International Research Journal of Social Sciences_____________________________________ ISSN 2319–3565Vol. 2(1), 6-14, January (2013) Int. Res. J. Social Sci. International Science Congress Association 7 Explaining Changing Trust Trends in America April K. Clark, Michael Clark, and Daniel Monzin3 Department of Political Science, Purdue University-Calumet, Indiana, UNITED STATES Department of Political Science, Northern Illinois University, Illinois, UNITED STATES Res. Asst., Dept. of Political Science, Purdue University-Calumet, Indiana, UNITED STATESAvailable online at: www.isca.in Received 24th December 2012, revised 31st December 2012, accepted 5th January 2013 Abstract It is understood that high levels of generalized trust are necessary for a well-functioning democracy. Since the mid-1980s, however, trust in America has declined dramatically and has not returned to the same levels since. What explains this trend? We use two approaches to explain changing attitudes toward trust in others using data from the NORC General Social Surveys. First, we conduct age-period-cohort (A-P-C) analyses showing that generational replacement is having a negative impact on trust levels. That is, more trusting generations of Americans have been dying and being replaced by younger, less trusting Americans. Second, we pool cross-sectional survey data to model declining trust, and we show how trust in others is influenced by individual factors, and how these patterns change over time. Keywords: Interpersonal trust, age, period, cohort, trends, socio-demographic influences. Introduction Trust is an idea which shapes and forms an individual’s perception and understanding of society. Trust and reciprocity are essential for a community to function properly. Individuals who see other members of society as trustworthy are more likely themselves to be trustworthy. Social trust facilitates the passage of information among individuals and encourages cooperation, providing a framework for future interactions. It is clear that trust has an enormous array of practical benefits to individuals and to communities. Trust brings good things politically, civically, and economically. Trusting societies are important because a common culture where individuals believe they share underlying values facilitates compromises on important issues. In short, having a high level of trust in others is imperative for a functioning democracy. It is understood then that societies with high levels of generalized trust and civic attitudes provide incentives for citizens to come together in both social and political spheres. However, as chronicled by others, the American population as a whole has reduced its trust in others and entering cohorts have shown rather steady declines ever since the late-1940s1-. While there is a general consensus that trust in America declined sharply in the mid-1980s and has not returned to the same levels since, the cause of declining trust in America remains unsettled. Some argue that the dramatic nature of the shift in interpersonal trust supports demographic explanations of change, such as aging or cohort replacement hypotheses2-6; while others link the decline to other factors in society2,7-10. Research Methodology Theories of the Origins of Social Trust: The cause and creation of trust and whether it takes place at the societal or individual level is a wide ranging debate within the academic community. There are several contending explanations in the literature. At the societal level trust is believed to be a property of society rather than the individual. Trust is created through interactions with both formal and informal organizations2,11; in the direct participation in the social networks of everyday life12; as well as in the characteristics and attitudes of the local community2,5. As for the influence on trust of the socio-economic environment, previous research finds that income inequality13-14, racial heterogeneity14-15, crime14 and wealth measured by GDP per capita15-16 are all important macro-level determinants. Individual factors such as demographic characteristics and also an individual’s perception of their overall well-being or success in life have previously shown correlations with trust2,9,13,17. The empirical evidence for a strong connection between individual factors is vast. For example, others have shown that at the individual level, education and income are positively linked to trust5,14. Various studies have also found that trust is expressed by the “winners” in society as measured in terms of money, status, and high levels of life satisfaction and subjective happiness2,8-9. In addition, numerous studies have argued that trust changes are due to aging or generational effects. One such study finds that the over-time decline in trust stems chiefly from two separate, but related, effects18. In this way, a cohort effect beginning in the late 1940s explains the reduction in trust among more recent International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 8 cohorts but the findings also support that some of the decline may be the result of a non-linear age pattern with trust levels increasing among adults between the ages