All You Need Is a Dollar and a Dream.

Explaining the Inverse Relationship

Between Socio-economic Position and

Lottery Play

JENS BECKERT

MARK LUTTER

Max Planck Institute for the Study of Societies

Paulstr. 3, 50676 Cologne, Germany

beckert@mpifg.de

lutter@mpifg.de

Paper presented at the conference

“Random Riches. Gambling and Speculation in Perspective”

Vienna, May 23-24, 2013

1

Sociologists generally agree that there is an inverse relationship between socioeconomic

position and lottery play.1 Low-income individuals spend a larger part of their incomes on

lottery tickets than those with higher incomes (Beckert and Lutter 2009; Clotfelter and Cook

1991; Livernois 1987), and they benefit less from lottery proceeds (Beckert and Lutter 2008;

Borg et al. 1991; Borg and Mason 1988; Rubenstein and Scafidi 2002). However, the question

remains open as to what factors explain this pattern. Why do the poor spend proportionally

more of their income on lottery tickets than their wealthier and better educated peers?

Much of the research on motivating factors in lottery participation stems from

cognitive psychology (e.g. Ariyabuddhiphongs 2011; Griffiths 1990; Rogers 1998; Rogers

and Webley 2001) and economics (e.g. Albers and Hübl 1997; Brenner and Brenner 1990;

Friedman and Savage 1948; McCaffery 1994; Ng 1965). In sociology, research provides

insightful historical or qualitative in-depth studies (Casey 2008; Garvía 2007; Garvía 2008;

Light 1977; Reith 1999; Rosecrance 1986; Zola 1967) on gambling motivations, but

quantitative accounts that test theories explaining lottery play are almost nonexistent. In this

study, we use survey data to test structural, cultural, and network approaches to explain why

the poor play the lottery. First, deprivation or strain theory explains lottery gambling through

structural factors of disadvantaged social position, and assumes that lottery participation

serves as compensation for and release of tensions arising from social inequalities and

feelings of deprivation (Bloch 1951; Devereux 1980; Frey 1984). Second, cultural

explanations view lottery participation as an expression of prevailing pro-gambling value

orientations such as superstitious beliefs; as a reliance on fate and luck for upward mobility;

and as a denial of the protestant work ethic (Binde 2007; Ellison and Nybroten 1999; Lutter

2011; Murell 1979; Thorner 1956). Third, network-analytical explanations emphasize the

1 This paper is an early version of Beckert and Lutter (2013). It presents previously unpublished additional findings and results.

2

influence of social network structures on the demand for lottery tickets (Adams 1996; Garvía

2007; Guillén, Garvía, and Santana 2012).

Using survey data from a nationwide probability sample, we test these three

approaches, and perform controls for alternative explanations (cognitive bias theory). We find

that all three accounts play a role in explaining the stratified patterns of lottery expenditure.

Factors such as educational attainment and self-perceived social deprivation both have major

effects, whereas cultural factors play a much lesser role. While lottery players show fatalistic

value orientations, it is not a lack of a Protestant work ethic that makes the poor spend

relatively more on lottery tickets. Among the strongest predictors are network factors. The

gambling behavior of an actor’s close social network clearly corresponds to intensified

spending patterns, measured both in terms of monthly lottery expenditure as well as

expenditure as a proportion of income. Interaction effects with income level show that all

factors are able to explain lottery expenditure among the poor. In particular, social

deprivation, peer group influences and fatalistic beliefs are relevant to explaining lottery

consumption among lower income households.

The first section briefly presents the main approaches to explaining gambling

behavior. We then develop arguments to explain the popularity of lotteries among the poor,

and discuss sociological accounts on lottery gambling. We derive several testable hypotheses

from this section. The article’s empirical section describes the data, variable definitions, and

the analytic strategy to test the hypotheses, followed by a presentation and discussion of the

results. The article concludes with a summary of its main findings and reflections on its

implications.

SOCIAL STRATIFICATION AND THE DEMAND FOR LOTTERY TICKETS

Why are members of the lower classes more likely to play the lottery? An answer to this

question must be based on those factors that motivate people to spend money on lotteries. It 3

must also show that people of a lower social status face situations which result in lottery

gambling having greater attractivity to them than to people of a higher social status.

There are two general approaches to explaining socially stratified demand in lottery markets,

individualistic and contextual theories. Individualistic theories reconstruct lottery participation

either as a rational investment decision or as a cognitively biased decision arrived at by an

incorrect understanding of the statistics of the game. Economists view lottery participation as

a rational wealth maximizing strategy for the middle and lower-middle classes (Friedman and

Savage 1948). Once basic needs are fulfilled and the opportunity costs for lottery tickets are

low, playing the lottery can be reconstructed as a rational choice for those who otherwise lack

the means to accumulate substantial wealth (McCaffery 1994). In contrast, cognitive

psychologists and behavioral economists view lottery participation as a function of incorrect

probability assessments, cognitive biases and heuristics (Griffiths 1990). Lottery players

systematically overestimate the very low probability of the game and have overly optimistic

assessments of the chance to win. It is assumed that biased knowledge on the statistical

properties of games of chance and limited cognitive abilities are more prevalent among lower

social classes.

