5-HTTLPR,Suicidal Behavior by Others,Depression,and Criminal Behavior During Adolescence

Abstract

Vicarious strains like suicidal behavior on the part of others have been shown to be predictive of both negative emotions and antisocial behavior during adolescence. Little research to date, however, has examined the role that biological factors play in moderating these relationships. Using a sample of adolescents drawn from the National Longitudinal Survey of Adolescent Health (N = 7,995), and drawing on two separate, but related, theories, I explore whether the serotonin transporter gene (5-HTTLPR) interacts with suicidal behavior by others to affect depression and self-reported crime. Results of ordinary least squares and negative binomial regressions reveal that suicide by others interacts with 5-HTTLPR to increase both depression and crime for males but not females, net of controls. Thus, 5-HTTLPR may be implicated in shaping negative emotions and antisocial behavior among males during adolescence by moderating the effects of suicide by others. Implications for theory are discussed.

Keywords

5-HTTLPR suicide biosocial theory general strain theory depression crime

Suicidal behavior by others has been shown to have effects on both the emotional states and behavior of those close to the person who attempted or completed suicide (Brent et al., 1995; Burton, Foy, Bwanausi, Johnson, & Moore, 1994; Cerel, Fristad, Weller, & Weller, 2000; Pfeffer et al., 1997; Schofield & Ratnarajah, 2007; Shepherd & Barraclough, 1976; Simone, 2008; Wilcox et al., 2010). To date, however, no research has offered or tested a theoretical rationale for why suicide on the part of others should shape emotional states and behavior. I argue that a popular theory in the literature on crime and delinquency, general strain theory (GST), offers an explanation for past findings.

Robert Agnew’s (1992, 2006) GST is one of the more popular theories of crime and delinquency. Agnew’s (1992) theory argues that negative stimuli or “strains” affect individuals by producing negative emotions, most notably anger and depression, which can then in turn increase delinquent behavior. Importantly, these strains need not be direct, as the vicarious experience of stress and trauma can also affect one’s emotions and behavior (Agnew, 2002, 2006).

In the present study, I argue that GST offers a theoretical model to explain the relationship between suicide by others, negative emotions, and antisocial behavior. In addition, I go past this model to show how these relationships are moderated by genetics. Recent research (Caspi et al., 2003; Simons et al., 2011) has already shown that the effects of stressful life events on emotional states and behavior are conditioned by genetics, but without the theoretical rationale provided in this study. I test my theoretical model using data from the National Longitudinal Study of Adolescent Health (Add Health). In so doing, this study advances the literatures on trauma, negative emotions, and theories of crime and delinquency. In the following sections, I will explicate my theoretical model more fully and describe the data set and methods in more detail.

Suicide by Others as Strain, Negative Emotions, and Crime

GST was proposed by Robert Agnew (1992) as an individual level, social-psychological explanation of crime and delinquency. The theory is an expansion of traditional anomie/strain theories, which focus on an individual’s ability to achieve culturally valued economic and class-based status goals, and the stress and deviance caused when individuals do not reach these goals (Cloward & Ohlin, 1960; Cohen, 1955; Merton, 1938). Agnew introduced two other types of strain likely to lead to deviant coping: the loss of valued persons or objects and the presentation of noxious (negative) stimuli. Importantly, Agnew has theorized that experiences of strain need not be direct to shape behavior, as the vicarious experience of stress, trauma, loss, and so forth can lead to deviant coping in much the same way as direct strain (Agnew, 2002).

I argue that suicidal behavior by others should be considered an important vicarious strain. Suicides and suicide attempts are often unexpected, traumatic, and violent (Schofield & Ratnarajah, 2007). Suicidal behavior by friends and family thus captures both the threatened or actual loss of valued persons and the presentation of noxious stimuli. Past research and anecdotal evidence have demonstrated the validity of friends and family suicidal behavior as a measure of vicarious strain. For example, the literature on the aftermath of suicide has shown that experiencing the death by suicide of a parent or close peer during adolescence can have various consequences for psychological functioning, most significant among them being higher risks for depression and posttraumatic stress disorder (Brent et al., 1995; Burton et al., 1994; Schofield & Ratnarajah, 2007; Shepherd & Barraclough, 1976; Simone, 2008; Wilcox et al., 2010). In addition, studies have shown that experiencing the suicide of a loved one in childhood or adolescence, most especially a parent, can have long-term consequences for antisocial behavior, including increased aggression and substance use (Cerel et al., 2000; Pfeffer et al., 1997; Wilcox et al., 2010).