of 18-47 and then leveling off. Distrust also tends to be expressed by victims of traumatic or overwhelming events. It has been shown that trust decreases for blacks and females (i.e. groups that have been historically discriminated against)14; it is lower for those who experienced crime or violence in past years; and those who are divorced or separated. Such life experiences, may lead to heightened suspiciousness of one’s surroundings and the motives of others19that has negative consequences for an individual’s development of trust in others. Although some studies show that trust is not influenced by religious preference, there is evidence that religious belief, belonging, and behavior matter for trust. Previous research has shown that individuals who claim a religious affiliation tend to be more trusting than those who do not14. It has also been argued that people who report frequent attendance of a religious service are more inclined to participate in trust-building activities such as giving both time and money, spending time with friends, and participating in volunteer associations. Research shows that individuals who are considered religious have a higher level of trust and civic participation than those who do not10. While none of these different theories are mutually exclusive or incompatible, our paper seeks to explain the trend of declining trust in the United States in two ways. First, we test three possible explanations: the life-cycle, generational-replacement, and period effects theories of changing public opinion in an effort to understand attitudinal movement on this issue. Second, we estimate a set of models that rest on the view that trust is an individual property associated with individual characteristics. In particular it is argued that social and demographic features influence trust and we examine if and how group attitudes have been changing and the effect on trust trends. Data and Measures: The social trust index employed here is based on responses to the three standard forced-choice items employed by the General Social Survey (GSS) for 1972 through 2010. The long duration of the GSS and the use of consistent survey language to measure interpersonal trust make it ideally suited for analyzing trends over time. To determine the cause of decline in interpersonal trust since the mid-1980s the data was split into two pooled datasets, 1972 through 1984 and 1986 through 2010. Running OLS regressions for both time frames demonstrates which variables influence trust for each time period and whether after the decline beginning in 1984 the same variables remained significant. Due to the adoption of a non-respondent, sub-sampling design in 2004, a weight must be employed to compensate for the change in study design. To ensure that these time series are nationally representative, all estimates are weighted (using the GSS weight WTSALL), and we drop the 1987 black oversamples. The three trust measures in the GSS used to indicate generalized trust are: “Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?” (TRUST) (1 = Most people can be trusted, 2 = Cannot be too careful, 3 = Other, depends); ‘‘Would you say that most of the time, people try to be helpful, or that they are mostly just looking out for themselves?’’(HELPFUL) (1 = Try to be helpful, 2 = Just look out for themselves, 3 = Depends); and ‘‘Do you think most people would try to take advantage of you if they got a chance, or would they try to be fair?’’(FAIR) (1 =Would take advantage of you, 2 = Would try to be fair, 3 = Depends). The answers were combined into an index of social trust ranging from 3 to 9, with higher values representing high trust in others. The internal consistency of the interpersonal trust scales ranges from .36 to .43 and a .67 measure of reliability (Cronbach's alpha). The data include 29,779 respondents who had trust scores across all survey years. Individual Characteristics: As indicated previously a number of factors are thought to influence trust. To begin, ag has been shown to have a positive effect on trust2,18,20. Respondents’ ages in the data pooled across all surveys range from 18-89. (The GSS does not distinguish among respondents ages 90 and above and simply assigns these individuals an age score of 89). The oldest cohort member was born in 1883, and the youngest was born in 1992. To test for effect of age, the variable was constructed into five year intervals to enable the age-period-cohort analyses, and for purposes of the OLS regression analyses the respondent’s age was divided into four dummy age groups with the youngest age group (18-29), omitted as the baseline. Trust has been linked with social inequality; the more inequality an individual feels the lower their level of trust2,10. In the past, minorities and women have experienced discrimination and been subject to inequality, so as a result they are less likely to trust others2,14. We created dummy variables for respondent’s race with one indicating “white” and gender coded so that one represented “female” respondents. It has been argued that trust is influenced by wealth as represented by those who “have” and the “have not’s”. We include several different