Immigrant Paradox? Generational Status, Alcohol Use, and Negative Consequences across College

Abstract

The current study examined linkages between generational status, alcohol use, wanting to get drunk, and negative alcohol-related consequences during college. We tested whether immigrant students’ longitudinal alcohol use trajectories converged to dominant unhealthy patterns or whether immigrant students maintained healthier patterns across college. We also examined if the weekend exerted equal risk for students of different generations. Furthermore, we explored whether patterns were consistent among Latinx and Asian American students. Stratified random sampling identified first-year students attending a US college. A longitudinal daily diary design was used; students completed web-based surveys for up to 14 days within each of 7 semesters. Each day, participants (N=689; n_{1st generation} =114; n_2nd generation=244; 51% female) reported their alcohol use and consequences (N=55,829 days). Multi-level models demonstrated that compared to 3^rd generation students, 1^st generation students were more likely to abstain from alcohol and less likely to binge drink and want to get drunk. First generation students also experienced fewer negative alcohol-related consequences. The protective effect of being 1^st generation was maintained across college semesters, with subgroup analyses focusing on Latinx and Asian American students largely supporting the main findings. However, for abstaining and negative consequences, the weekend effect was less pronounced for immigrants than later generation students. For example, the difference in negative consequences between 1^st (vs. 3^rd) generation students was largest on the weekend. Additional work is needed to understand how 1^st generation students leverage protective factors to abstain from alcohol use even when exposed repeatedly to “wet” drinking environments.

1. Introduction

Almost a quarter of college students in the United States are immigrants or children of immigrants (Staklis & Horn, 2012). Although a college degree can provide substantial economic and social returns (Hout, 2012), paradoxically, aspects of the college environment may pose a threat to the health of immigrant students. Attending college can expose students to a “wet” social environment that facilitates underage drinking and heavy episodic drinking. Indeed, college students are more at risk for binge drinking than non-college-going young adults (Slutske et al., 2004; White & Hingson, 2013) and each year, over 500,000 college students are injured and an estimated 1,800 die from alcohol-related causes (Hingson, Zha, & Weitzman, 2009).

Research indicates that immigrants to the United States demonstrate healthier alcohol use patterns than native-borns; they are less likely to drink alcohol, report alcohol-related problems, and experience alcohol abuse/dependence (Bacio, Mays, & Lau, 2013; Eitle, Wahl, & Aranda, 2009; Johnson, VanGeest, & Cho, 2002; Lee, Han, & Gfroerer, 2013; Lo, Cheng, & Howell, 2014; Peña et al., 2008; Prado et al., 2009; Szaflarski, Cubbins, & Ying, 2011).These findings are in line with the broader literature on the “immigrant health paradox.” The paradox is that, despite various disadvantages and challenges, immigrants often report better health behaviors and outcomes than their native-born counterparts (Markides & Coreil, 1986; Singh & Hiatt, 2006). Explanations for this paradox have been explored including immigrant selection (i.e., healthier people immigrate), acculturation stress, and US-born populations’ increased exposure to risky contexts. Furthermore, assimilation (or, unidirectional acculturation) may explain the paradox. That is, individuals may adopt the (unhealthy) beliefs and practices of US culture and relinquish those of their heritage culture (Gordon, 1964). Generational status is often considered a distal indicator of assimilation to the US. First generation are born outside of the US, 2^nd generation are born in the US to foreign-born parents and, 3^rd generation are US-born with US-born parents.

The US racial-ethnic groups with the largest number of immigrants are Hispanic or Latinos (hereafter Latinx) and Asians. Research with Latinx and Asian American college students has shown that alcohol use varies by generational status and acculturation levels. Studies have shown that among Latinx students, those who were more assimilated to US culture consumed more alcohol, drank heavily more frequently, and were more likely to binge drink (Des Rosiers, Schwartz, Zamboanga, Ham, & Huang, 2013; Schwartz et al., 2014). However, not all studies are consistent; some have documented no association between acculturation and alcohol use (Cano et al., 2015; Mercado, Ramirez, Sharma, Popan, & Latorre, 2017) and a meta-analysis of studies involving Latinx participants in the US (including college and non-college samples) suggested that acculturation predicted more alcohol use, but associations were small (Alvarez, Frietze, Ramos, Field, & Zárate, 2017). Among Asian American students, acculturation has been linked with current alcohol use, heavy episodic use, and alcohol problems (Hendershot, MacPherson, Myers, Carr, & Wall, 2005; Luk, Emery, Karyadi, Patock-Peckham, & King, 2013). A recent meta-analysis demonstrated that acculturation predicted alcohol use and hazardous drinking intensity among Asian Americans; associations were stronger during emerging adulthood than during adolescence or middle adulthood (Liu & Zamboanga, 2018). Yet, not all studies have accounted for generational status or acculturation (Iwamoto, Corbin, & Fromme, 2010) and most extant research exploring generational status and alcohol consumption during the high-risk college years has been cross-sectional, limiting understanding of within-person changes in alcohol use over time.

