1,099 12.5% Tell Police Truth Against Friend's Wishes 966 87.7% No Morality in Way One Makes Money 4,328 24.7%
Moral Attitudes towards Government Approve of Death Penalty 43,890 72.5% Not that Bad to Cheat on Taxes 2,460 15.6% Not that Bad to Cheat Government 2,508 5.2% Support Country if Country was Wrong 1,980 42.8%
Moral Attitudes towards Inequality Government Should Reduce Inequality 21,521 58.1% Government Should Help the Poor 14,386 54.5% Income Inequality in US is Too Large 4,074 78.5% Just that Rich get Better Health Care 914 36.1% Just that Rich Kids get Better Education 945 39.6%
38
TABLE 2: SUMMARY STATISTICS FOR THE VARIABLES OF INTEREST AND CONTROL VARIABLES Variable N Mean Std. Dev. Variables of Interest
Education, Years 52,816 12.73 3.17 Elementary Education, Years 52,816 7.86 0.76 High School Education, Years 52,816 3.36 1.30 College Education, Years 52,816 1.30 1.64 Postgraduate Education, Years 52,816 0.22 0.74
Biographic and Demographic Controls Age 52,816 45.32 17.24 Male 52,816 0.441 -- White 52,816 0.818 -- Black 52,816 0.137 -- Foreign Born 52,816 0.071 --
Parental Status Controls Mother's Education, Years 44,722 10.70 3.72 Father's Education, Years 38,360 10.53 4.34 Number of Grandparents Born in US 42,603 1.15 1.60 Number of Parents Born in US 45,009 1.71 0.67
Childhood Environment Controls Number of Siblings 52,816 3.93 3.11 Family Income at Age 16 was Below Average 41,307 0.328 -- Family Income at Age 16 was Above Average 41,307 0.168 -- Raised as Protestant 52,816 0.598 -- Raised as Catholic 52,816 0.277 -- Raised as Jew 52,816 0.020 -- Not Raised in Religion 52,816 0.048 -- Lived on a Farm at age 16 52,816 0.160 -- Lived in a Small Town at Age 16 52,816 0.316 -- Lived in a Medium Sized City at age 16 52,816 0.150 -- Lived in a Suburb of a Large City at age 16 52,816 0.107 -- Lived in a Large City at Age 16 52,816 0.155 --
39
TABLE 3: ESTIMATES OF EDUCATION’S IMPACT ON VARIABLES REVEALING MORALITY
Bivariate
Regression
Add Bio/Demo Controls
Add Parent Controls
Add Childhood Controls
Add Fixed Effects
(1) (2) (3) (4) (5) Nature of Morality
World is Evil -0.0229*** -0.0251*** -0.0202*** -0.0206*** -0.0201***
(0.0019) (0.0020) (0.0026) (0.0036) (0.0034)
11,569 11,569 7,533 3,928 3,928
Human Nature is Good 0.0200*** 0.0191*** 0.0151*** 0.0123*** 0.0122***
(0.0026) (0.0022) (0.0031) (0.0034) (0.0035)
8,814 8,814 5,780 3,925 3,925
Morality is a Personal Matter -0.0070*** -0.0076*** -0.0080*** -0.0075*** -0.0077***
(0.0018) (0.0016) (0.0016) (0.0022) (0.0021)
8,141 8,141 5,481 5,408 5,408
Science Breaks Down Morals+ -0.0426*** -0.0384*** -0.0341*** -0.0306*** -0.0295***
(0.0058) (0.0056) (0.0068) (0.0073) (0.0073)
1,408 1,408 917 911 911
Moral Reasoning Morality has Shades of Gray 0.0083*** 0.0084*** 0.0106*** 0.0100*** 0.0097***
(0.0018) (0.0019) (0.0021) (0.0023) (0.0022)
8,180 8,180 5,519 5,442 5,442
Sinners Should be Punished -0.0314*** -0.0300*** -0.0213*** -0.0200*** -0.0195***
(0.0026) (0.0021) (0.0023) (0.0029) (0.0029)
7,869 7,869 5,293 5,222 5,222
One Immoral Person Corrupts Society -0.0088*** -0.0078** -0.0065** -0.0063** -0.0067**
(0.0018) (0.0024) (0.0021) (0.0022) (0.0024)
8,200 8,200 5,522 5,447 5,447
Dissenters Should be Treated Leniently+ 0.0082 0.0069 0.0094 0.0087 0.0091
(0.0062) (0.0052) (0.0057) (0.0058) (0.0063)
1,201 1,201 790 787 787
Moral Attitudes towards Private Actions Premarital Sex is Wrong -0.0258*** -0.0173*** -0.0105*** -0.0075*** -0.0069***
(0.0020) (0.0013) (0.0012) (0.0012) (0.0012)
31,864 31,864 17,891 13,291 13,291
Homosexual Sex is Wrong -0.0394*** -0.0344*** -0.0279*** -0.0248*** -0.0230***
(0.0039) (0.0040) (0.0049) (0.0039) (0.0037)
31,059 31,059 17,441 13,415 13,415
Gays Have Right to Marry 0.0294*** 0.0256*** 0.0180*** 0.0159*** 0.0144***
(0.0030) (0.0029) (0.0043) (0.0036) (0.0024)
6,022 6,022 4,057 3,927 3,927
Woman Has Right to Abortion 0.0335*** 0.0320*** 0.0217*** 0.0198*** 0.0198***
(0.0013) (0.0012) (0.0011) (0.0009) (0.0009)
30,173 30,173 19,643 15,504 15,504
Terminal Has Right to Die 0.0180*** 0.0117*** 0.0027 -0.0017 -0.0025
(0.0029) (0.0021) (0.0015) (0.0019) (0.0021)
26,479 26,479 17,213 12,694 12,694
Sometimes Okay to Hit 0.0277*** 0.0210*** 0.0186*** 0.0172*** 0.0177***
(0.0015) (0.0015) (0.0022) (0.0021) (0.0020)
18,126 18,126 9,102 9,070 9,070
40
Bivariate Regression
Add Bio/Demo Controls
Add Parent Controls
Add Childhood Controls
Add Fixed Effects
(1) (2) (3) (4) (5) Moral Attitudes towards Responsibility and Loyalty
Personal Responsibility for Reducing Pain++ 0.0074 0.0092 0.0104 0.0139 0.0126
(0.0051) (0.0055) (0.0081) (0.0088) (0.0081)
1,390 1,390 889 889 889
Friend Has Right to False Testimony++ -0.0034 -0.0066* -0.0113** -0.0118** -0.0115**
(0.0029) (0.0031) (0.0041) (0.0038) (0.0037)
1,099 1,099 714 714 714