From the contextual viewpoint, purchasing decisions are socially anchored. They are

either structurally explained, arising from tensions and frustrations caused by disadvantaged

socio-economic positions, or culturally explained from beliefs in fate, luck and magic, or

perceived as a result of social contagion through peer group influences. All three contextual

theories maintain that actors of lower social position are at a higher risk of exposure to factors

that bring about lottery gambling. In the following part, we discuss these approaches and

derive testable hypotheses.

Structural factors: Lottery playing as tension management

4

The first approach to explaining socially stratified demand in lottery markets assumes factors

such as low levels of education and disadvantaged socio-economic positions are associated

with higher states of tension, leading to compensatory patterns of behavior. Gambling is such

a pattern. According to functionalist and deprivation theories (Bloch 1951; Devereux 1980;

Frey 1984), gambling is a socially accepted way of channeling frustrations and tensions

resulting from contradictory or unattainable demands imposed on the individual in modern

societies. Gambling, especially for the lower and lower middle classes, is a social practice in

which these tensions can be released without having a disruptive impact on the social order.

Lotteries provide a form of catharsis, giving an outlet for feelings of disadvantage and

dissatisfaction about one’s own status.

Lottery players “indulge in fantasies about what could be done with the prize money”

(Clotfelter and Cook 1991: 9). This makes lottery tickets not a monetary investment but rather

a trigger for daydreams, a vehicle for the momentary escape from reality (Binde 2012; Lutter

2010). Lottery players are motivated by the imaginative value of this consumer good (Beckert

2011; Beckert 2013; Campbell 1987; Lutter 2012). Fantasy worlds stemming from the

purchase of lottery tickets are comparatively cheap. Lower social strata are excluded from

most other “evocative” consumer goods that also create dream worlds, for example status

goods such as fine clothing, wines or luxury cars. Because of this relative exclusion from

alternative opportunities for imaginative goods consumption, individuals from lower social

strata are more likely to be drawn to lottery tickets than members of higher social strata

(Cohen 2001: 730ff.). The likelihood of players projecting unattainable desires onto lottery

tickets is higher for those from lower social status groups.

Two hypotheses follow from this. First, we assume that low socio-economic status

leads to higher personal states of tension and therefore more frequent lottery participation.

Education provides the main resource for material success in modern societies. Aspirations

for upward mobility are highly constrained by lower levels of education. At the same time, 5

lottery participation gives hope for social elevation that is independent of education. A second

hypothesis takes subjective indicators of deprivation into account. Gambling is a method of

sensation seeking to compensate for feelings of boredom and monotony in daily life.

Therefore, people who experience monotony and meaninglessness in their personal and

working lives devote higher expenditure to playing the lottery.

H 1.1: People with low levels of education spend more money on lottery tickets.

H1.2: People who feel their work and everyday lives to be boring and monotonous

spend more money on lottery tickets than those experiencing interesting and

eventful lives.

Cultural factors: Lottery as an expression of beliefs

A second approach to explaining socially stratified demand in lottery markets sees the lower

strata’s increased fascination with lotteries as a function of a culture that both emphasizes

beliefs in fate and luck instead of personal achievement, and lacks orientation towards the

values of a Protestant work ethic (Binde 2007; Ellison and Nybroten 1999; Murell 1979;

Thorner 1956). The Protestant work ethic proclaims virtues of diligence, thriftiness, efficiency

and profitability as well as productivity. From such a perspective, gambling is a waste of time

and money. Betting on random outcomes undermines the Protestant values of self-discipline,

prudence and sober rationality. The reliance on chance moves gambling close to fatalism,

superstition and magic.

The lottery is a game of pure chance in which individual skill, intelligence or habitual

difference have no influence on the opportunities for winning. It is not personal achievement,

talent, creativity or social and cultural capital that determines success in gambling. At the

same time randomness, as a distributive mechanism, stands in opposition to the achievement-

based justification of economic success in modern societies. The fact that the chances of 6

winning are completely equal is a central attraction of lotteries (McCaffery 1994: 88). This

attraction, however, varies with social position: Members of the lower social strata find the

egalitarian distribution of chance more attractive because their opportunities for success are

then greater than when meritocratic or ascriptive mechanisms prevail.

Participation in the lottery can be regarded as a flight from the all-embracing merit

principle of modern societies: “The element of chance also provides an escape from the

rationality of the culture, since gambling permits one to rely on fate and superstition” (Murell

1979: 92). This withdrawal from a Protestant work ethic is especially attractive to people from

lower social classes. While Protestant values of merit and achievement oppose the temptation

to play the lottery, a belief in fate and luck promotes playing games of chance. Two

hypotheses follow from this:

H2.1: People who deny the Protestant work ethic spend significantly more money on

the lottery.

H 2.2: People who express beliefs in fate and luck as a basis for personal success

spend more money on the lottery.

Network Factors: Lottery as a peer influence

A further explanation for the socially stratified demand in lottery markets focuses on the

causal effect of social network position. In the literature on gambling, the network perspective

has been explored in qualitative studies mainly on other games of chance and rarely in studies

on the lottery. Exceptions include the works of Light (1977) and Adams (1996) and especially

Roberto Garvía (2007; 2008). According to Garvía (2007), the exceptionally high demand for

lottery tickets in Spain can be traced back to the widespread tradition of sharing lottery tickets

by playing in syndicates among friends, relatives or colleagues. In a recent study, Guillén, 7

Garvía, and Santana (2012) analyze the motives for playing in syndicates. The authors show

that syndicate players share tickets not primarily out of economic reasons of maximizing their

chances, but mostly to establish cohesive social groups and to maintain friendships (see also

Humphreys and Perez 2013).