While the nature of the link between the suicidal behavior of parents and friends and negative emotional states like depression seems fairly obvious, the link between suicide by others and antisocial behavior in adolescence is less obvious. The reason suicidal behavior by friends and family can be consequential for offending has to do with its timing. Experiencing this type of vicarious strain during childhood or adolescence makes it consequential for antisocial behavior. Children and adolescents who experience the actual or attempted suicide of a friend or family member may internalize the experience if others surrounding the bereaved don’t specifically engage the individual in order to guide them through the bereavement process (Schofield & Ratnarajah, 2007). In addition, experiencing the suicide or suicide attempt of a loved one puts strains on one’s personal relationships and can lead to interpersonal conflict (Cerel et al., 2000). Overall, the literature suggests that antisocial behavior is one part of an overarching psychopathology that can develop after experiencing suicidal behavior by friends and family during adolescence, most especially if bereavement care is not readily available (Cerel et al., 2000; Schofield & Ratnarajah, 2007; Wilcox et al., 2010). Suicidal behavior by friends and family should thus be considered a vicarious strain worthy of study that can affect both negative emotions and antisocial behavior during adolescence, and GST offers a theoretical model for why this should be.

I also explore another important issue in the present study. Almost 20 years of research have provided extensive empirical support for Agnew’s theory (Agnew, 2002; Langton & Piquero, 2007; Mazerolle, Piquero, & Capowich, 2003; Piquero & Sealock, 2004). However, a key proposition in GST, that the strain-negative emotions-crime relationship might be conditioned by outside factors (Agnew, 2006), has been infrequently tested. In addition, when this proposition has been tested, the focus has been on sociological variables (Jang & Johnson, 2003, 2005), and there has thus far been little effort to integrate genetics as a potential conditioning factor into the strain-negative emotions-crime relationship.

In addition to arguing that a traditional GST model explains the relationship between suicide by others and negative emotions and antisocial behavior, I also argue in the present study that genetics is a factor that could condition this model. Could it be that some individuals, because of their genetic makeup, are more likely to react to suicidal behavior on the part of others by developing negative emotions and engaging in deviant coping responses to these emotions? I will now turn to a discussion of the issue of moderation of environmental effects (in this case, a vicarious strain) on psychological functioning and behavior by genes.

Gene-Environment Interactions: 5-HTTLPR, Strain, Depression, and Antisocial Behavior

Research exploring the genetic basis of neuropsychological functioning and antisocial behavior has tended to focus on variations in genes responsible for regulating neurotransmitters, such as serotonin and dopamine. One gene that has received extensive attention in this literature is the serotonin-transporter-linked polymorphic region (5-HTTLPR).

5-HTTLPR is a degenerate repeat polymorphic region in SLC6A4, which is the gene that codes for the transportation of serotonin. This makes 5-HTTLPR an important gene in studies of depression and antisocial behavior, because serotonin is thought to be linked to feelings of happiness and well-being (Belsky & Pluess, 2009). A polymorphism in the promoter region of 5-HTTLPR means there are two variants, the short (s) and long (l) alleles. The s-allele has been associated with reduced expression of the serotonin transporter molecule, and those with two copies of the s-allele (s/s) have been shown by prior research to be the most vulnerable to displaying psychological and behavioral issues when having to deal with adversity (Belsky & Pluess, 2009; Caspi et al., 2003).

Recent research has shown that 5-HTTLPR moderates the effects of environmental stressors on both depression and antisocial behaviors. In a groundbreaking study, Caspi et al. (2003) showed that 5-HTTLPR moderated the effects of stressful life events in early adulthood on depressive symptoms and the probability of suicidal ideation and attempts and episodes of major depression at 26 years of age. Those homozygous for the s-allele (s/s) proved the most adversely affected, while l-allele homozygotes (l/l) showed greater resilience, with little to no effect of stressful life events on depression among this group. Taylor et al. (2006) reported similar results in a sample of young adults, finding the same pattern when observing the effect of problematic child-rearing history and recent negative life events on depressive symptomatology. A similar pattern involving 5-HTTLPR and depression emerged in Eley et al.’s (2004) study on adolescent girls who were and were not exposed to risky family environments and in Brummett et al.’s (2008) investigation of middle-aged and aging adults who had and hadn’t served as the primary caregiver of a relative with Alzheimer’s disease.

While research on 5-HTTLPR has tended to focus on depression, several investigators have recently examined its association with antisocial behaviors. Similar to the research on depression, this line of studies has found that 5-HTTLPR moderates the effect of environmental stressors on aggressive and antisocial behaviors. For example, recent studies have shown that carriers of the s-allele are more at risk of aggression and violent criminal behavior when presented with an adverse social environment (Reif et al., 2007; Retz et al., 2008; Vaughn et al., 2009; Verona, Joiner, Johnson, & Bender, 2006). Most recently, Simons et al. (2011, 2012) showed that 5-HTTLPR (along with other genes) interacted with a composite measure of favorable and adverse environmental factors to effect anger, orientations toward hostility, and aggression.