Tell Police Truth Against Friend's Wishes++ 0.0033 0.0060 0.0124** 0.0134** 0.0128**
(0.0039) (0.0036) (0.0049) (0.0051) (0.0053)
966 966 631 631 631
No Morality in Way One Makes Money+++ -0.0283*** -0.0297*** -0.0285*** -0.0286*** -0.0289***
(0.0020) (0.0025) (0.0025) (0.0029) (0.0028)
4,328 4,328 2,947 2,935 2,935
Moral Attitudes towards Government and Citizenship Approve of Death Penalty -0.0015 -0.0055 -0.0097** -0.0105** -0.0104**
(0.0043) (0.0037) (0.0040) (0.0040) (0.0040)
43,890 43,890 25,983 19,752 19,752
Not that Bad to Cheat on Taxes -0.0037 -0.0053* -0.0055* -0.0032 -0.0037
(0.0025) (0.0027) (0.0026) (0.0046) (0.0048)
2,460 2,460 1,545 793 793
Not that Bad to Cheat Government -0.0062*** -0.0061*** -0.0067*** -0.0070*** -0.0073***
(0.0010) (0.0010) (0.0011) (0.0021) (0.0022)
2,508 2,508 1,575 800 800
Support Country if Country was Wrong -0.0329*** -0.0312*** -0.0353*** -0.0421*** -0.0411***
(0.0043) (0.0045) (0.0052) (0.0098) (0.0100)
1,980 1,980 1,365 590 590
Moral Attitudes towards Economic Inequality Government Should Reduce Inequality -0.0256*** -0.0246*** -0.0175*** -0.0165*** -0.0169***
(0.0022) (0.0023) (0.0025) (0.0024) (0.0025)
21,521 21,521 14,203 10,829 10,829
Government Should Help the Poor -0.0258*** -0.0242*** -0.0148*** -0.0143*** -0.0141***
(0.0029) (0.0026) (0.0030) (0.0030) (0.0031)
14,386 14,386 8,780 6,461 6,461
Income Inequality in US is Too Large -0.0042 -0.0033 -0.0004 0.0064 0.0065
(0.0030) (0.0028) (0.0028) (0.0047) (0.0049)
4,074 4,074 2,599 1,286 1,286
Just that Rich get Better Health Care++ -0.0115** -0.0099* -0.0035 -0.0024 -0.0045
(0.0044) (0.0052) (0.0089) (0.0082) (0.0085)
914 914 574 574 574
Just that Rich Kids get Better Education++ -0.0066 -0.0061 0.0049 0.0046 0.0051
(0.0060) (0.0070) (0.0066) (0.0059) (0.0061)
945 945 583 583 583
Each parameter estimate within the table corresponds to a separate OLS regression, where the dependent variable (representing morality) is listed in the left-hand column. Standard errors, corrected for clustering at the region level, are in parentheses below the parameter estimate, and the number of observations is listed below the standard error. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level. One cross (+) denotes that only region fixed effects, and not year fixed effects, were included in column (5). Two crosses (++) denote that only region fixed effects were included in column (5), and that controls for relative childhood income were not included in columns (4) and (5). Three crosses (+++) denote that controls for whether the respondent was born in the US, and the number of his parents and grandparents that were born in the US, were not included in columns (3) through (5).
41
TABLE 4A: REGRESSION RESULTS FOR “WORLD IS EVIL”
(1) (2) (3) (4) (5) Education, Years -0.0229*** -0.0251*** -0.0202*** -0.0206*** -0.0201***
(0.0019) (0.0020) (0.0026) (0.0036) (0.0034)
Age
-0.0027*** -0.0030*** -0.0034*** -0.0031***
(0.0003) (0.0004) (0.0004) (0.0005)
Male
0.0555*** 0.0586*** 0.0464** 0.0433**
(0.0085) (0.0101) (0.0161) (0.0165)
White
-0.0770** -0.0607 -0.0658 -0.0405
(0.0280) (0.0339) (0.0547) (0.0437)
Black
0.0259 0.0385 0.0453 0.0740
(0.0335) (0.0457) (0.0623) (0.0493)
Foreign Born
-0.0089 0.0194 -0.0230 0.0019
(0.0170) (0.0248) (0.0265) (0.0374)
Mother's Education, Years
-0.0045* -0.0060** -0.0055*
(0.0021) (0.0026) (0.0028)
Father's Education, Years
-0.0039** -0.0039 -0.0035
(0.0017) (0.0029) (0.0026)
Number of Grandparents Born in US
-0.0094 -0.0086 -0.0055
(0.0065) (0.0083) (0.0077)
Number of Parents Born in US
0.0033 -0.0064 -0.0107
(0.0113) (0.0163) (0.0166)
Number of Siblings
0.0035* 0.0030*
(0.0016) (0.0015)
Family Income at Age 16 was Below Average
0.0101 0.0113
(0.0195) (0.0196)
Family Income at Age 16 was Above Average
-0.0217 -0.0215*
(0.0121) (0.0109)
Raised as Protestant
0.0762* 0.0690*
(0.0337) (0.0320)
Raised as Catholic
0.0158 0.0200
(0.0360) (0.0334)
Raised as Jew
0.0214 0.0198
(0.0386) (0.0390)
Not Raised in Religion
0.0208 0.0136
(0.0405) (0.0413)
Lived on a Farm at age 16
-0.0550** -0.0656**
(0.0185) (0.0216)
Lived in a Small Town at Age 16
-0.0418 -0.0466*
(0.0279) (0.0253)
Lived in a Medium Sized City at age 16
-0.0349* -0.0436**
(0.0164) (0.0175)
Lived in a Suburb of a Large City at age 16
-0.0613* -0.0567*
(0.0292) (0.0286)
Lived in a Large City at Age 16
0.0099 0.0069
(0.0292) (0.0280)
Regional Fixed Effects No No No No Yes Year Fixed Effects No No No No Yes N 11,569 11,569 7,533 3,928 3,928 Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent believes the world is evil. Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
42
TABLE 4B: REGRESSION RESULTS FOR “MORALITY HAS SHADES OF GRAY”
(1) (2) (3) (4) (5) Education, Years 0.0083*** 0.0084*** 0.0106*** 0.0100*** 0.0097***
(0.0018) (0.0019) (0.0021) (0.0023) (0.0022)
Age