Therefore, we assume that syndicates positively affect demand by making lottery

participation part of social interaction and group life. Shared tickets and the activity around

playing the lottery constitute a group identity. As a group activity, the utility of a shared

lottery ticket is not defined primarily by the expected monetary return from a ticket—although

in the minds of the players winning remains a possibility—but by the secondary social effects

which evolve from membership in the informal group. The lottery ticket generates shared

experiences and emotions and forms a basis for communication. Affirmative attitudes toward

gambling are promoted in the group. Losses are less likely to stop future participation, since

quitting would threaten the continuity of the group or one’s own group membership. In

addition to the effects from these voluntary syndicates, we can also assume effects from social

contagion from a players’ social surroundings. The gambling habits of those in social

networks close to an individual should influence his or her own gambling behavior.

H 3.1: Playing in a syndicate positively correlates with intensified demand for lottery

tickets.

H 3.2: The more members of a respondent’s close network (friends, partner and

parents) play the lottery, the higher the probability that the respondent plays

the lottery more intensively.

DATA AND METHODS

8

The data used to test the hypotheses come from a CATI survey on the demand structure of

Lotto players in Germany (for details on the sample, see Beckert and Lutter 2013). Table 1

presents basic descriptive statistics of all variables used in this study. Our central dependent

variable is the amount of monthly expenditure for lottery tickets. We take the natural log of

this variable to account for skewness and to improve model fit. In addition, we construct this

variable as a percentage share of monthly income. In doing this, we capture lottery playing as

a measure of wasteful consumption.

We use the following set of indicators as independent variables. Following H1.1,

educational level is measured by the number of years spent in school and higher learning

institutions. H1.2 indicates an association between lottery gambling and work or life

dissatisfaction. We measure work/life dissatisfaction with a simple additive index of three

seven-point Likert items. The items are (a) “I often feel that I am doing something meaningful

in my work (in my everyday life)”, (b) “My work (everyday life) often bores me” and (c) “I

think my work (everyday life) is much more interesting than those of others.”2

Hypothesis H2.1 proposes a negative relation between gambling habits and the

Protestant work ethic. We use a simple additive index of the following seven-point Likert

items: (a) “If one works hard enough one is likely to make a good life for oneself.”, (b)

“There are few satisfactions equal to the realization that one has done his best at a job”, (c)

“The self-made person is likely to be more ethical than the person born to wealth.”3 As for

hypothesis H2.2, which assumes an association between the intensity of play and fatalistic

value orientations, we use a scale that consists of an additive index of the following three

seven-point Likert items: (a) “In order to get along in life, it is good to have luck on your

side”, (b) “Most people are not aware of how much of their life depends on chance and luck”

2 The items are taken from a scale by Firestone et al. (2005). 3 Items are taken from a scale by Mirels and Garrett (1971).

9

and (c) “Life is mostly controlled by fate.”4 All items were recoded in such a way that higher

values represent greater beliefs.

With regard to social network effects, we use two indicators. First, we use a dummy

variable that counts the number of respondents playing in a syndicate with friends, colleagues

or acquaintances (1 = player usually plays in a syndicate; 0 = player usually plays alone). In

order to assess the influence of the players’ social networks on the demand for lottery tickets

(H3.2), we generate ego-centric network data from our survey data. Respondents were asked

to consider two very close friends and how often each had played the lottery, if at all, within

the past year (coded as 0 = never, 8 = a few times a year, 18 = once or twice a month, 52 = at

least once a week). We also asked how often the respondent’s life partner and parents played

the lottery. We then matched the data of the four variables into a single additive index which

measures the intensity of lottery play in the respondent’s close social network.

In all statistical models, we use the following set of control variables: a dummy for

gender (1 = female), monthly available net income (transformed into logged values), age

(logged years), level of employment (where 1 = full time, 0 = otherwise), status of

employment (1 = blue-collar worker, 0 = otherwise), unemployed (1 = yes, 0 = no), retired (1

= yes, 0 = no), cohabitation (1 = living together with a partner, 0 = otherwise), urbanity

(1=living in region with more than 200.000 inhabitants, 0 = otherwise), Catholic (1 = yes, 0 =

no), Protestant (1 = yes, 0 = no), minority (1 = citizenship other than German, 0 = otherwise).

To control for possible cognitive biases, as suggested, we use a variable that measures

biased or overly optimistic assessments of the game’s winning probabilities. Survey

respondents were asked if they believe that winning the jackpot is either “likely” (coded as 1),

“unlikely” or “very unlikely” (both coded as 0). We assume that players who think they are

“likely” to win the lottery have an overly optimistic perception of the true winning chances of

the game.

4 The items are taken from a scale used by Pearlin (1981) and Darke and Freedman (1997). 10

INSERT TABLE 1 ABOUT HERE

RESULTS

Table 2 presents the results of our regressions on the log of the amount of lottery expenditure.