In summation, recent research has shown that carriers of the s-allele of 5-HTTLPR, but most especially those homozygous for the s-allele, are at greater risk of depression and antisocial behavior when presented with environmental stressors. But how do genes cause some individuals to be at risk when presented with negative stimuli like poor parenting, victimization, or suicidal behavior on the part of others? Recent theorizing by Belsky and his colleagues (Belsky, 1997, 2005; Belsky & Pluess, 2009; Belsky et al., 2009) have argued that this is because certain forms of a gene make individuals more susceptible to environmental influence. Belsky and Pluess (2009) observed that the genes most often studied in the GxE literature are involved in the serotonergic and dopaminergic systems. 5-HTTLPR is directly linked to serotonin activity in the brain, and the serotonergic system is implicated in sensitivity to punishment and displeasure. This means that certain individuals are more responsive to their environment because of their different thresholds for experiencing pleasure and displeasure. Essentially, some people are, because of their genetic makeups, more readily shaped by their environment than others. Thus, the person homozygous for the s-allele of 5-HTTLPR may be more likely than the person who is not to experience depression, and thus perhaps engage in more coping behaviors, whether legal or illegal, when they experience a vicarious strain like suicidal behavior on the part of others.

The Present Study

The present study extends the literatures on the effects of suicide by others and criminological theory in two important ways. First, I offer up and test a theoretical model (GST) to explain the relationship between suicidal behavior by others and negative emotions and antisocial behavior. Second, I integrate a GxE approach into this theoretical model. All three of the key elements present in GST, (vicarious) strain, negative emotions, and criminal offending have been present in the GxE literature, and they have all been specifically tied to 5-HTTLPR. This makes the integration of these two literatures a logical step and the testing of a combined model is a natural extension of prior theorizing and research.

Based on prior theorizing and research in the literature on suicidal behavior by others and its effects and the GxE literature, I expect that 5-HTTLPR genotype will condition the theoretical model of strain-negative emotions-crime that is identified in GST. Specifically, I expect that those who are homozygous for the s-allele will experience both more depression and engage in more offending after experiencing the death or near death of a friend or family member to suicide than will individuals with some other allelic combination of 5-HTTLPR (l/l, s/l, l/s). In addition, I expect that the suicidal behavior by others × 5-HTTLPR effect on offending behavior will be mediated by depressive symptoms, per the expectations of GST. The suicidal behavior by others × 5-HTTLPR effect on offending will thus act as a mediated moderator model that combines a GST model with previous findings in the GxE literature, with depression acting as the mediating variable.

I draw on these expectations to derive specific hypotheses about the relationship between 5-HTTLPR, suicidal behavior by others, depression, and criminal offending. Hypothesis 1 predicts that suicidal behavior by others will have a significant and positive effect on depressive symptoms. Hypothesis 2 predicts that suicidal behavior by others will have a significant and positive effect on criminal offending. Drawing on past research on 5-HTTLPR, Hypothesis 3 predicts that 5-HTTLPR will moderate the effect of suicidal behavior by others on depressive symptoms such that those who are homozygous for the s-allele will experience more depressive symptoms than non-s-allele homozygotes. Hypothesis 4 predicts that 5-HTTLPR will moderate the effect of suicidal behavior by others on criminal offending such that those who are homozygous for the s-allele will engage in more criminal offending than non-s-allele homozygotes. Finally, drawing on GST, Hypothesis 5 predicts that depressive symptoms will significantly mediate the effect that the interaction between 5-HTTLPR and friends and family suicidal behavior has on criminal offending.

Data and Method

Data

For the analyses presented below, I draw on data from the National Longitudinal Study of Adolescent Health (Add Health). Add Health is a nationally representative sample of American adolescents who were first recruited during the 1994-1995 school year while they were in Grades 7 to 12 (Harris et al., 2003; Udry, 2003). Add Health obtained a nationally representative sample of adolescents by utilizing a multistage stratified sampling process to select 80 high schools and 52 middle and junior high schools for inclusion in the study. More than 90,000 students completed in-school self-report surveys, and of this group a subsample was randomly chosen for the Wave I in-home component of Add Health. Ultimately, 20,745 adolescents and 17,700 of their primary caregivers participated in the Wave I in-home component of Add Health (Harris et al., 2003). Wave II data collection occurred approximately 1 to 2 years later, Wave III data were collected during 2001-2002, and Wave IV data were collected during 2007-2008.

During Wave IV in-home interviews, Add Health collected biological data from respondents. Among the data collected, Add Health took saliva swabs for DNA analysis. In conjunction with the Institute for Behavioral Genetics (IBG) in Boulder, Colorado, Add Health genotyped Wave IV interviewees for a set of genetic markers of interest to biosocial researchers. That Add Health is a large and nationally representative data set that contains variables measuring both the social environment and genetics makes it highly desirable for the present study. The present study includes Add Health respondents who were interviewed at Waves I, II, and IV who did not have any missing genetic data. After dealing with missing data, the present study includes the analysis of information gathered from 7,995 respondents in the Add Health data set, 3,581 males and 4,414 females.