1.1. Longitudinal Change and Weekend Variation

Longitudinal research is needed to understand whether the college environment elicits differential change in alcohol use and related consequences among 1^st, 2^nd, and 3^rd generation students. It may be that contact with “wet” college environments increases alcohol misuse among immigrant students, and over time their behaviors converge to unhealthy mainstream norms (Gordon, 1964). Alternatively, patterns may remain distinct, with immigrant students more likely to abstain from alcohol or consume in moderation across their college careers.

Just as alcohol misuse might change across college, so too can it vary across days of the week. Research has documented that students drink more alcohol and experience the worst consequences on the social weekend –-Thursday, Friday, and Saturday (Del Boca, Darkes, Greenbaum, & Goldman, 2004; Maggs, Williams, & Lee, 2011) –- when alcohol-related emergency department visits for young people also peak (Crane & Cai, 2017; Elder, Shults, Swahn, Strife, & Ryan, 2004). It is possible that all students —regardless of generational status—increase their alcohol use similarly on weekend days compared to weekdays. Alternatively, being an immigrant may be especially protective during heightened periods of risk making the increase in alcohol consumption on the weekends more pronounced for later generations.

1.2. The Current Study

Using daily longitudinal survey data from students attending a college with drinking-conducive traditions, we tested whether: (1) generational status is associated with alcohol use and negative alcohol-related consequences; (2) generational status is associated with longitudinal alcohol use trajectories from first through fourth year of college; (3) the weekend exacerbates the risk of alcohol use equally across generations. Finally, (4) in exploratory analyses the sample was disaggregated to examine associations among subsamples of Latinx and Asian American students.

2. Methods

2.1. Data

Data from the University Life Study (ULS) were used. These data have been described extensively elsewhere (Greene & Maggs, 2014; Howard, Patrick, & Maggs, 2015; Patrick, Maggs, & Lefkowitz, 2014); however, this is the first study to focus on generational status in the ULS. Briefly, first-semester students attending a college in the Northeastern US were identified using a stratified random sampling procedure (by gender and the four largest racial/ethnic groups in the US). Students who were U.S. citizens or permanent residents were eligible to participate; international students were ineligible. Recruited students (N=744) were followed longitudinally for seven college semesters. Each semester, students completed one longer survey and up to 14 consecutive daily surveys. All surveys were web-based. The daily links were emailed at ~4am and referred to the previous day’s activities. Students were followed longitudinally regardless of enrollment status and 82% were retained at semester 7.

The current paper focuses on participants who provided information about their and their parents’ birthplaces (i.e., US-born versus foreign-born), collected during Semesters 2 and 3. Students who did not participate in either of these semesters were excluded as was one student who was missing daily survey information, leaving a final sample of 689 participants (Ndays=55,829). On average, participants completed 81 daily surveys out of a possible 98. Analyses included all occasions that participants completed daily diaries because the multi-level modeling strategy allows for missing data at level 1.

Participants averaged 18.5 years old (SD=.43) in semester 1 and 21.5 years old (SD=.42) in semester 7. About half (51.1%) were female and one quarter (25.3%) participated in Greek life. Due to the sampling strategy, the sample was more diverse in terms of race-ethnicity and generational status than the university from which it was drawn. Approximately 25.4% self-identified as Latinx or Hispanic, 26.9% as Non-Hispanic (NH) white, 15.5% as NH African American or Black, 23.4% as NH Asian American or Hawaiian/Pacific Islander, and 8.9% indicated that they were NH and multiracial. In addition, the sample was 16.6% 1^st generation (n=114), 35.4% 2^nd generation (n= 244) and 48.0% 3^rd generation (n=331).