The first model estimates the socioeconomic determinants of lottery play. A number of

statistically significant differences can be observed, revealing insights into the

sociodemographic composition of lottery players. Income, age, employment status, and

cohabitation are significant factors. Monthly income varies positively with lottery

expenditure, but with a disproportionate growth rate. Because of the double log-linear model

specification (Gujarati 2003: 175f.), we can interpret the coefficient as an estimator of income

elasticity. The result reveals that for a 10 percent increase in monthly income, we can expect

an increase of roughly 3 percent in expenditure. This means that low income households

spend a relatively higher proportion of their income on lottery tickets than higher income

households.

Age is another nonlinear predictor of lottery play. Lottery players tend to be older but in a

disproportionate relation to spending. We can expect an approximate 8 percent increase in

expenditure when age increases by 10 percent. Furthermore, lottery players are significantly

more likely to work full-time and tend to cohabitate.

In line with our first hypothesis, education proves to be a significant factor in

explaining the demand for lottery tickets. Since the dependent variable was transformed into

logged values, the coefficients can be interpreted in terms of percentage change. With every

additional year of education, the average expenditure drops by roughly six percent. People

with low educational levels spend significantly more on lottery tickets than the better

educated.

11

Hypothesis H1.2 concerns the subjective assessment of strain arising from socially

disadvantaged positions, and proposes an association between lottery gambling and work or

life dissatisfaction. As the model results show, dissatisfaction with one’s work is significantly

related to lottery expenditure. Those who lack autonomy and express feelings of senselessness

in their everyday routines spend significantly more money on lottery tickets. These findings

underline the assumption that lottery gambling functions as an outlet for strain by offering an

easy dream of escape from daily drudgery.

The cultural factors proposed by hypotheses H2.1 and 2.2 cannot sufficiently explain

lottery expenditure. While fatalistic value orientations have a clear impact on expenditure,

spending patterns do not correspond to work-ethic value orientations. Although lottery players

maintain a culture of beliefs in fate and luck, they do not play because they lack or deny a

Protestant work ethic.

By far the strongest predictors of lottery expenditure are the network-related factors,

peer play and syndicate play. This is shown by the substantial increase in model fit when

including network variables in Model 4. Adjusted R-square doubles from .117 to .232. Both

coefficients are highly significant. The effect of education diminishes to some extent when

these predictors are included. Therefore, controlling for social network variables affects the

social composition of lottery playing with regard to educational attainment.

INSERT TABLE 3 ABOUT HERE

Table 3 displays the results of OLS regressions on expenditure as a proportion of income.

Measuring the variable in this way allows us to examine the degree of “wastefulness” of this

form of demand behavior. Findings are in line with the proposed hypotheses. Spending a high

proportion of income on lottery tickets is primarily a function of disadvantaged social

positions, self-perceived work and life dissatisfaction, beliefs in fate and luck and network 12

influences from close peers. The Protestant work ethic as a cultural explanation again shows

no significant effect. Furthermore, the finding that class differences are moderated mostly by

network variables emerges more clearly in these models. Adding network variables in the

fourth model diminishes the negative effect of education, which becomes insignificant.

Overall, one can speak of a curvilinear relation of lottery expenditure and

socioeconomic position. Lottery players are most likely to come not from the poorest, but

from upper segments of the lower classes. It is not the unemployed or those in insecure work

that are most likely to spend money on lottery tickets. It is rather middle aged full-time

employees with lower educational backgrounds but with mid-range incomes. In line with Tec

(1964), this can be interpreted in line with the deprivation approach to gambling (also see

Beckert and Lutter 2007). Those in the higher segments of the lower classes are to a greater

extent than others situated in a tension-filled status arrangement: Compared to the poorest

layer of social strata, they are integrated more deeply into the meritocratic value system of a

capitalist society and have greater personal contact to people of higher statuses. This exposes

them to social comparison. They perceive the success of others and develop aspirations for

upward mobility which remain unfulfilled because they lack educational credentials.

Participation in the lottery gives hope for social elevation independently of society’s

constraints.

INSERT TABLE 3 ABOUT HERE

We now analyze how the effects of predictors alter over different levels of income. For

this we estimate regressions with interaction effects for income and all substantial predictors

used in this study.5 Figure 1 shows the conditional effects of the variables on the absolute

expenditure. Figure 2 shows the conditional effects of the variables on relative expenditure

5 Detailed regression results can be made available upon request. 13

measured as a share of income. In these figures, each panel plots the estimated regression

coefficients of the respective predictor for the whole range of values of the income variable.

This approach reveals how effects change with income, both in their direction and in their

statistical significance. The outer lines indicate the lower and upper limits of the 95-percent

confidence interval of the respective coefficient at a specific value of income. If the interval is

beyond the zero line, then the coefficient is significant at the 95-percent level. If the zero line

intersects the confidence interval, the coefficients are insignificant.

INSERT FIGURE 1 ABOUT HERE

Firstly, the outcomes reveal interesting results with regard to strain/deprivation theory.