Measures

Criminal behavior

The key dependent variable consists of nine items drawn from Wave II that asked respondents about various criminal activities they have engaged in during the prior year. These items are a mixture of both violent and property offending. Measures of violence included questions asking how often respondents used or threatened to use a weapon on someone to get something from them, pulled or actually used a knife or gun on someone, used a weapon during a fight, or hurt someone so badly in a fight that they needed medical attention. Property offending measures included asking respondents how often they deliberately damaged other’s property, stole cars, burglarized buildings, and stole items worth more than US$50. Inspection of the distributions for each item revealed extensive skew, with relatively few respondents reporting frequent involvement. I thus recoded the nine items into binary measures of whether the respondent reported engaging in these behaviors in the past year (1 = yes), and summed them into a measure of criminal behavior that emphasizes the prevalence of offending (α = .70). This type of scaling has been recommended by researchers in the past (Hindelang, Hirschi, & Weis, 1981; Osgood, McMorris, & Potenza, 2002), and this particular scale is closely related to others that researchers have developed for use in the Add Health data set in past studies (Guo, Roettger, & Cai, 2008; Hagan & Foster, 2003; Haynie, 2001, 2003; Kaufman, 2009).

Suicidal behavior by friends and family

Suicidal behavior by friends and family combines four items that asked respondents if any friends or family members tried or succeeded in killing themselves during the past 12 months. While the original questions asked separately whether respondents had friends or family who had attempted or completed suicide, very few respondents had a friend or family member that had attempted or completed suicide, so all four measures were combined into one measure. The result is a 0/1 item reflecting whether a respondent had neither a friend or family member attempt or complete suicide (0) or either a friend or family member or both a friend and family member attempt or complete suicide (1) in the past 12 months. A total of 1,609 of the 7,995 respondents scored a 1 on this measure.

Depressive symptoms

I measure depressive symptoms with 19 items from the Center for Epidemiologic Studies–Depression (CES-D) Scale, administered during Wave I interviews (range = 0-57; α = .85). This scale includes items that ask about feelings of loneliness, fear, sadness, fearfulness, and depression in the week before being interviewed, with possible responses ranging from 0 (never or rarely) to 3 (most of the time or all the time). The use of depressive symptoms in the present study is consistent with the literature on the effects of suicidal behavior by others and with recent theorizing and research on GST (Brezina, 1996; Broidy, 2001; Jang & Johnson, 2003; Kaufman, 2009; Moon et al., 2009), as well as prior research on the combination of 5-HTTLPR and stressful life events (Carver, Johnson, Joormann, LeMoult, & Cuccaro, 2011; Caspi et al., 2003; Goldman, Glei, Lin, & Weinstein, 2010).

5-HTTLPR

The genotype at 5-HTTLPR located on chromosome 17q11.1-q12 has a functional polymorphism in the variable repeat sequence in the promoter region (Bradley, Dodelzon, Sandhu, & Philibert, 2005; Simons et al., 2011). Prior research on 5-HTTLPR has focused on three variants of the gene: those homozygous for the long allele (l/l), those carrying one short allele (s/l, l/s), and those homozygous for the short allele (s/s).

Those homozygous for the short allele have been shown by past research to be the most reactive to environmental stressors (Caspi et al., 2003; Manuck, Flory, Ferrell, & Mulddon, 2004). Accordingly, I coded the 5-HTTLPR variable to reflect the nonpresence (0) or presence (1) of two s-alleles. Based on this coding the distribution for 5-HTTLPR was such that 71.6% of respondents were not s-allele homozygotes (l/l, s/l, l/s), and 28.4% were homozygous for the s-allele. The Hardy-Weinberg equilibrium test showed that the distribution of 5-HTTLPR did not differ significantly from that predicted on the basis of simple Mendelian inheritance.

Controls

I also include several general controls in all analyses that have been shown to correlate with involvement in crime: Age; dummy variables for Hispanic, non-Hispanic Black, Native American, Asian, and Other (with non-Hispanic White as the reference category); parent’s education (1 = 4 year degree or more); and parent receiving public assistance (1 = yes). About 55% of the sample is non-Hispanic White, 14.7% Hispanic, 20% non-Hispanic Black, with the remainder comprising Native American (2.5%), Asian (6.8%), and members of Other racial/ethnic groups (1.1%). About 23.4% of parents are college graduates, while 7.3% report receiving public assistance. I utilized multiple imputation techniques to fill in missing values on the parental education and public assistance measures.

Analytic Strategy

I test my hypotheses using ordinary least squares (OLS) and negative binomial (NB) regression techniques where appropriate. I utilize NB regression techniques when looking at the key dependent variable of interest because it is a highly skewed count measure, thus violating the assumption of normality required for OLS regression (Gardner, Mulvey, & Shaw, 1995). I utilize the appropriate weight, cluster, and strata variables in all analyses to account for the complex Add Health survey design. Tests using variance inflation factors (VIFs) showed that multicollinearity was not a problem in any of the equations. I present separate models for males and females rather than just controlling for gender, as the GST literature has argued that strains may affect males and females differently (Broidy & Agnew, 1997; Jang, 2007).