-0.0002 -0.0001 0.0002 0.0002
(0.0003) (0.0005) (0.0004) (0.0005)
Male
-0.0204* -0.0224* -0.0228* -0.0219*
(0.0094) (0.0110) (0.0117) (0.0116)
White
-0.0169 -0.0238 -0.0196 -0.0390
(0.0222) (0.0201) (0.0195) (0.0244)
Black
-0.0241 -0.0311 -0.0126 -0.0230
(0.0261) (0.0226) (0.0217) (0.0260)
Foreign Born
-0.0031 -0.0377* -0.0343 0.0068
(0.0081) (0.0179) (0.0202) (0.0291)
Mother's Education, Years
-0.0014 -0.0015 -0.0016
(0.0018) (0.0018) (0.0017)
Father's Education, Years
0.0004 -0.0007 -0.0009
(0.0011) (0.0015) (0.0014)
Number of Grandparents Born in US
0.0121 0.0003 -0.0037
(0.0068) (0.0057) (0.0053)
Number of Parents Born in US
0.0054 0.0052 0.0005
(0.0064) (0.0048) (0.0058)
Number of Siblings
-0.0011 -0.0007
(0.0025) (0.0024)
Family Income at Age 16 was Below Average
-0.0080 -0.0060
(0.0080) (0.0081)
Family Income at Age 16 was Above Average
0.0054 0.0055
(0.0121) (0.0127)
Raised as Protestant
-0.0473 -0.0470
(0.0373) (0.0375)
Raised as Catholic
0.0259 0.0173
(0.0415) (0.0401)
Raised as Jew
0.0754 0.0700
(0.0462) (0.0504)
Not Raised in Religion
-0.0111 -0.0098
(0.0273) (0.0287)
Lived on a Farm at age 16
-0.0361 -0.0401**
(0.0211) (0.0164)
Lived in a Small Town at Age 16
-0.0146 -0.0173
(0.0120) (0.0105)
Lived in a Medium Sized City at age 16
-0.0128 -0.0124
(0.0195) (0.0179)
Lived in a Suburb of a Large City at age 16
-0.0005 -0.0043
(0.0150) (0.0125)
Lived in a Large City at Age 16
0.0141 0.0086
(0.0084) (0.0077)
Regional Fixed Effects No No No No Yes Year Fixed Effects No No No No Yes N 8,180 8,180 5,519 5,442 5,442 Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent believes that morality has shades of gray (as opposed to viewing moral issues in black and white). Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
43
TABLE 4C: REGRESSION RESULTS FOR “WOMAN HAS A RIGHT TO ABORTION”
(1) (2) (3) (4) (5) Education, Years 0.0335*** 0.0320*** 0.0217*** 0.0198*** 0.0198***
(0.0013) (0.0012) (0.0011) (0.0009) (0.0009)
Age
-0.0011*** -0.0001 0.0001 0.0001
(0.0002) (0.0003) (0.0003) (0.0004)
Male
0.0099 0.0023 0.0051 0.0063
(0.0072) (0.0091) (0.0104) (0.0104)
White
0.0489 0.0634* 0.0270 0.0181
(0.0428) (0.0324) (0.0298) (0.0272)
Black
0.0328 0.0865** 0.0440 0.0455
(0.0472) (0.0364) (0.0307) (0.0275)
Foreign Born
0.0186 -0.0385 -0.0208 0.0077
(0.0221) (0.0287) (0.0284) (0.0261)
Mother's Education, Years
0.0095*** 0.0067** 0.0062**
(0.0021) (0.0021) (0.0020)
Father's Education, Years
0.0098*** 0.0067*** 0.0060**
(0.0013) (0.0019) (0.0021)
Number of Grandparents Born in US
0.0208*** 0.0165** 0.0093*
(0.0061) (0.0062) (0.0042)
Number of Parents Born in US
-0.0210 -0.0079 -0.0062
(0.0126) (0.0125) (0.0120)
Number of Siblings
-0.0069** -0.0063**
(0.0022) (0.0022)
Family Income at Age 16 was Below Average
0.0222** 0.0247**
(0.0089) (0.0092)
Family Income at Age 16 was Above Average
0.0265* 0.0267*
(0.0142) (0.0141)
Raised as Protestant
-0.0069 0.0015
(0.0234) (0.0243)
Raised as Catholic
-0.0520** -0.0578***
(0.0194) (0.0167)
Raised as Jew
0.2057*** 0.1983***
(0.0267) (0.0227)
Not Raised in Religion
0.0888*** 0.0865**
(0.0255) (0.0272)
Lived on a Farm at age 16
-0.0260** -0.0151
(0.0113) (0.0113)
Lived in a Small Town at Age 16
0.0114 0.0088
(0.0094) (0.0083)
Lived in a Medium Sized City at age 16
0.0579*** 0.0550***
(0.0143) (0.0164)
Lived in a Suburb of a Large City at age 16
0.1203*** 0.1097***
(0.0193) (0.0182)
Lived in a Large City at Age 16
0.0825*** 0.0748***
(0.0167) (0.0151)
Regional Fixed Effects No No No No Yes Year Fixed Effects No No No No Yes
N 30,173 30,173 19,643 15,504 15,504 Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent believes that women have the right to have an abortion for any reason. Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
44
TABLE 4D: REGRESSION RESULTS FOR “PERSONAL RESPONSIBILITY FOR REDUCING PAIN”
(1) (2) (3) (4) (5) Education, Years 0.0074 0.0092 0.0104 0.0139 0.0126
(0.0051) (0.0055) (0.0081) (0.0088) (0.0081)
Age
0.0011 0.0009 0.0004 0.0004
(0.0008) (0.0007) (0.0008) (0.0008)
Male
-0.0650** -0.0491 -0.0403 -0.0397
(0.0244) (0.0304) (0.0315) (0.0311)
White
-0.0441 -0.1592*** -0.1833*** -0.1733***
(0.0411) (0.0350) (0.0458) (0.0451)
Black
-0.0379 -0.1567* -0.1736** -0.1516*
(0.0580) (0.0743) (0.0668) (0.0759)
Foreign Born
-0.0857** -0.0233 -0.0354 -0.0590
(0.0379) (0.0572) (0.0542) (0.0940)
Mother's Education, Years
0.0006 0.0012 0.0004
(0.0077) (0.0072) (0.0074)
Father's Education, Years
-0.0022 -0.0001 0.0017
(0.0049) (0.0046) (0.0045)
Number of Grandparents Born in US
0.0127 0.0297** 0.0349**
(0.0093) (0.0126) (0.0121)
Number of Parents Born in US
0.0813*** 0.0735*** 0.0791***
(0.0173) (0.0170) (0.0160)
Number of Siblings
0.0052 0.0045
(0.0068) (0.0071)
Raised as Protestant
0.1772 0.1793*
(0.0971) (0.0929)
Raised as Catholic
0.1144 0.1318
(0.1059) (0.1091)
Raised as Jew
0.0785 0.1025