In Figure 1, the negative effect of education and the influence of work and life dissatisfaction

are particularly strong at higher income levels. Subjective feelings of work/life dissatisfaction

only moderately increase the (already quite high) level of lottery expenditure among the lower

income classes but lead to drastic changes among the higher classes. People from higher

income levels who feel socially disadvantaged probably experience more acute subjective

feelings of status inconsistency. This produces a higher level of strain and tension which in

turn results in more intensive gaming behavior. The same pattern can be observed in the

negative education effect. When income levels increase, lower levels of education play a

stronger role in determining lottery expenditure. Therefore, frustration or status discrepancy

has a greater effect on playing behavior and leads to higher amounts of expenditure among

those classes who are able to afford it. This result vividly portrays strain theory: Lottery

playing seems to serve as an outlet for the strain that people experience from status

inconsistencies.

On the other hand, peer play imposes increased lottery involvement on people from

lower classes. While the composition of organized syndicates only moderately varies with 14

income, peer influence is a strong explanatory factor among lower income households.

Network factors seem to have a strong clustering effect with regard to class.

Figure 2 displays interaction effects of regressions on expenditure as a percentage of

monthly income. Here, contrary to absolute amounts of expenditure, all predictors except for

the work-ethic factor reveal their strongest effects among the lower income classes. Again,

peer play proves to be the most significant factor. Socioeconomic position has an effect on the

constitution of an actor’s network, and the characteristics of the network affect the individual

behavior of its members. A lower position in the social strata of a society clearly bonds

networks of lottery players, which in turn affects the playing behavior of the respondent.

Therefore, network-ties play an important role in the formation of this market by creating and

maintaining demand, and moreover, in affecting socially skewed patterns of wasteful

consumption through lottery play.

INSERT FIGURE 2 ABOUT HERE

CONCLUSION

Each of the discussed approaches partially contributes to an explanation of the demand for

lottery tickets. We were able to show that the explanatory mechanisms identified by the

theories are related to income positions. The explanatory relevance of each approach does,

nevertheless, differ as does their relationship to income status.

Structural factors of educational disadvantage and self-perceived deprivation prove to

be relevant in explaining lottery expenditure. People who feel alienated at work spend more

on the lottery than those who are satisfied with their everyday life. Moreover, the conditional

relationship between levels of income and gambling expenditure also corroborates the theory.

Interestingly, feelings of dissatisfaction have greater effects on absolute lottery expenditure 15

from higher income classes. This is probably due to status inconsistencies being felt more

intensely. In contrast, dissatisfaction is highly relevant among lower income classes when

expenditure is measured as a share of income.

Cultural factors, on the other hand, play only a minor role in explaining why the poor

play the lottery. While beliefs in fate and luck can explain spending patterns both for lower

and higher income classes, the Protestant work ethic is a relevant cultural factor only for

higher income classes in deterring gambling. The lack or denial of this ethic is not a

characteristic of the poor. Therefore, a central finding of this study is that a lack of the “right”

values is not a reason why the poor play the lottery.

Compared with all other factors, the network approach makes the best fit to the data.

Network variables can explain approximately half of the variance and are therefore the

strongest predictors of lottery expenditure. Expenditure is significantly increased when there

is an increase in the number and frequency of close peers playing the lottery, with the effect

being most pronounced among lower income individuals. Hence, the poor play more because

network factors impart influences on spending patterns, both measured absolutely and relative

to income.

The intention of this article was to provide a more comprehensive examination for the causes

of socially stratified demand behavior in lottery markets. The results show that in addition to

networks, factors of social position and deprivation as well as beliefs on fate and luck are

relevant. The relationship between the intensity of lottery play and objective criteria such as

sociodemographic status has been known from previous research (Brown, Kaldenberg, and

Browne 1992; Clotfelter and Cook 1991). However, the associations between lottery play and

subjective indicators of status frustration and their respective mediation effects have not been

detected in this form before. Some of the findings from previous empirical studies by Tec

(1964), Downes et al. (1976), Smith et al. (1976) and King (1985) do point in a similar

16

direction, but their results are either statistically insignificant or refer to other forms of

gambling.

Explaining spending patterns for lottery tickets among different income groups

provides insights into the social stratification of market demand. Our findings point to the

importance of social context – in contrast to individual player characteristics as stressed in

economic and psychological approaches – in explaining demand patterns within the lottery

market, and hence reveal the social influences that explain the stratification of certain patterns

of consumption.

17

TABLES

TABLE 1: Descriptive statistics

of all dependent and independent variables

Variable Mean Std. Dev. Min. Max.

Expenditure (euros) 10.66 20.33 0 216 Expenditure (ln) 1.49 1.38 0 5.39 Expenditure (% of income) 0.83 2.25 0 35.29 Female .48 .50 0 1 Income (ln) 6.96 .59 4.42 9.43 Age (ln) 3.77 .37 2.89 4.47 Employment full-time .46 .50 0 1 Blue-collar worker .24 .43 0 1 Unemployed .05 .21 0 1 Retired .21 .41 0 1 Cohabitation .63 .48 0 1 Urbanity .25 .43 0 1 Catholic .32 .47 0 1 Protestant .38 .49 0 1 Minority .02 .14 0 1 Education 12.10 3.23 8 17 Work dissatisfaction 6.97 3.42 3 21 Work ethic 7.19 2.95 3 21 Belief in fate & luck 12.86 4.03 3 21 Syndicate play .17 .37 0 1 Peer play 37.66 39.97 0 208 Optimism .05 .22 0 1