indicators of success and well-being including education, social class and life satisfaction. Education is a measure ranging from 0 to 20, representing the actual number of years of education a person has. Using respondent’s self-reported social class, we construct four dummy variables – lower, working, middle, and upper (baseline group) – to test the belief that people who feel vulnerable or disadvantaged find it riskier to trust because they’re less well-fortified to deal with the consequences of misplaced trust. Life satisfaction has consistently proved a predictor of trust2,5,13. To measure life satisfaction GSS asks respondents “Taken all together, how would you say things are these days--would you say that you are very happy, pretty happy, or not too happy?” The variable was coded into three dummy variables with "not too happy" used as the baseline. International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 9 We also model individuals’ decisions to trust as a function of their region of residence and marital status. On average, Southerners are less trusting than non-Southerners, and the research suggests that these differences occur even after controlling for regional variation in other factors related to trust. Region is coded as six regional dummies (New England States, Mid-Atlantic States, North Central States, Mountain States, Pacific States and Southern States as the reference group). The effect of marital status is twofold. First, married individuals have been found to be more likely to express trust. Such differences in trust are likely a consequence of greater social interaction with family and support in married households. Second, and in contrast, distrust is more common among those whose relationship status is divorced or separated. It's easy to conceive that individuals experiencing relationship trauma such as a divorce or separation from their spouse, believe that no one is trustworthy and that close relationships will only end up hurting them. As such, respondent’s marital status is coded as two dummy variables one indicating married or not, and a different measure coded so that one represents whether the respondent is separated or divorced. Variables were also included to capture religious preference and religious behavior. Religious tradition consists of five dummy variables: Protestant, Catholic, Jewish, Other, and Seculars (or the non-religious; omitted as the baseline group). To measure religiosity (religious behavior) the GSS asks how often the respondent attends a religious service. Respondents were allowed to respond “never, less than once a year, about once or twice a year, several times a year, once a month, 2-3 times a month, nearly every week, every week, and several times a week”. These responses were then coded as individuals who reported attending a religious service “2-3 times a month” to “several times a week” as having “regular attendance”. Individuals who responded “about once or twice a year” to “once a month” were seen as attending “sometimes.” Those who claimed “never” to “less than once a year” were placed as individuals who “never attend.” Individuals who never attend a religious service are the baseline for the regression model. Results and Discussion Explaining Change over Time: Age-Period-Cohort Effect (A-P-C): Several studies have noted concerns about the condition of U.S. civic culture, especially among the young. In a cohort comparison of today’s young adults, not with today’s older adults, but with the young adults of the past—a common occurrence we find is evidence of diminished civic attachment. Most scholars attribute the decline in trust as stemming from either generational replacement or period effects, or a combination of the two effects. To better understand the dramatic nature of the shift in interpersonal trust due to period effects, the overall average trends (see figure 1 for mean trust levels of all adults – 1972-2010) shows that trust levels appear to be irregular, but markedly higher in the 1970s and early 1980s. Levels of trust then become substantially lower but relatively stable until the mid-90s when trust falls by about a 1/3 of a point, after which trust rebounds but not to the same level it had been before the trust decay (i.e. in 1984, the mean for the combined trust scale was 6.37; by 2010, it had fallen to 5.85). These average period trends confirm that Americans today are significantly less trusting than was true a generation ago.Figure-1 Overall Trust Levels 1972-2010 International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 10 It is also plausible that the relatively high levels of trust among the Greatest Generation results from differences in their formative socialization as compared to that of more recent cohorts. For example, it has been argued that people born before 1930 are more trusting and civic-minded as a result of their big coming-of-age experience (World War II), while successive generations are less trusting as a result of theirs (take your pick: Vietnam, Watergate, the coarsening of the popular culture, television, suburbanization)1,21. The usual assumption of the cohort interpretation is that certain events in history