2.2. Measures

2.2.1. Alcohol Use.

Each day, participants responded to several questions about their alcohol consumption and consequences which were used to create the following measures. Participants who did not drink on a given day were coded as having abstained. Binge drinking occurred when participants drank 4+ (females) or 5+ (males) alcoholic drinks in 2 hours. Participants also reported whether they had wanted to get drunk (yes/no), and whether they experienced any of six negative physical or behavioral consequences (e.g., had a hangover, lost control) as a result of drinking (Patrick & Maggs, 2011). The sum was computed to create a count variable, negative consequences. Days that students did not drink were coded as zero for negative alcohol-related consequences and wanting to get drunk.

2.2.2. Generational Status.

In line with Census Bureau (2016) definitions, first generation students were born outside of the US to one (or more) parents who were also foreign born. Second generation students were born in the US to at least one foreign-born parent. Third generation students were born in the US to US-born parents. Participants who were not born in the US and who reported no biological parent information (n=13) were presumed adopted and coded as 3^rd generation. Participants with valid parent information who were missing information about their own birthplace (n=4) were coded as born in the US.

2.2.3. Timing Variables.

Semester in college was coded 1 through 7. The social weekend was coded as Thursday, Friday, Saturday = 1 and Sunday through Wednesday = 0.

2.2.4. Control Variables.

Our models include gender, Greek membership, and race/ethnicity. Prior research has indicated that men average more drinks than women, members of fraternities or sororities drink more than non-members, and White college students misuse alcohol at higher rates than non-White students (Borsari, Murphy, & Barnett, 2007). Gender was dichotomized. A participant was classified as Greek if he or she reported any participation in a fraternity or sorority during college. Based on self-report, participants were coded as White, NH (reference group), Latinx (regardless of race), Black or African American NH, Asian American or Hawaiian Pacific Islander NH, or multi-racial NH.

2.3. Analytical Strategy

Multi-level mixed effects generalized linear models were used given the longitudinal data and corresponding correlated errors (Hoffman, 2015; Raudenbush & Bryk, 2002). Days (level 1) were nested within semesters (level 2) that were nested within people (level 3). Analyses were computed using the meglm command in Stata 14. Significant interactions were probed using the margins command. Odds ratios were estimated for binary outcomes whereas incident rate ratios were estimated for negative consequences. For each outcome, we tested the association between immigrant status (level 3) and the outcome, controlling for gender, Greek, and race/ethnicity (level 3), semester (i.e., the longitudinal time trend, level 2), and whether it was a weekend day (level 1). Subsequently, we tested interactions of generational status with semester as well as whether the day was a weekend.

After conducting these models, we computed subgroup analyses to assess the consistency of findings among Asian American and Latinx students (N=366). First generation (i.e., foreign-born) were contrasted with US-born (i.e., 2^nd and 3^rd generation students) given small numbers of 3^rd generation Asians. Models controlled for gender, race/ethnicity, Greek status, semester, and weekend. Subsequently, we examined whether the foreign birthplace-alcohol association varied across time or race/ethnicity (i.e., Foreign-born × Semester interaction and a Foreign-born × Race/Ethnicity interaction). Exploratory within group analyses for Latinx (n=175) and Asian (n=161) students further examined the role of foreign-born status.

3. Results

We began by conducting multi-level models examining associations between generational status and the four alcohol use outcomes, controlling for relevant covariates (aim 1). These results demonstrated that, compared to 3^rd generation students, 1^st generation students displayed healthier alcohol consumption patterns (see Table 1). First generation students were more likely to abstain, less likely to binge drink and want to get drunk, and they experienced fewer negative alcohol-related consequences than their 3^rd generation counterparts. Alcohol use and consequences did not differ significantly between 2^nd and 3^rd generation students when relevant covariates were included.

Table 1.