18

TABLE 2: Results of OLS regressions

on log of lottery expenditure

(1) (2) (3) (4) (5) Expen-

ditures, ln Expen-

ditures, ln Expen-

ditures, ln Expen-

ditures, ln Expen-

ditures, ln Female 0.084 0.069 0.085 0.129 0.110 (0.976) (0.799) (0.986) (1.597) (1.372) Income (ln) 0.304*** 0.311*** 0.360*** 0.315*** 0.320*** (3.589) (3.664) (4.246) (3.979) (4.091) Age (ln) 0.787*** 0.809*** 0.692*** 0.661*** 0.656*** (4.933) (5.081) (4.311) (4.413) (4.427) Employment full-time 0.234* 0.253* 0.213* 0.141 0.136 (2.137) (2.318) (1.971) (1.394) (1.358) Blue-collar worker 0.170 0.140 0.083 0.052 0.047 (1.711) (1.391) (0.818) (0.554) (0.503) Unemployed 0.081 0.010 -0.086 0.005 0.017 (0.387) (0.047) (-0.403) (0.026) (0.086) Retired 0.030 0.008 -0.021 0.083 0.080 (0.189) (0.052) (-0.133) (0.561) (0.541) Cohabitation 0.215* 0.216* 0.252** 0.112 0.108 (2.452) (2.454) (2.865) (1.349) (1.311) Urbanity -0.142 -0.132 -0.091 -0.091 -0.093 (-1.515) (-1.407) (-0.972) (-1.046) (-1.072) Catholic -0.133 -0.104 -0.148 -0.154 -0.168 (-1.266) (-0.992) (-1.417) (-1.578) (-1.741) Protestant 0.017 0.031 0.006 -0.045 -0.053 (0.174) (0.315) (0.059) (-0.495) (-0.586) Minority 0.393 0.341 0.260 0.362 0.305 (1.288) (1.121) (0.827) (1.236) (1.052) Education -0.060*** -0.060*** -0.053*** -0.033* -0.029* (-4.199) (-4.214) (-3.727) (-2.438) (-2.163) Work dissatisfaction 0.028* 0.032* 0.027* 0.026* (2.255) (2.569) (2.331) (2.232) Work ethic -0.027 -0.022 -0.018 (-1.855) (-1.619) (-1.322) Belief in fate & luck 0.046*** 0.035*** 0.031** (4.392) (3.577) (3.191) Syndicate play 0.865*** 0.829*** (8.230) (7.955) Peer play 0.008*** 0.008*** (8.094) (8.069) Optimism 0.852*** (4.883) Constant -3.106*** -3.441*** -3.826*** -3.842*** -3.883*** (-4.426) (-4.825) (-5.260) (-5.664) (-5.787) R2 0.096 0.102 0.131 0.245 0.262 Adjusted R2 0.086 0.090 0.117 0.232 0.249 Observations 1090 1080 1043 1043 1043

Notes: t statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001 (two-sided tests).

19

TABLE 3: Results of OLS regressions

on lottery expenditure as a percentage of income

(1) (2) (3) (4) (5) Expen-

ditures, % of income

Expen- ditures, % of

income

Expen- ditures, %

income

Expen- ditures, % of

income

Expen- ditures, % of

income Female 0.004* 0.003* 0.003* 0.003* 0.003* (2.305) (2.010) (2.000) (2.326) (2.122) Income (ln) -0.009*** -0.008*** -0.005*** -0.006*** -0.006*** (-5.536) (-5.131) (-3.841) (-4.252) (-4.229) Age (ln) 0.015*** 0.015*** 0.012*** 0.012*** 0.012*** (5.066) (5.317) (4.413) (4.412) (4.422) Employment full-time 0.001 0.001 -0.000 -0.001 -0.001 (0.453) (0.498) (-0.118) (-0.486) (-0.538) Blue-collar worker 0.001 0.001 0.001 0.001 0.001 (0.362) (0.355) (0.759) (0.545) (0.497) Unemployed -0.003 -0.004 -0.007 -0.006 -0.006 (-0.744) (-0.987) (-1.937) (-1.748) (-1.709) Retired -0.002 -0.003 -0.002 -0.001 -0.001 (-0.557) (-0.962) (-0.706) (-0.286) (-0.313) Cohabitation 0.002 0.001 0.003* 0.002 0.002 (0.968) (0.900) (2.248) (1.243) (1.206) Urbanity -0.003 -0.003 -0.002 -0.002 -0.002 (-1.626) (-1.565) (-1.131) (-1.157) (-1.182) Catholic -0.001 -0.001 -0.000 -0.000 -0.000 (-0.479) (-0.276) (-0.150) (-0.151) (-0.288) Protestant -0.000 -0.001 -0.000 -0.001 -0.001 (-0.239) (-0.363) (-0.290) (-0.620) (-0.704) Minority 0.008 0.007 0.007 0.007 0.006 (1.427) (1.272) (1.252) (1.439) (1.269) Education -0.001* -0.001* -0.001* -0.000 -0.000 (-2.545) (-2.507) (-2.326) (-1.515) (-1.252) Work dissatisfaction 0.000* 0.001* 0.000* 0.000* (2.033) (2.398) (2.168) (2.072) Work ethic -0.000 -0.000 -0.000 (-1.021) (-0.821) (-0.539) Belief in fate & luck 0.000* 0.000 0.000 (2.465) (1.861) (1.490) Syndicate play 0.007*** 0.006*** (3.645) (3.350) Peer play 0.000*** 0.000*** (5.169) (5.112) Optimism 0.014*** (4.515) Constant 0.021 0.012 0.001 0.001 -0.000 (1.630) (0.902) (0.065) (0.050) (-0.006) R2 0.083 0.083 0.094 0.134 0.150 Adjusted R2 0.072 0.071 0.080 0.118 0.135 Observations 1090 1080 1043 1043 1043