make an indelible imprint on the young, whereas older individuals are better equipped to resist its influence by their previous life experiences. A cohort explanation argues that differences in attitudes are due to the social and cultural climate prevailing so that members of a generation share the same experiences, distinctive from the experiences of their adjacent counterparts22-23. The cohort explanation holds that, as new birth cohorts enter the electorate with different life experiences, they may develop different political attitudes that influence them for a lifetime. This effect implies that through the process of cohort replacement, mistrustful attitudes may grow in the general population. On the other hand, these patterns may not be reflective of cohort effects at all but rather, these results might be explained, fully or partly, by aging and life-cycle factors. Research provides evidence of an aging - net of period and cohort - effect on trust with younger age groups predisposed to be less trusting than older age groups. The results show that Americans begin adult life with low levels of trust, become more trusting as they grow into middle age, and then maintain this higher level of trust for their remaining years18. We begin by analyzing trust rates by age since 1972 and then we examine across-cohort changes in trust and changes over time. In Table 1 (graphical representations available upon request), each column represents age differences in trust. Similarly, each row gives the period - i.e. year-to-year - differences within each age group. A clear pattern emerges: younger age groups are less trusting than the older age groups with a strong tendency for trust to increase with age until the mid-40's, and then remain fairly stable until the oldest ages. The modest increase in trust that appears after age 70 may be a sign of a genuine age effect but this pattern can also be attributed to differential mortality (i.e. less trusting persons dying at younger ages than persons with higher trust). Similarly, the year-to-year differences within each age group shows random fluctuation until the mid-1980s when trust drops. This decline is followed by periods of randomness but trust never completely recovers and trust levels within the age groups are much lower than in the years prior to the decline in the late 1980s (see “Year Totals” row in table-1). The evidence here clearly points to the conclusion that the life-cycle model does not explain changes in social trust. While the patterns support a strong, consistent relationship between age and trust - with the young generally less trusting than the old – there is no indication of growing convergence (a requirement of the life-cycle model) with the young moving closer in line with their elders. Table-1 Trust Index Means by Age and Year of Survey Year of survey Age ‘72-76 ‘77-81 ‘82-86 ‘87-91 ‘92-96 ‘97-01 ‘02-10 Age totals 20-24 5.59 5.33 5.84 5.38 4.96 5.00 5.08 5.31 25-29 6.02 6.16 5.96 5.66 5.25 5.19 5.26 5.64 30-34 6.23 6.15 5.94 5.86 5.37 5.76 5.40 5.82 35-39 6.33 6.65 6.38 6.21 5.95 6.02 5.49 6.15 40-44 6.51 6.42 6.55 6.39 5.84 6.02 5.91 6.24 45-49 6.48 6.87 6.54 6.55 6.33 6.15 6.11 6.43 50-54 6.40 7.04 6.51 6.20 6.33 6.35 5.97 6.40 55-59 6.27 6.37 6.47 6.31 6.38 6.37 6.16 6.33 60-64 6.36 6.46 6.61 6.37 6.18 6.24 6.41 6.38 65-69 6.01 6.55 6.42 6.83 6.42 6.39 6.64 6.46 70-74 6.27 6.60 6.30 6.55 6.35 6.33 6.38 6.40 75+ 6.43 6.73 6.81 6.60 6.31 6.51 6.65 6.58 Year totals 6.20 6.34 6.29 6.15 5.87 5.95 5.86 6.18 Note. Each diagonal approximates a “standard cohort table.” For example, the 1952-1956 birth cohort is represented in the 20-24 year olds interviewed in 1972-1976 with a mean trust of 5.59. Five years later, this cohort is 25-29 years old and their mean (6.16) appears in the second row in the 1977-1981 column. This same cohort appears in the third row (5.94 mean trust), in the 1982-1986 survey, and so on. Source: General Social Survey, 1972-2010. International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 11 In Table 1 the birth cohorts are roughly displayed in the upper left-to-lower right diagonal cells. The boxed columns approximate a standard cohort table, in which the 5-year distance between surveys is equal to the width of the age categories. For example, the 1952-1956 birth cohort is represented in the 20-24 year olds interviewed in 1972-1976 with a mean trust of 5.59. Five years later, this cohort is 25-29 years old and their mean (6.16) appears in the second row in the 1977-1981 column. This same cohort appears in the third row (5.94 mean trust), in the 1982-1986 survey, and so on. Table 1 demonstrates with stunning clarity the net effect of age-differences on the erosion of trust. As chronicled by others, entering cohorts have shown rather steady declines ever since the late-1940s1-4. We see that when controlling for age (but not for period), the older cohorts have variable, but high levels of trust up until the cohorts of the 1940s, after which each succeeding cohort displays lower trust than the cohorts that precede it. The pattern indicates that more recent cohorts are less trusting and the decline appears to have increased in recent cohorts. The