(0.1361) (0.1459)
Not Raised in Religion
0.0580 0.0592
(0.1046) (0.1034)
Lived on a Farm at age 16
0.0895* 0.0902*
(0.0439) (0.0461)
Lived in a Small Town at Age 16
-0.0193 -0.0056
(0.0296) (0.0328)
Lived in a Medium Sized City at age 16
0.0312 0.0318
(0.0635) (0.0690)
Lived in a Suburb of a Large City at age 16
-0.0249 -0.0202
(0.0572) (0.0565)
Lived in a Large City at Age 16
-0.1315 -0.1281
(0.0782) (0.0823)
Regional Fixed Effects No No No No Yes
N 1,390 1,390 889 889 889
Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent believes that he or she has a personal responsibility for reducing pain in the world. Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
45
TABLE 4E: REGRESSION RESULTS FOR “SUPPORT COUNTRY IF COUNTRY WAS WRONG”
(1) (2) (3) (4) (5) Education, Years -0.0329*** -0.0312*** -0.0353*** -0.0421*** -0.0411***
(0.0043) (0.0045) (0.0052) (0.0098) (0.0100)
Age
0.0028*** 0.0031*** 0.0039*** 0.0036**
(0.0005) (0.0009) (0.0010) (0.0012)
Male
0.0102 0.0225 0.0385 0.0437
(0.0121) (0.0246) (0.0315) (0.0298)
White
-0.0000 0.0084 -0.0051 -0.0004
(0.0314) (0.0435) (0.0604) (0.0668)
Black
0.0020 0.0332 -0.0543 -0.0411
(0.0325) (0.0394) (0.0939) (0.1060)
Foreign Born
0.0654** 0.0857* -0.0097 -0.0574
(0.0263) (0.0450) (0.0601) (0.1151)
Mother's Education, Years
-0.0043 -0.0077 -0.0092
(0.0041) (0.0077) (0.0079)
Father's Education, Years
0.0050 0.0004 0.0011
(0.0033) (0.0034) (0.0037)
Number of Grandparents Born in US
-0.0219 -0.0320 -0.0321
(0.0183) (0.0306) (0.0290)
Number of Parents Born in US
-0.0515 -0.0402 -0.0207
(0.0414) (0.0472) (0.0384)
Number of Siblings
-0.0057 -0.0071
(0.0099) (0.0089)
Family Income at Age 16 was Below Average
0.0173 0.0250
(0.0292) (0.0294)
Family Income at Age 16 was Above Average
0.0684 0.0554
(0.0526) (0.0524)
Raised as Protestant
-0.1619* -0.1470*
(0.0721) (0.0714)
Raised as Catholic
-0.1138* -0.0841
(0.0553) (0.0518)
Raised as Jew
-0.0002 0.0192
(0.0884) (0.0966)
Not Raised in Religion
-0.1689** -0.1710**
(0.0599) (0.0579)
Lived on a Farm at age 16
0.0190 0.0046
(0.0749) (0.0801)
Lived in a Small Town at Age 16
0.0662 0.0522
(0.0848) (0.0916)
Lived in a Medium Sized City at age 16
-0.0206 -0.0264
(0.0721) (0.0743)
Lived in a Suburb of a Large City at age 16
0.1315* 0.1093
(0.0691) (0.0780)
Lived in a Large City at Age 16
-0.0524 -0.0576
(0.0708) (0.0789)
Regional Fixed Effects No No No No Yes Year Fixed Effects No No No No Yes
N 1,980 1,980 1,365 590 590
Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent would support the country in actions that the respondent personally believes are wrong. Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
46
TABLE 4F: REGRESSION RESULTS FOR “INEQUALITY IN US IS TOO LARGE”
(1) (2) (3) (4) (5) Education, Years -0.0042 -0.0033 -0.0004 0.0064 0.0065
(0.0030) (0.0028) (0.0028) (0.0047) (0.0049)
Age
0.0002 -0.0007 -0.0007 -0.0010
(0.0003) (0.0005) (0.0009) (0.0010)
Male
-0.0691*** -0.0939*** -0.1171*** -0.1191***
(0.0113) (0.0154) (0.0268) (0.0292)
White
-0.0212 -0.0144 0.0054 0.0274
(0.0356) (0.0567) (0.0739) (0.0729)
Black
-0.0176 -0.0058 0.0123 0.0341
(0.0403) (0.0575) (0.0641) (0.0655)
Foreign Born
0.0708*** 0.0808* 0.1107** 0.1593**
(0.0215) (0.0376) (0.0405) (0.0593)
Mother's Education, Years
-0.0018 0.0005 -0.0004
(0.0025) (0.0039) (0.0038)
Father's Education, Years
-0.0042 -0.0071 -0.0081*
(0.0025) (0.0040) (0.0039)
Number of Grandparents Born in US
0.0022 0.0068 0.0095
(0.0054) (0.0129) (0.0121)
Number of Parents Born in US
-0.0089 -0.0014 -0.0050
(0.0203) (0.0348) (0.0401)
Number of Siblings
-0.0009 0.0005
(0.0032) (0.0026)
Family Income at Age 16 was Below Average
0.0485 0.0438
(0.0421) (0.0418)
Family Income at Age 16 was Above Average
-0.0270 -0.0278
(0.0468) (0.0460)
Raised as Protestant
0.0145 -0.0021
(0.0608) (0.0495)
Raised as Catholic
0.0112 -0.0086
(0.0538) (0.0493)
Raised as Jew
-0.1552 -0.1613
(0.1407) (0.1444)
Not Raised in Religion
0.0734 0.0487
(0.0929) (0.0865)
Lived on a Farm at age 16
0.0150 0.0058
(0.0456) (0.0489)
Lived in a Small Town at Age 16
-0.0116 -0.0131
(0.0450) (0.0453)
Lived in a Medium Sized City at age 16
-0.0465 -0.0558
(0.0606) (0.0593)
Lived in a Suburb of a Large City at age 16
0.0359 0.0236
(0.0421) (0.0423)
Lived in a Large City at Age 16
0.0348 0.0220
(0.0565) (0.0486)
Regional Fixed Effects No No No No Yes Year Fixed Effects No No No No Yes
N 4,074 4,074 2,599 1,286 1,286
Parameter estimates reported in this table were obtained through OLS regressions of the variables listed in the left-hand column against a binary variable capturing whether the respondent believes that inequality in the United States is too large. Standard errors, corrected for clustering at the region level, are listed in parentheses below parameter estimates. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level.