Alcohol Use and Related Consequences by Generational Status: Adjusted Odds Ratios and Incident Rate Ratios

	Abstain	Binge Drink	Wanted to get Drunk	Negative Consequences
	OR	OR	OR	IRR
Fixed Effects
Generational status
2nd generation	1.085	0.894	0.777	1.132
	(0.189)	(0.186)	(0.182)	(0.243)
1st generation	1.897**	0.464**	0.314***	0.562*
	(0.418)	(0.124)	(0.096)	(0.155)
Male	1.186	0.781	0.797	0.883
	(0.152)	(0.120)	(0.139)	(0.140)
Greek	0.459***	2.882***	3.461***	2.483***
	(0.068)	(0.502)	(0.680)	(0.446)
Race-ethnicity
Black/African American	2.601***	0.213***	0.261***	0.280***
	(0.559)	(0.055)	(0.076)	(0.075)
Latinx	0.850	1.020	1.171	0.955
	(0.170)	(0.241)	(0.311)	(0.232)
Asian American	2.089**	0.339***	0.334***	0.325***
	(0.490)	(0.095)	(0.107)	(0.095)
Multiracial	1.433	0.591+	0.639	0.667
	(0.369)	(0.180)	(0.219)	(0.208)
Semester	0.874***	1.096***	1.053***	1.046**
	(0.009)	(0.013)	(0.014)	(0.016)
Weekend	0.079***	17.144***	16.975***	10.545***
	(0.003)	(0.848)	(0.934)	(0.592)
Constant	82.930***	0.005***	0.004***	0.006***
	(12.56)	(0.001)	(0.001)	(0.001)
Random Effects
Level 3 Variance	2.260***	3.080***	3.904***	2.995***
	(0.573)	(0.250)	(0.342)	(0.268)
Level 2 Variance	0.167***	0.515***	0.539***	0.894***
	(0.042)	(0.049)	(0.056)	(0.087)
N(days)	55829	55829	55810	55762

⁺p<.10,

^*p<.05,

^**p<.01,

^***p<.001

Notes. Exponentiated coefficients (>1 indicates positive association, <1 indicates negative association); Standard errors in parentheses. Reference category for generational status is 3^rd generation and race-ethnicity is white. Ln alpha is 1.537(.040) for negative consequences.

Next, models tested whether the alcohol use patterns of the 1^st, 2^nd, and 3^rd generations converged over time (aim 2) or varied for weekends versus weekdays (aim 3). The results demonstrated that differences between generations were maintained across college; the interactions with time were not significant for any of the four outcomes. Thus, the protective effects of being an immigrant did not change as students progressed through college (results not shown, but available on request).

Furthermore, for abstaining and negative consequences, the impact of generational status varied by day of the week. All students were more likely to abstain from alcohol use on weekdays than weekends, but the weekend effect on drinking (versus not) was larger for 3^rd and 2^nd generation students than for 1^st generation students. The marginal predictive mean difference for weekend versus weekday between the three groups were −.225(.009), −.202(.010), and −.133 (.012), respectively (see Figure 1). Likewise, regardless of generational status, negative consequences were higher on weekend days. As with abstaining, the predicted mean difference for weekend versus weekday was larger for 3^rd and 2^nd generation than for 1^st generation students (M_contrast=.479 (.080), .393(.073), and .202(.049), respectively; see Figure 2).

Figure 1.

Predicted probability of abstaining on a given weekday versus weekend day by generational status (N=55,829 days).

Figure 2.

Predicted negative consequences on a given weekday versus weekend day by generational status (N=55,762 days).

3.1. Subgroup Analyses of Ethnic Groups

Like the full sample models, models limited to Latinx and Asian Americans showed strong associations between being foreign-born and alcohol use outcomes (Table 3). Also similar to the full sample, subgroup analyses demonstrated that the protective effect of being foreign born was consistent across time (i.e., no Foreign-born × Semester interactions) and race-ethnicity (i.e., no evidence of Foreign-born × Race-ethnicity interactions). Exploratory within-group analyses demonstrated that, for Asian American students, being foreign-born was protective across all outcomes (i.e., foreign-born Asian Americans were more likely to abstain and less likely to binge drink, want to get drunk, and experience negative consequences than US-born Asians). However, the associations were not statistically significant for Latinx students, although there was a trend suggesting that foreign-born Latinx were less likely to report wanting to get drunk than US-born Latinx. In addition, the effect of being foreign-born was consistent across semesters for both groups (i.e., no significant Foreign-born × Semester interactions for Latinx or Asians, consistent with main analyses). Additional exploratory analyses demonstrated that associations between foreign-born status and alcohol use were generally similar for men and women. Only one significant interaction of gender and foreign birthplace was observed: among Latinx, being foreign born predicted fewer negative consequences for men, but not women.

Table 3.