importance of A-P-C in explaining variations in trust stems from the understanding that the specific source of change can have important - albeit different and possibly, negative - consequences on society. Age differences in trust are not particularly troubling if the relative disparity between old and young is persistent. If trust changes in predictable patterns such as increases with age (life-cycle or age effect) then the impact on societal trust levels is negligible, even though individuals have become more trusting as they grow older. However, if each new birth cohort enters adult life with different experiences and characteristics affecting them for the remainder of their lives, a society will change even if no individual changes during his or her lifetime1-2. It becomes possible then, through the process of generational replacement, that as more trusting cohorts die and are replaced by less trustful cohorts, the average trust in a society will gradually decline, even if all individuals maintain a steady level of trust throughout their lifetimes. A simple calculation allows us to see how much of the change in trust opinions over time is due to generational replacement and, how much change is due to other factors. The approach requires dividing the data into two time periods (conducted years apart) and statistically weight the results of the second poll to match the population composition at time 1 to estimate what public opinion would have looked like in time 2 if no generational replacement had occurred. Put simply, one can estimate the impact of cohort replacement by determining what public opinion would have looked like without it. How much would responses differ if the composition of the sample by generation was held constant? Table-2 Estimation of Opinion without Generational Replacement Cohort Age in 1984 Mean Trust in 1984 Percent of 1984 Pop. Mean Trust in 2004 Percent of 2004 Pop. Product 1980-1984 0-4 -- 0.0 5.21 0.09 0.0 1975-1979 5--9 -- 0.0 5.22 0.10 0.0 1970-1974 10--14 -- 0.0 5.57 0.10 0.0 1965-1969 15-19 5.77 0.02 5.47 0.10 0.11 1960-1964 20-24 5.87 0.15 6.02 0.11 0.90 1955-1959 25-29 5.92 0.12 6.65 0.11 0.80 1950-1954 30-34 6.03 0.10 6.61 0.10 0.66 1945-1949 35-39 6.56 0.11 6.27 0.09 0.69 1940-1944 40-44 6.64 0.09 6.36 0.07 0.57 1935-1939 45-49 6.79 0.08 6.37 0.05 0.51 1930-1934 50-54 6.66 0.06 5.85 0.03 0.35 1925-1929 55-59 6.60 0.07 7.42 0.03 0.52 1920-1924 60-64 6.74 0.06 6.58 0.02 0.39 Pre-1920 65+ 6.68 0.14 6.78 0.01 0.95 Total 6.37 1.00 6.00 1.00 6.46 Total mean trust change: 6.00 – 6.37 = -0.37, Change due to generational replacement: 6.00 – 6.46 = -0.46, Percentage change due to generational replacement: -0.46/-0.37 = 24% International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 12 As the data in Table 2 indicate, mean trust levels dropped by more than a 1/3 of a point between 1984 (6.37) and 2004 (6.00); and the decline in trust was brought about in part by generational replacement. Specifically, in column 4 we write the population distribution by cohort as it existed at time 1 (1984); in column 5 we enter mean trust levels by cohort in time 2(2004), and then sum across all cohorts. The product (6.46) is our estimate of what trust opinions would have looked like in 2004 if no generational replacement had occurred since 1984. By subtracting this figure from the actual population attitude at time 2 (6.00), we obtain an estimate of the effects of population turnover. In this case, 24% of the trust decline is due to generational replacement. Thus, generational replacement certainly matters in the development of attitudes towards others but this model alone does not explain trust trends. Other factors are also important and we must look elsewhere for a more comprehensive understanding of the origins of trust. We now employ an approach to investigating what it was about American politics and society that changed trusting opinions. In particular, we examine how much of the observed variation is the result of changing individual characteristics and beliefs. Explaining Change over Time: Individual Factors: Fully explaining why trust changes is beyond the scope of this paper. However, we can get a start by examining how the attitudes of various social and demographic groups have changed in the past four decades. That is the task we undertake in this section. Descriptive analyses between 1984 and 2010 confirm that interpersonal trust declined among every social and demographic group examined (graphical representations available upon request). However, the rate of decline varied with the most notable trust losses occurring by educational attainment, gender, marital status, region, and personal happiness. A consistent finding in studies of generalized trust is the great effect that education has in explaining the development of more trusting attitudes2,5. More educated people might trust others more either because they associate with other more educated and trustworthy people or because education raises social skills and status and thus increases the ability to punish or reward others24. While we find greater trust is associated with higher levels of attainment, the education patterns also