Notes: t statistics in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001 (two-sided tests).

20

FIGURES

FIGURE 1: Conditional effects of the predictors on the log of

lottery expenditure at different levels of income,

with 95-% confidence intervals

-.2-.1

0.1

-.10

.1.2

-.20

.2

0.0

5.1

.15

0.5

11.

52

0.0

05.0

1.0

15.0

2

-2 0 2 -2 0 2 -2 0 2

Education Work dissatisfaction Work ethic

Belief in fate & luck Syndicate play Peer play

effect on lottery expenditures95%-confidence interval

Income

21

FIGURE 2: Conditional effects of the predictors on

expenditure as % of income, at different levels of income,

with 95-% confidence intervals

-.4-.2

0.2

-.20

.2.4

-.2-.1

0.1

.2

-.20

.2.4

-20

24

-.02

0.0

2.0

4

-2 0 2 -2 0 2 -2 0 2

Education Work dissatisfaction Work ethic

Belief in fate & luck Syndicate play Peer play

effect on expenditures (as % of income)95%-confidence interval

Income

22

REFERENCES Adams, Douglas James. 1996. Playing the Lottery. Social Action, Social Networks and

Accounts of Motive. Ph.D. Dissertation, Department of Sociology, University of Arizona.

Albers, Norman and Lothar Hübl. 1997. "Gambling Market and Individual Patterns of Gambling in Germany." Journal of Gambling Studies 13:125-144.

Ariyabuddhiphongs, V. 2011. "Lottery Gambling: A Review." Journal of Gambling Studies 27:15-33.

Beckert, Jens. 2011. "The Transcending Power of Goods: Imaginative Value in the Economy." Pp. 106-128 in The Worth of Goods. Valuation and Pricing in the Economy, edited by J. Beckert and P. Aspers. Oxford: Oxford University Press.

Beckert, Jens. 2013. "Imagined Futures: Fictional Expectations in the Economy." Theory and Society 42:219-240.

Beckert, Jens and Mark Lutter. 2007. "Wer spielt, hat schon verloren? Zur Erklärung des Nachfrageverhaltens auf dem Lottomarkt." Kölner Zeitschrift für Soziologie und Sozialpsychologie 59:240-270.

Beckert, Jens and Mark Lutter. 2008. "Wer spielt Lotto? Umverteilungswirkungen und sozialstrukturelle Inzidenz staatlicher Lotteriemärkte." Kölner Zeitschrift für Soziologie und Sozialpsychologie 60:233-264.

Beckert, Jens and Mark Lutter. 2009. "The Inequality of Fair Play. Lottery Gambling and Social Stratification in Germany." European Sociological Review 25:475-488.

Beckert, Jens and Mark Lutter. 2013. "Why the Poor Play the Lottery: Sociological Approaches to Explaining Class-based Lottery Play." Sociology 47:1152-1170.

Binde, Per. 2007. "Gambling and Religion: Histories of Concord and Conflict." Journal of Gambling Issues 20:145-165.

Binde, Per. 2012. "Why people gamble: a model with five motivational dimensions." International Gambling Studies 13:81-97.

Bloch, Herbert A. 1951. "The Sociology of Gambling." The American Journal of Sociology 57:215-221.

Borg, Mary O. and Paul M. Mason. 1988. "The Budgetary Incidence of a Lottery to Support Education." National Tax Journal 41:75-85.

Borg, Mary O., Paul M. Mason, and Stephen L. Shapiro. 1991. The Economic Consequences of State Lotteries. New York: Praeger.

Brenner, Reuven and Gabrielle A. Brenner. 1990. Gambling and Speculation. A Theory, a History, and a Future of Some Human Decisions. New York: Cambridge University Press.

Brown, Daniel J., Dennis O. Kaldenberg, and Beverly A. Browne. 1992. "Socioeconomic-Status and Playing the Lotteries." Sociology and Social Research 76:161-167.

Campbell, Colin. 1987. The Romantic Ethic and the Spirit of Modern Consumerism. Oxford, New York: B. Blackwell.

Casey, Emma. 2008. Women, pleasure and the gambling experience. Aldershot, U.K.: Ashgate.

Clotfelter, Charles T. and Philip J. Cook. 1991. Selling Hope. State Lotteries in America. Cambridge, MA: Harvard University Press.

Cohen, Lloyd R. 2001. "The Lure of the Lottery." Wake Forest Law Review 36:705-745. Darke, Peter R. and Jonathan L. Freedman. 1997. "The Belief in Good Luck Scale." Journal

Of Research In Personality 31:486-511.