47
TABLE 5A: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON PERCEPTIONS OF THE NATURE OF MORALITY
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
World is Evil
Elementary Education -0.0075 -0.0049 -0.0117 -0.0048 -0.0013
(0.0055) (0.0071) (0.0201) (0.0166) (0.0161)
High School Education -0.0221*** -0.0311*** -0.0275** -0.0351*** -0.0340***
(0.0056) (0.0059) (0.0088) (0.0072) (0.0075)
College Education -0.0309*** -0.0321*** -0.0254*** -0.0262*** -0.0261***
(0.0043) (0.0042) (0.0037) (0.0056) (0.0054)
Post-Graduate Education -0.0101** -0.0059 -0.0025 0.0058 0.0060
(0.0040) (0.0042) (0.0041) (0.0062) (0.0063)
N 11,569 11,569 7,533 3,928 3,928
Human Nature is Good Elementary Education 0.0168* 0.0079 0.0227 0.0342 0.0328
(0.0078) (0.0064) (0.0161) (0.0191) (0.0189)
High School Education 0.0282*** 0.0312*** 0.0230** 0.0214** 0.0195*
(0.0061) (0.0060) (0.0084) (0.0085) (0.0086)
College Education 0.0183*** 0.0178*** 0.0150*** 0.0099** 0.0110**
(0.0025) (0.0023) (0.0039) (0.0038) (0.0037)
Post-Graduate Education 0.0109 0.0079 0.0052 0.0008 0.0004
(0.0062) (0.0061) (0.0061) (0.0043) (0.0047)
N 8,814 8,814 5,780 3,925 3,925
Morality is a Personal Matter Elementary Education 0.0106** 0.0083* -0.0042 -0.0062 -0.0067
(0.0037) (0.0044) (0.0063) (0.0071) (0.0071)
High School Education -0.0037 -0.0052 -0.0059 -0.0031 -0.0037
(0.0062) (0.0065) (0.0095) (0.0104) (0.0104)
College Education -0.0170*** -0.0171*** -0.0140*** -0.0139*** -0.0141***
(0.0035) (0.0034) (0.0029) (0.0028) (0.0028)
Post-Graduate Education 0.0032 0.0047 0.0021 0.0026 0.0031
(0.0058) (0.0059) (0.0074) (0.0072) (0.0071)
N 8,141 8,141 5,481 5,408 5,408
Science Breaks Down Morals+ Elementary Education 0.0110 0.0211 0.0016 0.0092 0.0059
(0.0257) (0.0242) (0.0530) (0.0529) (0.0546)
High School Education -0.0518*** -0.0416*** -0.0393 -0.0358 -0.0326
(0.0112) (0.0097) (0.0232) (0.0267) (0.0267)
College Education -0.0537*** -0.0522*** -0.0394*** -0.0372*** -0.0371***
(0.0083) (0.0089) (0.0074) (0.0071) (0.0074)
Post-Graduate Education -0.0242** -0.0288** -0.0256** -0.0199* -0.0170
(0.0095) (0.0092) (0.0088) (0.0097) (0.0097)
N 1,408 1,408 917 911 911
Each set of four parameter estimates (arranged vertically) within the table corresponds to a separate OLS regression, where the dependent variable (representing morality) is listed in the left-hand column. Standard errors, corrected for clustering at the region level, are in parentheses below the parameter estimates, and the number of observations is listed below each set of four parameter estimates and standard errors. One asterisk (*) denotes statistical significance at the p < 10% level, two asterisks (**) denote significance at the p < 5% level, and three asterisks (***) denote significance at the p < 1% level. A cross (+) denotes that only region fixed effects, and not year fixed effects, were included in the regression in column (5). Two crosses (++) denote that only region fixed effects were included in column (5), and that controls for relative childhood income were not included in columns (4) and (5). Three crosses (+++) denote that controls for whether the respondent was born in the US, and the number of his parents and grandparents that were born in the US, were not included in columns (3) through (5).
48
TABLE 5B: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON MORAL REASONING
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
Morality has Shades of Gray
Elementary Education 0.0126** 0.0133** 0.0067 0.0072 0.0061
(0.0052) (0.0052) (0.0054) (0.0055) (0.0058)
High School Education 0.0072 0.0071 0.0125 0.0109 0.0101
(0.0069) (0.0069) (0.0068) (0.0072) (0.0072)
College Education 0.0063** 0.0063** 0.0101** 0.0098** 0.0101**
(0.0026) (0.0027) (0.0031) (0.0036) (0.0034)
Post-Graduate Education 0.0127* 0.0130* 0.0116 0.0108 0.0100
(0.0058) (0.0058) (0.0072) (0.0069) (0.0067)
N 8,180 8,180 5,519 5,442 5,442
Sinners Should be Punished Elementary Education 0.0047 0.0025 -0.0085 -0.0092 -0.0108
(0.0074) (0.0068) (0.0059) (0.0064) (0.0063)
High School Education -0.0436*** -0.0430*** -0.0107 -0.0095 -0.0033
(0.0084) (0.0076) (0.0082) (0.0090) (0.0090)
College Education -0.0338*** -0.0303*** -0.0247*** -0.0233*** -0.0240***
(0.0053) (0.0055) (0.0046) (0.0049) (0.0052)
Post-Graduate Education -0.0326*** -0.0323*** -0.0295*** -0.0264*** -0.0271***
(0.0080) (0.0071) (0.0072) (0.0067) (0.0076)
N 7,869 7,869 5,293 5,222 5,222
One Immoral Person Corrupts Society Elementary Education 0.0057 0.0117 0.0103 0.0053 0.0082
(0.0078) (0.0067) (0.0088) (0.0090) (0.0099)
High School Education -0.0170** -0.0154* -0.0121 -0.0102 -0.0109
(0.0053) (0.0078) (0.0123) (0.0124) (0.0131)
College Education -0.0065 -0.0068 -0.0051 -0.0039 -0.0056
(0.0037) (0.0037) (0.0050) (0.0048) (0.0045)
Post-Graduate Education -0.0134* -0.0129* -0.0130 -0.0136 -0.0126
(0.0060) (0.0064) (0.0092) (0.0091) (0.0093)
N 8,200 8,200 5,522 5,447 5,447
Dissenters Should be Treated Leniently+ Elementary Education -0.0199 -0.0174 0.0060 0.0043 0.0026
(0.0205) (0.0181) (0.0305) (0.0296) (0.0300)
High School Education 0.0107 0.0080 -0.0052 0.0007 0.0001
(0.0194) (0.0184) (0.0278) (0.0261) (0.0276)
College Education 0.0039 0.0026 0.0086 0.0074 0.0077
(0.0105) (0.0104) (0.0114) (0.0108) (0.0115)
Post-Graduate Education 0.0359 0.0342 0.0235 0.0192 0.0210
(0.0198) (0.0207) (0.0222) (0.0224) (0.0225)
N 1,201 1,201 790 787 787
See notes to Table 5a.