demonstrate that the decline in trust is vast regardless of educational attainment. In fact, between 1984 and 2010, trust dropped by more than a ½ point on the 7-point index regardless of differences in schooling. The analogous decline is somewhat surprising given that education in particular figures prominently in the literature on trust, with studies suggesting a strong effect associated with respondents’ cognitive abilities to trust5,14. Groups that have been historically discriminated against, blacks and females, also register trust declines. While there are significant racial differences in trust, with whites substantially more likely to trust others, (resulting in a trust deficit of more than a point in 2010) the falloff in trust since 1984 is also much greater for white respondents. As a consequence, mean trust levels dropped from 6.60 in 1984 to 6.11 in 2010. Among Blacks however, and despite considerable variation over the years, trust rates are nearly identical in 2010 to those recorded in 1984 (5.05 compared to 5.04 in 2010). Largely absent are gender differences in trust. Men were about a fifth of a point lower on the trust index in 1972, and despite considerable trust variability over time showing greater trust among women, the gender gap all but disappeared by 2010. While both sexes experienced trust declines, among females the downward slide was more prominent. As discussed earlier, success and well-being is associated with greater trust. Our findings make it abundantly clear that inequality diminishes individual levels of interpersonal trust. First, we find that social class and life satisfaction are positively related to trust. The effect of social class is that the less well-off are least trusting of others and the trust gap is growing. The difference between the lower and upper classes is a massive 2.36 on the trust index. Moreover, the trust gap between the middle and working classes as compared to the upper class has also grown substantially since the mid-1980s. Second, happiness results in individuals who are more trusting. People who are “very” or “pretty” happy are much more likely to be trusters than folks who are unhappy in their personal lives. Finally, we find life experience such as marital status and one’s surroundings (i.e. place of residence) contributes to the development of trust in others. Marriage contributes to trust - probably due to the lack of relationship trauma experience – so that distrusting individuals are those who have never been married, separated or divorced. It's easy to conceive that individuals experiencing relationship distress (e.g. a divorce or separation) believe that no one is trustworthy and that close relationships will only end up hurting them. Trust also varies with region. Southerners are the least trusting, and New Englanders, closely followed by the residents of mountain states, are most trusting of others. The distrust among Southerners is no doubt partly explained by the higher percentage of Blacks who live in the South. Moreover, recent research has shown that trust is associated with a wide range of macro-level phenomena, including economic development and civic engagement2,7 and it may be that some of the regional differences in trust are explained by regional variations in these social and economic aspects. The patterns for religious preference indicate trust differs by affiliation, but largely missing is any trust disparity between the religious and the nonaffiliated. While there is considerable variation by tradition over time, the findings demonstrate that trust is substantially lower today than in the mid-1980s. The lone exception to this pattern is among Jews, but this is nothing new - i.e. Jews are typically more trusting than the other traditions25. In the case of Jews however, the sample sizes are quite small and when we divide our sample into three International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 13 broad periods (to see if the pattern appears with reasonably large sample sizes) we find the same trend that appears among the other religious traditions is observed for Jews. In contrast, religiosity shows genuine trust differences. Regular attendees are much more likely to express trust than those who “sometimes” or “never” attend. And, while trust fell among all groups, in 2010 the trust gap closed to about a 1/5 of a point between those who never attend and attend regularly - suggesting a diminishing influence of church attendance on trust. Multivariate Regression Model: We now turn to the two regression models explaining trusting attitudes. Since trust opinions among the various groupings changed over time, we estimate two models – one for the period of high trust - 1972-1984, and the other for the period of declining trust - 1986-2010. Pooling the GSS data into these two sub-samples provides the opportunity to look at how trust opinions have changed from the more trusting mid-1980s period to the distrustful years of 1986 to 2010. The various determinants of trust discussed above are included in these models in Table 3. Table-3 Explaining Trust Trends 1972-1984 Coefficient s.e. 1986-2010 Coefficient s.e. Significant Diff. Over Time