Results

Table 1 presents the descriptive statistics for the study variables for males and females separately, with mean comparisons for the two groups. Levels of criminal offending are very low for both males and females. Not surprisingly, males on average report more criminal behavior than do females. Females were significantly more likely than males to report that a friend or family member or both had attempted or completed suicide in the past year. While about 25% of females said this had occurred, only about 14% of males had experienced this vicarious strain. Females on average reported more depressive symptoms in the week prior to Wave I interviews than did males. Among the controls, there are significantly more Black females than Black males in the sample, and there are significantly more Asian males than females in the sample. Males were on average slightly older than females at Wave II, and females were slightly more likely to have a target parent receiving public assistance at Wave I.

Table 1.

Descriptive Statistics and Mean Comparisons by Gender.

		Males	Females
		(n = 3,581)	(n = 4,414)
Variables	Range	M (SE)	M (SE)
Dependent variable
Criminal behavior W2	0-9	0.54 (0.03)	0.27 (0.02)**
Independent variable
Friends and family suicidal behavior	0/1	0.14 (0.01)	0.25 (0.01)**
5-HTTLPR genotype	0/1	0.28 (0.01)	0.29 (0.01)
Mediating variable
Depressive symptoms	0-51	9.94 (0.11)	11.86 (0.12)**
Controls
White	0/1	0.56 (0.01)	0.54 (0.01)
Hispanic	0/1	0.15 (0.01)	0.15 (0.01)
Black	0/1	0.18 (0.01)	0.22 (0.01)**
Native American	0/1	0.03 (0.00)	0.02 (0.00)
Asian	0/1	0.08 (0.00)	0.06 (0.00)**
Other	0/1	0.01 (0.00)	0.01 (0.00)
Age W2	11-21	16.11 (0.03)	15.93 (0.02)**
Parent’s education	0/1	0.27 (0.01)	0.26 (0.01)
Parent receiving public assistance	0/1	0.07 (0.00)	0.09 (0.00)**

Note. Because these statistics are weighted and adjusted for survey design, standard errors are produced rather than standard deviations. W2 = Wave II.

p < .01, mean one-way ANOVAs denote significant gender comparisons.

Table 2 presents the correlation matrix for the study variables, along with means and standard errors for the full sample. The correlation matrix serves to check for gene-environment correlations (rGEs). rGE refers to a nonrandom distribution of environments among different genotypes (Simons et al., 2011). These rGEs can potentially confound GxE effects (Guo et al., 2008; Guo, Tong, & Cai, 2008). Table 2 shows that there is not a significant correlation between suicidal behavior by friends and family and 5-HTTLPR genotype. This suggests an absence of rGE effects in these data.¹ As expected, suicidal behavior by friends and family is significantly and positively related to depressive symptoms at Wave I and criminal behavior at Wave II. 5-HTTLPR genotype does not significantly correlate with criminal behavior. As expected, the Depressive Symptoms Scale is significantly and positively correlated with criminal behavior. The only correlation with 5-HTTLPR genotype to note is a significant, positive, but extremely weak correlation with depression symptoms.

Table 2.

Correlation Matrix for the Study Variables for the Full Sample (N = 7,995).

		1	2	3	4	5	6	7	8	9	10	11	12	13	14
1.	Criminal behavior W2	X
2.	Friends and family suicidal behavior	.10**	X
3.	5-HTTLPR genotype (1 = s/s)	.00	.00	X
4.	Depressive symptoms	.12**	.18**	.03*	X
5.	Male	.16**	−.13**	.00	−.13**	X
6.	White	−.03**	.04**	−.08**	−.13**	.02	X
7.	Hispanic	.04**	.01	.05**	.07**	.01	−.46**	X
8.	Black	.00	−.06**	−.07**	.04**	−.05**	−.55**	−.21**	X
9.	Native American	.04**	.04**	−.01	.02*	.00	−.18**	−.07**	−.08**	X
10.	Asian	−.02*	−.02*	.21**	.07**	.04**	−.30**	−.11**	−.13**	−.04**	X
11.	Other	.02	−.01	.02	.01	.00	−.12**	−.04**	−.05**	−.02	−.03*	X
12.	Age W2	−.04**	−.01	.03*	.13**	.06**	−.07**	.07**	−.01	−.01	.07**	.00	X
13.	Parent’s education	−.02	−.01	.00	−.09**	.02	.03**	−.15**	.01	−.02	.14**	.01	−.03**	X
14.	Parent receiving public assistance	.01	.00	.00	.06**	−.04**	−.13**	.06**	.13**	.03*	−.04**	.00	−.02	−.15**	X
M		.38	.20	.28	11.00	.45	.55	.15	.20	.03	.07	.01	16.01	.26	.08
SE		.01	.00	.01	.08	.01	.01	.00	.00	.00	.00	.00	.02	.00	.00

Note. W2 = Wave II.

p < .05. **p < .01.