49
TABLE 5C: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON MORAL ATTITUDES TOWARDS PRIVATE ACTIONS
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
Premarital Sex is Wrong Elementary Education -0.0114 -0.0051 -0.0016 -0.0052 -0.0028
(0.0062) (0.0047) (0.0076) (0.0045) (0.0035)
High School Education -0.0459*** -0.0228*** -0.0148*** -0.0094** -0.0054
(0.0036) (0.0034) (0.0040) (0.0034) (0.0039)
College Education -0.0174*** -0.0142*** -0.0039 -0.0003 -0.0011
(0.0031) (0.0025) (0.0024) (0.0034) (0.0032)
Post-Graduate Education -0.0187*** -0.0258*** -0.0260*** -0.0253*** -0.0260***
(0.0049) (0.0042) (0.0048) (0.0061) (0.0062)
N 31,864 31,864 17,891 13,291 13,291 Homosexual Sex is Wrong
Elementary Education -0.0029 0.0057 0.0182*** 0.0111** 0.0084
(0.0034) (0.0033) (0.0039) (0.0045) (0.0069)
High School Education -0.0307*** -0.0157*** -0.0019 0.0040** 0.0105***
(0.0024) (0.0021) (0.0022) (0.0017) (0.0017)
College Education -0.0563*** -0.0550*** -0.0404*** -0.0368*** -0.0352***
(0.0051) (0.0050) (0.0054) (0.0045) (0.0037)
Post-Graduate Education -0.0398*** -0.0455*** -0.0457*** -0.0465*** -0.0472***
(0.0039) (0.0035) (0.0028) (0.0047) (0.0049)
N 31,059 31,059 17,441 13,415 13,415 Gays Have Right to Marry
Elementary Education -0.0005 -0.0027 -0.0101 -0.0037 0.0051
(0.0076) (0.0074) (0.0077) (0.0094) (0.0076)
High School Education 0.0308*** 0.0170* 0.0043 -0.0008 -0.0115
(0.0082) (0.0078) (0.0097) (0.0100) (0.0106)
College Education 0.0359*** 0.0330*** 0.0214*** 0.0187*** 0.0169***
(0.0051) (0.0044) (0.0049) (0.0042) (0.0033)
Post-Graduate Education 0.0300*** 0.0359*** 0.0354*** 0.0321*** 0.0336***
(0.0062) (0.0058) (0.0066) (0.0055) (0.0052)
N 6,022 6,022 4,057 3,927 3,927 Woman Has Right to Abortion
Elementary Education 0.0072 0.0051 0.0045 0.0117** 0.0096**
(0.0054) (0.0055) (0.0048) (0.0050) (0.0042)
High School Education 0.0360*** 0.0317*** 0.0124** 0.0072 0.0058
(0.0029) (0.0023) (0.0038) (0.0044) (0.0045)
College Education 0.0413*** 0.0407*** 0.0287*** 0.0263*** 0.0273***
(0.0018) (0.0017) (0.0017) (0.0020) (0.0018)
Post-Graduate Education 0.0254*** 0.0267*** 0.0225*** 0.0222*** 0.0228***
(0.0040) (0.0039) (0.0040) (0.0055) (0.0058)
N 30,173 30,173 19,643 15,504 15,504 Terminal Patient Has Right to Die
Elementary Education 0.0295*** 0.0203*** 0.0078 0.0024 0.0009
(0.0048) (0.0036) (0.0073) (0.0075) (0.0071)
High School Education 0.0313*** 0.0216*** 0.0124** 0.0089 0.0050
(0.0040) (0.0034) (0.0053) (0.0064) (0.0064)
College Education 0.0108*** 0.0062** -0.0020 -0.0076*** -0.0076***
(0.0031) (0.0023) (0.0018) (0.0016) (0.0017)
Post-Graduate Education 0.0029 0.0025 0.0015 -0.0015 -0.0003
(0.0038) (0.0036) (0.0039) (0.0047) (0.0049)
N 1,408 1,408 917 911 911 Sometimes Okay to Hit Another
Elementary Education 0.0116* -0.0013 -0.0113 -0.0106 -0.0090
(0.0059) (0.0063) (0.0100) (0.0103) (0.0102)
High School Education 0.0444*** 0.0324*** 0.0251*** 0.0218*** 0.0232***
(0.0033) (0.0033) (0.0049) (0.0051) (0.0052)
College Education 0.0240*** 0.0220*** 0.0204*** 0.0192*** 0.0192***
(0.0035) (0.0028) (0.0039) (0.0037) (0.0037)
Post-Graduate Education 0.0082 0.0102* 0.0187*** 0.0182*** 0.0186***
(0.0053) (0.0055) (0.0044) (0.0043) (0.0044)
N 18,126 18,126 9,102 9,070 9,070
See notes to Table 5a.