23

Devereux, Edward C. Jr. 1980. Gambling and the Social Structure. A Sociological Study of Lotteries and Horse Racing in Contemporary America, vol. I & II. New York: Arno Press.

Downes, David M., B.P. Davies, M.E. David, and P. Stone. 1976. Gambling, Work and Leisure. London: Routledge and Kegan Paul Ltd.

Ellison, Christopher G. and Kathleen A. Nybroten. 1999. "Conservative Protestantism and Opposition to State-sponsored Lotteries: Evidence from the 1997 Texas Poll." Social Science Quarterly 80:356-369.

Firestone, Juanita M., Raymond T. Garza, and Richard J. Harris. 2005. "Protestant Work Ethic and Worker Productivity in a Mexican Brewery." International Sociology 20:27-44.

Frey, James H. 1984. "Gambling: A Sociological Review." Annals of the American Academy of Political and Social Science 474:107-121.

Friedman, Milton and L. J. Savage. 1948. "The Utility Analysis of Choices Involving Risk." The Journal of Political Economy 56:279-304.

Garvía, Roberto. 2007. "Syndication, Institutionalization, and Lottery Play." American Journal of Sociology 113:603-652.

Garvía, Roberto. 2008. Loterías. Un Estudio Desde La Nueva Sociología Económica, vol. 254. Madrid: CIS.

Griffiths, Mark D. 1990. "The Cognitive Psychology of Gambling." Journal of Gambling Studies 6:31-42.

Guillén, Mauro F., Roberto Garvía, and Andrés Santana. 2012. "Embedded Play: Economic and Social Motivations for Sharing Lottery Tickets." European Sociological Review 28:344-354.

Gujarati, Damodar N. 2003. Basic Econometrics. Boston: McGraw Hill. Humphreys, Brad R. and Levi Perez. 2013. "Syndicated Play in Lottery Games." The Journal

of Socio-Economics 45:124-131. King, Kim M. 1985. "Gambling: Three Forms and Three Explanations." Sociological Focus

18:235-248. Light, Ivan. 1977. "Numbers Gambling Among Blacks: A Financial Institution." American

Sociological Review 42:892-904. Livernois, J. R. 1987. "The Redistributive Effects Of Lotteries - Evidence From Canada."

Public Finance Quarterly 15:339-351. Lutter, Mark. 2010. Märkte für Träume. Die Soziologie des Lottospiels. Frankfurt/Main, New

York: Campus. Lutter, Mark. 2011. "Konkurrenten auf dem Markt für Hoffnung: Religiöse Wurzeln der

gesellschaftlichen Problematisierung von Glücksspielen." Soziale Probleme 22:28-55. Lutter, Mark. 2012. "Tagträume und Konsum: Die imaginative Qualität von Gütern am

Beispiel der Nachfrage für Lotterien." Soziale Welt 63:233-251. McCaffery, Edward J. 1994. "Why People Play Lotteries and Why It Matters." Wisconsin

Law Review 71:71-122. Mirels, Herbert L. and James B. Garrett. 1971. "Protestant Ethic as a Personality Variable."

Journal of Consulting and Clinical Psychology 36:40-44. Murell, May E. 1979. "Why People Gamble: A Sociological Perspective." Pp. 84-105 in

Gambling Today, edited by D. Lester. Springfield, Ill.: Thomas. Ng, Yew Kwang 1965. "Why do People Buy Lottery Tickets? Choices Involving Risk and the

Indivisibility of Expenditure." The Journal of Political Economy 73:530-535. . Pearlin, Leonard I., Elizabeth G. Menaghan, Morton A. Lieberman, and Joseph T. Mullan.

1981. "The Stress Process." Journal of Health and Social Behavior 22:337-356. Reith, Gerda. 1999. The Age of Chance. Gambling in Western Culture. London: Routledge.

24

Rogers, Paul. 1998. "The Cognitive Psychology of Lottery Gambling. A Theoretical Review." Journal of Gambling Studies 14:111-134.

Rogers, Paul and Paul Webley. 2001. ""It could be us!": Cognitive and Social Psychological Factors in UK National Lottery Play." Applied Psychology 50:181-199.

Rosecrance, John. 1986. "Why Regular Gamblers Don't Quit: A Sociological Perspective." Sociological Perspectives 29:pp. 357-378. .

Rubenstein, Ross and Benjamin Scafidi. 2002. "Who Pays and Who Benefits? Examining the Distributional Consequences of the Georgia Lottery for Education." National Tax Journal 55:223-238.

Smith, Ronald W., Frederick W. Preston, and Harry L. Humphries. 1976. "Alienation from Work. A Study of Casino Card Dealers." Pp. 229-246 in Gambling and Society, edited by W. R. Eadington. Springfield, Ill.: Thomas.

Tec, Nechama. 1964. Gambling in Sweden. Totowa: Bedminster Press. Thorner, Isidor. 1956. "Ascetic Protestantism, Gambling And The One-Price System."

American Journal Of Economics And Sociology 15:161-172. Zola, Irving Kenneth. 1967. "Observations on Gambling in a Lower-Class Setting." Pp. 19-32

in Gambling, edited by R. D. Herman. New York: Harper & Row.

25