To assess the possibility that suicidal behavior by friends and family alone and in combination with 5-HTTLPR genotype effects depression, Table 3 presents OLS regression models with Wave I depressive symptoms regressed on friends and family suicidal behavior, 5-HTTLPR genotype, and the controls separately for males and females. Model 1 examines the direct effects of friends and family suicidal behavior and 5-HTTLPR genotype on depressive symptoms. This model shows that suicidal behavior by friends and family has a highly significant and positive effect on depressive symptoms among males and females, supporting Hypothesis 1. 5-HTTLPR genotype does not exert a direct effect on depressive symptoms.

Table 3.

Depressive Symptoms Regressed on Friends and Family Suicidal Behavior, 5-HTTLPR Genotype, and Controls.

	Model 1		Model 2
	Males	Females	Males	Females
Variables	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Environment and genetic variables
Friends and family suicidal behavior	2.77 (0.50)**	3.10 (.33)**	2.22 (.52)**	3.14 (.40)**
5-HTTLPR (1 = s/s)	0.33 (0.29)	0.25 (0.33)	0.01 (0.31)	0.29 (0.34)
Two-way interaction
Friends and family suicidal			1.99 (.99)*	−0.16 (0.83)
Behavior × 5-HTTLPR
Controls
Hispanic	1.30 (0.39)**	1.87 (0.58)**	1.33 (0.39)**	1.87 (0.57)**
Black	1.41 (0.43)**	1.75 (0.50)**	1.43 (0.43)**	1.75 (0.49)**
Native American	1.74 (1.08)	1.95 (1.27)	1.74 (1.08)	1.95 (1.28)
Asian	3.25 (0.64)**	2.35 (0.98)*	3.27 (0.64)**	2.35 (0.98)*
Other	2.50 (1.36)	−0.56 (1.22)	2.48 (1.33)	−0.57 (1.22)
Parent’s education	−1.18 (0.28)**	−1.35 (0.35)**	−1.17 (0.27)**	−1.35 (0.35)**
Parent receiving public assistance	1.06 (0.51)*	1.83 (0.61)**	1.05 (0.52)*	1.83 (0.61)**
Age W1	0.57 (.09)**	0.71 (0.10)**	0.57 (0.09)**	0.71 (0.10)**
Constant	0.34 (1.41)	−0.17 (1.45)	0.44 (1.42)	−0.17 (1.45)
R ²	.07	.08	.07	.08

Note. Non-Hispanic White is the reference category for all race/ethnic groups. This table includes unstandardized coefficients (linearized standard errors) from OLS models. OLS = ordinary least squares; W1 = Wave I.

p < .05. **p < .01.

Model 2 incorporates the interaction between friends and family suicidal behavior and 5-HTTLPR genotype into the model. The interaction term is significant and positive for males only, suggesting that compared with males who are not homozygous for the s-allele of 5-HTTLPR, those who are experience more depressive symptoms when confronted with this vicarious strain, supporting Hypothesis 3 among this group. The direct effect of friends and family suicidal behavior remains highly significant in this model for both males and females. These results suggest that experiencing the death or near death of a friend, family member, or both due to suicide negatively affects adolescent males and females. However, the effect is almost twice as big for those males who are homozygous for the s-allele (from Model 2 for males in Table 3, the coefficient for friends and family suicidal behavior plus the interaction term coefficient) compared with males who are not homozygous for the s-allele. Among the controls, there are significant contrasts among different racial groups in both models for both males and females, with Hispanics, Blacks, and Asians reporting higher levels of depressive symptoms than non-Hispanic Whites. In addition, males and females whose primary caregiver reported having a college degree report lower levels of depression, males and females whose primary caregiver reported being on public assistance at Wave I report more depression, and older respondents from both groups report more depression.

Turning to criminal behavior, Table 4 presents NB regression models with Wave II criminal behavior regressed on friends and family suicidal behavior, 5-HTTLPR genotype, and controls separately for males and females. The baseline model in Model 1 shows that suicidal behavior by friends and family has a significant and positive effect on criminal behavior for males and females, supporting Hypothesis 2. Like with depression, 5-HTTLPR genotype does not affect criminal behavior for either group. Model 2 introduces the GxE term, and like with depression this interaction of friends and family suicidal behavior and 5-HTTLPR genotype has a significant and positive effect on criminal offending among males but not females, supporting Hypothesis 4 among males only. The direct effect of friends and family suicidal behavior remains positive and highly significant for both groups. Like with depression, experiencing the death or near death of a friend or family member or both due to suicide increases subsequent criminal behavior among everyone in this sample, but this effect is almost twice as big for those males who are homozygous for the s-allele (from Model 2 in Table 4, the coefficient for friends and family suicidal behavior plus the interaction term coefficient) compared with males who aren’t.

Table 4.

Criminal Behavior W2 Regressed on Friends and Family Suicidal Behavior, 5-HTTLPR Genotype, Depressive Symptoms, and Controls.