50
TABLE 5D: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON MORAL ATTITUDES TOWARDS RESPONSIBILITY
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
Personal Responsibility for Reducing Pain++
Elementary Education -0.0016 -0.0039 0.0007 0.0071 0.0103
(0.0195) (0.0217) (0.0367) (0.0343) (0.0349)
High School Education -0.0201* -0.0181* -0.0390** -0.0327* -0.0316*
(0.0101) (0.0098) (0.0126) (0.0150) (0.0151)
College Education 0.0071 0.0096 0.0050 0.0078 0.0074
(0.0075) (0.0072) (0.0110) (0.0130) (0.0128)
Post-Graduate Education 0.0555** 0.0545** 0.0654*** 0.0665*** 0.0609***
(0.0189) (0.0195) (0.0193) (0.0186) (0.0186)
N 1,390 1,390 889 889 889
Friend Has Right to False Testimony++ Elementary Education -0.0113 -0.0120 -0.0419 -0.0409 -0.0363
(0.0224) (0.0216) (0.0269) (0.0266) (0.0278)
High School Education -0.0077 -0.0224** -0.0208 -0.0204 -0.0197
(0.0094) (0.0085) (0.0152) (0.0138) (0.0149)
College Education 0.0067 0.0027 -0.0043 -0.0058 -0.0056
(0.0056) (0.0061) (0.0084) (0.0084) (0.0079)
Post-Graduate Education -0.0216 -0.0086 -0.0089 -0.0084 -0.0096
(0.0161) (0.0166) (0.0155) (0.0158) (0.0169)
N 1,099 1,099 714 714 714
Tell Police Truth Against Friend's Wishes++ Elementary Education 0.0087 0.0143 0.0502* 0.0519* 0.0502*
(0.0181) (0.0169) (0.0225) (0.0256) (0.0257)
High School Education 0.0043 0.0116 0.0135 0.0115 0.0094
(0.0098) (0.0091) (0.0135) (0.0143) (0.0148)
College Education -0.0046 -0.0014 0.0030 0.0064 0.0072
(0.0078) (0.0074) (0.0082) (0.0071) (0.0070)
Post-Graduate Education 0.0224 0.0147 0.0175 0.0155 0.0132
(0.0133) (0.0150) (0.0153) (0.0155) (0.0155)
N 966 966 631 631 631
No Morality in Way One Makes Money+++ Elementary Education -0.0131 -0.0084 0.0104 0.0106 0.0096
(0.0087) (0.0072) (0.0188) (0.0189) (0.0191)
High School Education -0.0262*** -0.0312*** -0.0393*** -0.0401*** -0.0402***
(0.0066) (0.0071) (0.0085) (0.0087) (0.0088)
College Education -0.0463*** -0.0466*** -0.0423*** -0.0423*** -0.0424***
(0.0044) (0.0038) (0.0049) (0.0047) (0.0049)
Post-Graduate Education 0.0058 0.0071 0.0032 0.0022 0.0022
(0.0075) (0.0077) (0.0067) (0.0072) (0.0072)
N 4,328 4,328 2,947 2,935 2,935
See notes to Table 5a.
51
TABLE 5E: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON MORAL ATTITUDES TOWARDS GOVERNMENT
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
Approve of Death Penalty
Elementary Education 0.0211*** 0.0120 0.0161* 0.0148 0.0134
(0.0061) (0.0067) (0.0086) (0.0090) (0.0086)
High School Education 0.0212*** 0.0227*** 0.0255*** 0.0241*** 0.0239***
(0.0060) (0.0045) (0.0046) (0.0057) (0.0059)
College Education -0.0157*** -0.0199*** -0.0207*** -0.0222*** -0.0217***
(0.0034) (0.0030) (0.0026) (0.0024) (0.0024)
Post-Graduate Education -0.0272*** -0.0332*** -0.0314*** -0.0331*** -0.0327***
(0.0031) (0.0033) (0.0027) (0.0034) (0.0036)
N 43,890 43,890 25,983 19,752 19,752
Not that Bad to Cheat on Taxes Elementary Education -0.0032 -0.0023 0.0082 0.0022 -0.0051
(0.0102) (0.0092) (0.0091) (0.0148) (0.0170)
High School Education 0.0177*** 0.0097 0.0308*** 0.0480*** 0.0446***
(0.0054) (0.0056) (0.0084) (0.0110) (0.0104)
College Education -0.0189** -0.0191** -0.0263*** -0.0311*** -0.0291***
(0.0063) (0.0066) (0.0056) (0.0081) (0.0085)
Post-Graduate Education 0.0059 0.0102 0.0135 0.0268* 0.0232
(0.0103) (0.0110) (0.0080) (0.0126) (0.0133)
N 2,460 2,460 1,545 793 793
Not that Bad to Cheat Government Elementary Education 0.0053 0.0081 0.0121* 0.0224 0.0169
(0.0113) (0.0125) (0.0056) (0.0173) (0.0161)
High School Education -0.0050 -0.0087 -0.0032 -0.0025 -0.0029
(0.0052) (0.0053) (0.0062) (0.0113) (0.0117)
College Education -0.0107*** -0.0096*** -0.0107*** -0.0102** -0.0101**
(0.0024) (0.0018) (0.0021) (0.0035) (0.0036)
Post-Graduate Education 0.0005 0.0029 -0.0034 -0.0066 -0.0072
(0.0042) (0.0044) (0.0026) (0.0067) (0.0066)
N 2,508 2,508 1,575 800 800
Support Country if Country was Wrong Elementary Education -0.0267* -0.0241 -0.0243 -0.0312 -0.0325*
(0.0143) (0.0160) (0.0243) (0.0173) (0.0165)
High School Education -0.0188* -0.0105 -0.0199 -0.0275 -0.0227
(0.0096) (0.0093) (0.0171) (0.0220) (0.0218)
College Education -0.0386*** -0.0365*** -0.0384*** -0.0457*** -0.0451**
(0.0081) (0.0085) (0.0105) (0.0127) (0.0139)
Post-Graduate Education -0.0354** -0.0410*** -0.0393*** -0.0444** -0.0440**
(0.0132) (0.0116) (0.0106) (0.0165) (0.0150)
N 1,980 1,980 1,365 590 590
See notes to Table 5a.