Constant 3.07 3.36 30-49 0.13 0.06 0.13 0.04 *** 50-64 0.13 0.06 0.17 0.05 65+ 0.10 0.07 0.19 0.05 *** Female 0.04 0.04 0.02 0.03 Education (�12 years) 0.15 0.05 0.15 0.03 White 0.16 0.07 0.16 0.05 ** Married 0.02 0.06 0.00 0.04 Divorced/Separated -0.01 0.09 -0.03 0.05 Very Happy 0.22 0.07 0.18 0.05 Pretty Happy 0.18 0.07 0.14 0.05 Middle Class 0.07 0.13 -0.03 0.08 Working Class -0.02 0.13 -0.13 0.08 Lower Class -0.06 0.16 -0.10 0.10 New England States 0.06 0.10 0.05 0.07 Mid-Atlantic States 0.06 0.06 0.06 0.05 North Central States 0.12 0.05 0.11 0.04 Mountain States 0.06 0.10 0.06 0.06 Pacific States 0.05 0.07 0.05 0.05 Protestant -0.01 0.09 0.02 0.05 Catholic -0.03 0.10 0.00 0.06 Jewish -0.01 0.17 0.00 0.11 Religion Other -0.02 0.20 0.00 0.09 Attend Regularly 0.07 0.06 0.08 0.04 Attend Sometimes 0.04 0.06 0.02 0.04 R 2 0.16 0.16 N 10,114 19,665 Note. The bolded figures represent the dummy variables that have a statistically significant effect on trust (p.05 or less). ***p .000; **p .01 for difference between 1972-1984 and 1986-2010 equations International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 14 To begin, the centrality of some of these relationships to building trust is apparent. Beginning with the demographic variables, we see that their influence is roughly similar in the two models. Trust rates are relatively low in the early ages (18-29 year olds, the omitted age group). The coefficients for age are significant and positive, confirming earlier results that indicate trust increases with age. The change in the size of the coefficient between the time periods points to the growing importance of age on the creation of trust. Those who are better educated (as measured by having more than 12 years of schooling) are more trusting. Contrary to expectations, women are more trusting than men. These significant differences held in both time periods. Not unexpectedly, we find that whites, relative to blacks, are more trusting. Moreover, the impact of race significantly increased from the 1970s and 1980s to the 2000s. It is worth noting that the coefficients capturing racial differences in trust are quite large and similar to those found elsewhere14-15. Marital status mattered but not entirely in the manner we expected. Married respondents were no more likely to trust than those who were unmarried (e.g. divorce, separated, widowed, or never married). However, those who were specifically divorced or separated were less trusting and the influence became statistically significant in the more recent time period suggesting the growing importance of relationship trauma. The indicators of success and well-being have the strongest influence on trust. In fact, one’s state of happiness proves to be the most powerful force driving trust. Irrespective of time period, happier individuals are more trusting individuals. In addition, reported social class has been steadily increasing in influence over the survey years, and it is the strongest influence that reduces trust between the two time periods. The coefficients for working and lower class (as compared to the omitted upper class variable) in the 1986-2010 time period are quite large, the largest negative influence, accounting for some of the decline in trusting orientations. Region also mattered as expected, and this held true in both time periods. Southerners (the comparison group) are the least trusting than those living elsewhere in the country, whereas the North Central states are the most trusting. Beyond this there is virtually no trust variance associated with place of residence. While we do not find evidence of a relationship between religious preference and trust, we see that religious commitment has a significant influence on trust. Those who are more committed (as measured by attendance at services) are more trusting, and this relationship held steady over time. These findings are in line with studies that link religious behavior as a source of higher levels of trust and civic participation2,10. Conclusion To summarize, the evidence presented here confirms that Americans today are significantly less trusting than was true a generation ago. The substantial over-time decrease in the extent to which Americans find others benign and trustworthy is partly attributed to a negative nonlinear net cohort effect. For cohorts born post-1948, the characteristic pattern is trust declines and about of quarter of the decline results from generational replacement. While there is a clear tendency for trust to increase with age until the mid-40's, the evidence demonstrates that, despite this positive age effect, no improvement in trust occurred. The simple explanation is that each cohort leaving the population is replaced by a new cohort with much lower levels of trust, and as more trusting cohorts die and are replaced by less trustful cohorts, average trust in society gradually declines even if all individuals maintain a steady level of trust throughout their lifetimes. These are the classic age-period-cohort features of trust as observed in two prior large scale studies18,20. The remaining loss in trust originated elsewhere. We find that apart from some greater effects for the