	Model 1		Model 2		Model 3
	Males	Females	Males	Females	Males	Females
Variables	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Environment and genetic variables
Friends and family suicidal behavior	.54 (0.11)**	.65 (0.11)**	.52 (0.14)**	.58 (0.12)**	.45 (0.15)**	.42 (0.12)**
5-HTTLPR (1 = s/s)	.02 (0.10)	−0.15 (0.11)	−0.07 (0.10)	−0.24 (0.14)	−0.08 (0.10)	−0.27 (0.14)
Two-way interaction
Friends and family suicidal			.42 (0.21)*	.26 (0.23)	.36 (0.20)	.33 (0.23)
Behavior × 5-HTTLPR
Mediating variable
Depressive symptoms					.04 (0.01)**	.04 (0.01)**
Controls
Hispanic	.16 (0.17)	.69 (0.14)**	.17 (0.17)	.69 (0.14)**	.12 (0.17)	.59 (0.15)**
Black	.15 (0.13)	.50 (0.19)**	.15 (0.13)	.50 (0.19)**	.12 (0.13)	.42 (0.18)*
Native American	.52 (0.24)*	.40 (0.23)	.51 (0.24)*	.42 (0.23)	.40 (0.23)	.35 (0.24)
Asian	−0.26 (0.21)	.22 (0.31)	−0.26 (0.22)	.23 (0.31)	−0.34 (0.22)	.23 (0.36)
Other	.27 (0.35)	.95 (0.39)*	.27 (0.35)	.95 (0.38)*	.22 (0.35)	.82 (0.33)*
Parent’s education	−0.23 (0.11)*	−0.35 (0.13)*	−0.23 (0.11)*	−0.35 (0.13)*	−0.22 (0.11)*	−0.34 (0.13)*
Parent receiving public assistance	.13 (0.15)	−0.25 (0.20)	.13 (0.15)	−0.24 (0.20)	.09 (0.15)	−0.31 (0.20)
Age W2	.00 (0.03)	−0.18 (0.04)**	.00 (0.03)	−0.18 (0.04)**	−0.03 (0.03)	−0.21 (0.03)**
Constant	−0.70 (0.52)	1.12 (0.59)	−0.66 (0.52)	1.15 (0.59)	−0.58 (0.52)	1.19 (0.56)*

Note. Non-Hispanic White is the reference category for all race/ethnic groups. This table includes unstandardized coefficients (linearized standard errors) from negative binomial models. W2 = Wave II.

p < .05. **p < .01.

Model 3 serves as a test of whether depressive symptoms mediate the significant GxE effect found among males in Model 2. For both males and females, depressive symptoms at Wave I have a significant and positive effect on criminal behavior at Wave II. As predicted and supporting Hypothesis 5 for males, the Depressive Symptoms Scale mediates the GxE term. The GxE term is no longer significant at the .05 level, and the coefficient has been reduced by approximately 15.2%. The direct effect of friends and family suicidal behavior remains highly significant for males and females. In these analyses, the biosocial GST model tested is only supported among males, while the traditional GST model that doesn’t account for genetics is supported among females.

Discussion

In this article, I proposed and tested an integrated theoretical model to explain past findings concerning 5-HTTLPR genotype, stressful events like suicidal behavior by others, negative emotions, and antisocial behavior. I specifically sought to test whether suicidal behavior by others has direct effects on depressive symptoms and criminal behavior, whether these effects are moderated by 5-HTTLPR genotype, and whether depressive symptoms mediate these GxE effects on criminal behavior. I utilized data from the Add Health study (Udry, 2003) that measured adolescents’ experience of suicidal behavior by others in the past year, depressive symptoms, criminal offending, 5-HTTLPR genotype, and numerous control variables. The results of both OLS and NB regression models reveal a significant GxE among males, where male adolescents who are faced with the actual or near loss of valued persons to suicide are more likely to experience depressive symptoms and act out in antisocial ways if they are homozygous for the s-allele of 5-HTTLPR versus some other allelic combination (l/l, s/l, l/s). In addition, the GxE effect on criminal behavior among males is mediated by the scale of depressive symptoms. While there is not a significant GxE among females, suicidal behavior by others does increase both depressive symptoms and criminal behavior. Before turning to the implications of this study, the limitations within should be noted.