Foreign Birthplace and Alcohol Use among Latinx and Asian American College Students

	Abstain	Binge Drink	Wanted to get Drunk	Negative Consequences
	OR	OR	OR	IRR
Fixed Effects
Foreign born	1.892**	0.505**	0.351***	0.507**
	(0.375)	(0.125)	(0.104)	(0.133)
Asian American	2.305***	0.350***	0.287***	0.369***
	(0.402)	(0.075)	(0.073)	(0.084)
Male	1.488*	0.625*	0.682	0.743
	(0.253)	(0.131)	(0.167)	(0.165)
Greek	0.360***	3.981***	4.388***	3.291***
	(0.072)	(0.962)	(1.229)	(0.841)
Semester	0.884***	1.092***	1.047*	1.045+
	(0.013)	(0.019)	(0.020)	(0.025)
Weekend	0.086***	15.014***	14.362***	10.348***
	(0.005)	(1.052)	(1.110)	(0.872)
Constant	62.917***	0.006***	0.004***	0.006***
	(10.834)	(0.001)	(0.001)	(0.001)
Random Effects
Level 3 Variance	1.921***	2.778***	3.725***	2.817***
	(.208)	(.326)	(.474)	(.368)
Level 2 Variance	.589***	.550***	.479***	1.005***
	(0.060)	(0.073)	(0.079)	(0.141)
N(days)	27528	27528	27521	27488

⁺p<.10,

^*p<.05,

^**p<.01,

^***p<.001

Note. Exponentiated coefficients; Standard errors in parentheses. Reference groups are US born, Latinx, females, and weekdays. Ln alpha is 1.679(.058) for negative consequences.

4. Discussion

Using intensive diary data from across seven college semesters, the current study examined whether generational status was linked to alcohol use and alcohol-related consequences. Our results demonstrated a clear protective effect of being an immigrant. This effect was consistent across different alcohol use dimensions including any drinking, binge drinking, subjective plans to get drunk, and experiencing alcohol-related consequences. Furthermore, students sustained these differences across college.

The nature of the university’s drinking culture gave us strong reason to expect that any generational differences in alcohol use and consequences might decrease and ultimately disappear over time. Previous research has linked acculturation factors to drinking motives and more liberal drinking norms, and in turn alcohol use (Mills & Caetano, 2012). Yet, even though all students were exposed to a heavy drinking environment, we found no evidence that the protective effect of 1^st generation status deteriorated over time. As they progressed through college, 1^st generation students continued to drink less, binge drink less, and experience fewer physical and behavioral consequences due to alcohol. Notably, all students were US citizens or permanent residents living in the US prior to college; those who moved to the US on student visas were not included in the sample. Much prior work exploring alcohol use among college students has not considered visa status and, therefore, potentially combined international students on temporary visas with immigrants who moved to the US to live permanently. Future work should disaggregate international and immigrant students because their alcohol use patterns may differ.

The findings of generational differences in alcohol use were largely consistent whether we focused on all students or the subsample of Latinx and Asian American students, supporting our conclusions. Although we did not find significant interactions between race/ethnicity and birthplace, the within-group analyses suggested that foreign-born status may be especially protective for Asian American students. Our findings are in line with prior research documenting higher risk of alcohol abuse and dependence among US-born Asians compared to foreign-born Asians (Breslau & Chang, 2006). Thus, it is imperative that research on nativity status and alcohol use among Asian Americans continues, especially given that this group has been understudied in the literature (Iwamoto, Grivel, Cheng, & Zamboanga, 2016). Special attention should be focused on US-born Asians who may be at heightened risk for heavy alcohol use and alcohol-related consequences (Iwamoto, Kaya, Grivel, & Clinton, 2016). Of course, the small sample sizes make our within-group results exploratory; replication is necessary. There remains an important need to explore within-race alcohol use trajectories to better understand the role of foreign birth, whether it changes longitudinally during college, and whether it varies for female and male students. Indeed, even within Latinx and Asian American populations, substance use and proximal correlates have been shown to vary by national group/region of origin (Lum, Corliss, Mays, Cochran, & Lui, 2009; Ramisetty-Mikler, Caetano, & Rodriguez, 2010; Shih et al., 2015). Thus, future research should continue to examine whether country of origin shapes alcohol consumption among immigrants and children of immigrants.