measures of age and race in the latter time period, the two equations yield virtually identical results. Both models show that many individual characteristics influence trust and the effects are quite similar with those found elsewhere. Consistent with previous research1,5, we find life satisfaction, education, social class, region, gender, and race are important determinants of interpersonal trust. People who live in states with high levels of trust are substantially more likely to be trusters themselves. Trust increases with education and social class whereas divorced or separated respondents express less trust. However, the most powerful determinant of trust is personal happiness2,8. The results emphasize a close connection between social trust, happiness, and well-being. Our analyses suggest that if the patterns uncovered here continue, mistrust will become more widespread presenting serious consequences for U.S. society. While it has been established that many individual traits influence trust, the effects are relatively comparable in both models suggesting the differences we find do not explain the dramatic decrease in social trust witnessed since the mid-1980s, and other factors must be responsible. Therefore, while the results do not fully explain why trust decreased, they do tell us where to look for explanations. For these reasons, we should continue to track changes in American trust. References 1.Putnam R.D., Bowling alone: America’s declining social capital, Journal of Democracy,6, 65–78 (1995)2.Putnam R.D, Bowling Alone: The Collapse and Revival of American Community, Simon and Schuster, New York (2000) International Research Journal of Social Sciences____________________________________________________ISSN 2319–3565Vol. 2(1), 7-13, January (2013) Int. Res. J. Social Sci. International Science Congress Association 15 3.Smith T.W., Factors relating to misanthropy in contemporary American society, Social Science Research,26, 170–196 (1997)4.Rahn W.M. and Transue J.E., Social trust and value change: The decline of social capital in American youth, 1976–1995, Political Psychology19, 545–565 (1998)5.Brehm J., and Rahn W., Individual-level evidence for the causes and consequences of social capital, American Journal of Political Science41, 999–1023 (1997)6.Paxton Pamela, Trust in Decline?, Contexts,4(1), 40-45(2005)7.Fukuyama F., Trust: The Social Virtues and the Creation of Prosperity, Free Press, New York (1995)8.Inglehart Ronald, Trust, well-being and democracy, in Warren, Mark E. (ed.) Democracy and Trust, Cambridge University Press, 88-120 (1999) 9.Patterson Orlando, Liberty against the democratic state: on the historical and contemporary sources of American distrust, in Mark E. Warren (ed.) Democracy and Trust, Cambridge University Press (1999)10.Uslaner Eric M., The Moral Foundations of Trust, New York, Cambridge University Press (2002)11.Delhey, Jan and Kenneth Newton. Who Trusts? The Origins of Social Trust in Seven Nations, European Societies,5(2), 93-137 (2002)12.Yamagishi Toshio and Yamagishi Midori, Trust and commitment in the United States and Japan, Motivation and Emotion, 18(2), 129-66 (1994)13.Uslaner Eric M., Producing and Consuming Trust, Political Science Quarterly, 115, 569-590 (2000)14.Alesina A. and E. La Ferrara, Who trusts others?, Journal of Public Economics, 85, 207-234 (2002)15.Shaleva Anna., How Does External Conflict Impact Social Trust? Evidence from the 9/11 Attacks as a Natural Experiment in the US (2011) Available at SSRN: http://dx.doi.org/10.2139/ssrn.1769023 (2011)16.Zak P. and S. Knack, Trust and Growth, Economic Journal, Royal Economic Society, 111(470), 295-321 (2001)17.Rotter J.B. Interpersonal trust, trustworthiness, and gullibility, American Psychologist,35, 1–7 (1980)18.Robinson Robert V. and Elton F. Jackson, Is Trust in Other Declining in America? An Age-Period-Cohort Analysis, Social Science Research,30, 117-145 (2001)19.Kramer R.M., The sinister attribution error: Paranoid cognition and collective distrust in organizations,Motivation and Emotion,18, 199-230 (1994)20.Clark April K. and Marie Eisenstein, Interpersonal Trust: An Age-Period-Cohort Analysis Revisited. Social Science Research Available at: http:/ /dx.doi.org /10.1016 /j.ssresearch.2012.09.006 (2012)21.Taylor Paul, Cary Funk and April K. Clark, Americans and Social Trust: Who, Where and Why? Pew Research Center, Social and Demographic Trends Report available online: http://pewsocialtrends.org/pubs/414/americans-and-social-trust-who-where-and-why (2007)22.Mannheim Karl, The problem of generations. In The new pilgrims, ed. Philip G. Altbach and Robert S. Laufer, Reprint, New York: David McKay, 101-37 ([1928] 1972)23.Jennings M. Kent and Richard G. Niemi, Generations and Politics: A Panel Study of Young Adults and Their Parents. Princeton, NJ: Princeton University Press, 118-123 (1981)24.Glaeser E., Laibson D., Scheinkman J. and Soutter C., What is social capital? Quarterly Journal of Economics, 65, 811-846 (2000) 25.Uslaner Eric M., Daphna Canetti-Nisim and Ami Pedahzur, Terrorism and Trust: Sustained Violence and the Social Fabric in Israel (paper presented for the Midwest Political Science Association) (2005)