52
TABLE 5F: ESTIMATES OF EDUCATION’S IMPACT AT VARIOUS LEVELS ON MORAL ATTITUDES TOWARDS INEQUALITY
Bivariate
Regression Add Bio/Demo
Controls Add Parent
Controls Add Childhood
Controls Add Fixed
Effects
(1) (2) (3) (4) (5)
Government Should Reduce Inequality
Elementary Education -0.0069 -0.0013 0.0106 0.0164 0.0162
(0.0063) (0.0068) (0.0116) (0.0140) (0.0144)
High School Education -0.0302*** -0.0382*** -0.0363*** -0.0345*** -0.0356***
(0.0052) (0.0062) (0.0083) (0.0086) (0.0090)
College Education -0.0423*** -0.0396*** -0.0350*** -0.0344*** -0.0346***
(0.0023) (0.0022) (0.0031) (0.0029) (0.0031)
Post-Graduate Education 0.0208*** 0.0284*** 0.0329*** 0.0333*** 0.0326***
(0.0043) (0.0045) (0.0033) (0.0047) (0.0048)
N 21,521 21,521 14,203 10,829 10,829
Government Should Help Poor Elementary Education -0.0037 0.0034 0.0082 0.0115 0.0118
(0.0086) (0.0081) (0.0129) (0.0116) (0.0123)
High School Education -0.0411*** -0.0544*** -0.0415*** -0.0425*** -0.0426***
(0.0052) (0.0053) (0.0059) (0.0048) (0.0045)
College Education -0.0359*** -0.0298*** -0.0264*** -0.0235*** -0.0230***
(0.0053) (0.0038) (0.0043) (0.0053) (0.0053)
Post-Graduate Education 0.0198*** 0.0308*** 0.0317*** 0.0276*** 0.0270**
(0.0057) (0.0046) (0.0052) (0.0079) (0.0089)
N 14,386 14,386 8,780 6,461 6,461
Inequality in US is Too Large Elementary Education 0.0013 0.0048 0.0117 0.0452 0.0439
(0.0063) (0.0075) (0.0130) (0.0316) (0.0314)
High School Education 0.0100 0.0110 0.0116 -0.0030 -0.0076
(0.0095) (0.0093) (0.0115) (0.0099) (0.0108)
College Education -0.0175** -0.0169** -0.0085* -0.0053 -0.0037
(0.0062) (0.0064) (0.0045) (0.0092) (0.0094)
Post-Graduate Education 0.0079 0.0086 0.0054 0.0317 0.0330
(0.0101) (0.0096) (0.0102) (0.0184) (0.0187)
N 4,074 4,074 2,599 1,286 1,286
Just that Rich get Better Health Care++ Elementary Education -0.0206 -0.0055 -0.0314 -0.0234 -0.0105
(0.0387) (0.0405) (0.0916) (0.0906) (0.0899)
High School Education -0.0151 -0.0100 -0.0041 -0.0033 -0.0078
(0.0230) (0.0229) (0.0393) (0.0390) (0.0412)
College Education -0.0095 -0.0074 0.0026 0.0032 0.0022
(0.0106) (0.0114) (0.0169) (0.0158) (0.0164)
Post-Graduate Education -0.0080 -0.0181 -0.0099 -0.0094 -0.0158
(0.0256) (0.0269) (0.0386) (0.0403) (0.0390)
N 914 914 574 574 574
Just that Rich Kids get Better Education++ Elementary Education 0.0215 0.0360 -0.0053 -0.0076 -0.0014
(0.0320) (0.0381) (0.0971) (0.0971) (0.0938)
High School Education -0.0216 -0.0170 0.0061 0.0048 0.0105
(0.0210) (0.0204) (0.0341) (0.0378) (0.0374)
College Education -0.0101 -0.0093 0.0063 0.0066 0.0070
(0.0145) (0.0157) (0.0178) (0.0181) (0.0190)
Post-Graduate Education 0.0131 0.0024 0.0027 0.0022 -0.0016
(0.0247) (0.0245) (0.0360) (0.0373) (0.0384)
N 945 945 583 583 583
See notes to Table 5a.
53
Reference List Ahmadi, Ali A. and Azita Ashrafjahani (2011) “Factors Affecting Ethical Perceptions and Attitudes of Managers,” African Journal of Business Management 5(26), 10452-10461 Card, David (1999) “The Causal Effect of Education on Earnings,” Handbook of Labor Economics 3(30), 1801-1863 Dee, Thomas S. (2004) “Are There Civic Returns to Education?” Journal of Public Economics, 88(9-10), 1697–1720 Hodgkinson, Virginia A. and Weitzman, Murray S. (1988) Giving and Volunteering in the United States: Findings from a National Survey (Independent Press, 1988 ed.) Haveman, Robert H. and Barbara L. Wolfe (1993) “Children’s Prospects and Children’s Policy,” Journal of Economic Perspectives 7(4), 153-174 Haveman, Robert H. and Barbara L. Wolfe (2002) “Social and Nonmarket Benefits from Education in an Advanced Economy,” Federal Reserve Bank of Boston Conference Series 47, 97-131 Hersh, Richard H., Diana P. Paolitto, and Joseph Reimer (1990) Promoting Moral Growth: From Piaget to Kohlberg (Waveland Press Inc., 2d ed.) Hjalmarsson, Randi, Helena Holmlund, and Matthew J. Lindquist (2011) “The Effect of Education on Criminal Convictions and Incarceration: Causal Evidence from Micro-Data” CEPR Discussion Paper No. DP8646, available at Social Science Research Network: http://ssrn.com/abstract=1964136 Lochner, Lance and Enrico Moretti (2004) “The Effect of Education on Crime: Evidence from Prison Inmates, Arrests and Self-Reports,” American Economic Review 94(1), 155-189. Machin, Stephen, Olivier Marie, and Sunčica Vujić (2011) “The Crime Reducing Effect of Education,” Economic Journal 121, 463-484 McMahon, Walter W. (1999) Education and Development: Measuring the Social Benefits (Oxford University Press), 92-124, 141-151 Milligan, Kevin, Enrico Moretti, and Philip Oreopoulos (2004) “Does Education Improve Citizenship? Evidence from the U.S. and the U.K.” Journal of Public Economics 88(9-10), 1667–1695
54
Moretti, Enrico (2004) “Estimating the Social Return to Higher Education: Evidence from Longitudinal and Repeated Cross-Sectional Data,” Journal of Econometrics 121, 175-212 Oreopoulos, Philip and Kjell G. Salvanes (2011) “Priceless: The Nonpecuniary Benefits of Schooling,” Journal of Economic Perspectives 25(1), 159-184