Although we do not know the underlying mechanisms responsible for sustained generational differences in alcohol outcomes, previous research has highlighted the role of pre-migration experiences - particularly characteristics of the drinking culture in the heritage country (Cook, Bond, Karriker-Jaffe, & Zemore, 2013) as an important predictor of alcohol use among the foreign-born. Furthermore, 1^st generation students may leverage protective factors such as familism or collectivism values or adherence to heritage cultural traditions enabling them to withstand the risks of the college environment and demonstrate healthier alcohol use patterns (Schwartz et al., 2011). Supporting this idea, one longitudinal study of Latinx adolescents in the US found that Hispanic cultural orientation protected against increases in alcohol use over time (Unger, Schwartz, Huh, Soto, & Baezconde-Garbanati, 2014).

Importantly, our results also demonstrated that the protective effect of being 1^st generation differed across weekends versus weekdays for abstaining and experienced negative consequences. Weekends are when young people are most likely to drink alcohol and experience negative consequences (Elder et al., 2004; Maggs et al., 2011). The current study added to the literature by demonstrating that the weekend was riskier for everyone, immigrants and non-immigrants alike. However, for abstaining and negative consequences, this weekend effect was larger for 3^rd generation and 2^nd generation students. On weekdays, the differences between generations were small as few students drank or experienced negative consequences. The weekends exacerbated risk more for 3^rd generation students than 1^st generation students.

Important limitations of the study must be noted. Data came from one university and thus the extent to which results generalize is unknown. Furthermore, although the data contain questions about birthplace, they do not include information about pre-migration experiences, timing of migration, or heritage cultural values/practices. Thus, we could not capture the multi-dimensional nature of the acculturation process (Schwartz, Unger, Zamboanga, & Szapocznik, 2010), test the possible protective role of heritage values, or differentiate between students who immigrated to the US as young children and those who immigrated as adolescents. Future research could use longitudinal designs to understand whether these characteristics shape changes in alcohol use across the college years. Examining social characteristics of the university environment - as well as the cultural congruity between the individual and the college environment (Cano et al., 2015) - might prove fruitful for future research as well.

The current study has notable strengths. The ecologically valid repeated daily diary design provided impressively detailed measurement across over 55,000 days, among a sample diverse with regard to generational status and race-ethnicity. Furthermore, the longitudinal data enabled us to document that the impact of generational status was stable through the fall of students’ fourth years in college, although it mattered more on weekends, at least for abstaining and negative consequences. To our knowledge, this is the first study to systematically examine the alcohol use of immigrant college students in the US using a longitudinal daily diary design. Future research should continue to explore the temporal variation in alcohol use and consequences among diverse samples of college students as well as the mechanisms that help to explain differences across generations.

Table 2.

Alcohol Use Predicted by Interactions between Generational Status and Weekend: Adjusted Odds Ratios and Incident Rate Ratios

	Abstain	Binge Drink	Wanted to get Drunk	Negative Consequences
	OR	OR	OR	IRR
Fixed Effects
Generational status
2nd generation	1.137	0.961	0.801	1.262
	(0.211)	(0.217)	(0.204)	(0.292)
1st generation	1.569+	0.453**	0.290***	0.347**
	(0.374)	(0.138)	(0.103)	(0.114)
Weekend	0.078***	17.620***	17.07***	10.576***
	(0.004)	(1.164)	(1.241)	(0.796)
Interactions
2nd Generation	0.944	0.919	0.965	0.871
×Weekend	(0.075)	(0.095)	(0.111)	(0.099)
1st Generation	1.266*	1.027	1.093	1.776**
×Weekend	(0.146)	(0.175)	(0.223)	(0.376)
Random Effects
Level 3 Variance	2.260***	3.080***	3.904***	2.997***
	(0.167)	(0.249)	(0.342)	(0.268)
Level 2 Variance	.574***	.515***	.539***	.898***
	(0.042)	(0.049)	(0.056)	(0.087)
N(days)	55829	55829	55810	55762

⁺p<.10,

^*p<.05,

^**p<.01,

^***p<.001

Notes. Exponentiated coefficients (>1 indicates positive association, <1 indicates negative association); Standard errors in parentheses. Reference category for generational status is 3^rd generation.

Models control for gender, Greek, race-ethnicity and semester in college. Ln alpha is 1.534(.040) for negative consequences.