While the present study makes important contributions to the GxE and criminological literatures, two key limitations should be noted. First, this study focuses on suicidal behavior by others because it has been underexplored and a theoretical explanation for its effects have been lacking. While this is a serious vicarious strain that past research has tied to both experiences of depression and antisocial behavior (Brent et al., 1995; Burton et al., 1994; Cerel et al., 2000; Kaufman, 2009; Pfeffer et al., 1997; Schofield & Ratnarajah, 2007; Shepherd & Barraclough, 1976; Simone, 2008; Wilcox et al., 2010), the GST literature describes many other strains that are of importance in shaping negative emotions and antisocial behavior. The central focus of this study was looking at suicidal behavior by others, so any support provided for the larger biosocial GST model tested in this study should be considered tentative until further research explores more strains and how their effects on negative emotions and offending are conditioned by 5-HTTLPR. Second, while the measure of depressive symptoms includes items tapping emotions like fear, anxiety, and guilt, these emotions may be worth exploring on their own in future research. These results tie together previous findings concerning the effects of suicidal behavior by others, the moderating effect of 5-HTTLPR on stressful life events, and the GST literature. Future studies would do well to consider the potential role of other strains and how their effects on negative emotions and antisocial behaviors are moderated by 5-HTTLPR genotype. Moreover, future research must account for and seek to explain gender differences in GxE effects. In the present study, the direct effects of suicidal behavior by others on depressive symptoms and criminal behavior are significant for both males and females, but the GxE effects are only significant for males. Why this should be is not clear, and more genetics research and theorizing are needed to understand what is occurring when GxE effects hold for one of these groups and not the other. One possibility is that identical allelic variations effect males and females differently, while another possibility is that the effects of one gene, in this case 5-HTTLPR, may be partly dependent on the allelic variations of other genes an individual carries (Priess-Groben & Hyde, 2013). These possibilities deserve further exploration in the future.

In conclusion, this study provides evidence that experiencing the death or near death of a friend or family member interacts with 5-HTTLPR genotype to increase risks for experiencing depressive symptoms and engaging in criminal behavior among adolescent males, while among females, only the direct effect of suicidal behavior by others is significant. These findings support the utility of GST as an explanation of previous findings concerning the effects of suicidal behavior by others on negative emotional states and antisocial behavior. In addition, these findings offer very tentative support for the utility of an integrated biosocial GST model whereby the effects of strain on negative emotions and antisocial behavior are moderated by 5-HTTLPR genotype. Theoretically, this study is important in showing the utility in combining traditional theories with a biosocial modeling approach. This kind of theory building and modeling strategy is important because it helps further develop ideas centered on the concept that the environment and biology are always interacting to shape how we experience and react to our world.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Stephen J. Watts is an assistant professor in the Department of Criminology and Criminal Justice at the University of Memphis. His research focuses on the testing of criminological theories and the integration of these theories into a biosocial framework for explaining criminal and deviant behaviors.

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-88.

Agnew

(2002). Experienced, vicarious, and anticipated strain: An exploratory study on physical victimization and delinquency. Justice Quarterly, 19, 603-632.

Agnew

(2006). Pressured into crime: An overview of general strain theory. Oxford, UK: Oxford University Press.

Belsky

(1997). Variation in susceptibility to rearing influences: An evolutionary argument. Psychological Inquiry, 8, 182-186.

Belsky

(2005). Differential susceptibility to rearing influences: An evolutionary hypothesis and some evidence. In Ellis

Bjorklund

(Eds.), Origins of the social mind: Evolutionary psychology and child development (pp. 139-163). New York, NY: Guildford.

Belsky

Jonassaint

Pluess

Stanton

Brummett

Williams

(2009). Vulnerability genes or plasticity genes? Molecular Psychiatry, 14, 746-754.

Belsky

Pluess

(2009). Beyond diathesis stress: Differential susceptibility to environmental influences. Psychological Bulletin, 135, 885-908.

Bradley

S. L.

Dodelzon

Sandhu

H. K.

Philibert

R. A.

(2005). Relationship of serotonin transporter gene polymorphisms and haplotypes to mRNA transcription. American Journal of Medical Genetics, Part B: Neuropsychiatric Genetics, 136, 58-61.

Brent

D. A.

Perper

J. A.

Moritz

Liotus

Richardson

Canobbio

. . . Roth

(1995). Posttraumatic stress disorder in peers of adolescent suicide victims: Predisposing factors and phenomenology. Journal of the American Academy of Child & Adolescent Psychiatry, 34, 209-215.

10.

Brezina

(1996). Adapting to strain: An examination of delinquent coping responses. Criminology, 34, 39-60.

11.

Broidy

(2001). A test of general strain theory. Criminology, 39, 9-36.

12.

Broidy

Agnew

(1997). Gender and crime: A general strain theory perspective. Journal of Research in Crime & Delinquency, 34, 275-306.

13.

Brummett

B. H.

Boyle

S. H.

Siegler

I. C.

Kuhn

C. M.

Ashley-Koch

Jonassaint

C. R.

. . . Williams

R. B.

(2008). Effects of environmental stress and gender on associations among symptoms of depression and the serotonin transporter gene linked polymorphic region (5-HTTLPR). Behavior Genetics, 38, 34-43.

14.

Burton

Foy

Bwanausi

Johnson

Moore

(1994). The relationship between traumatic exposure, family dysfunction, and post-traumatic stress symptoms in male juvenile offenders. Journal of Traumatic Stress, 7, 83-93.

15.

Carver

C. S.

Johnson

S. L.

Joormann

LeMoult

Cuccaro

M. L.

(2011). Childhood adversity interacts separately with 5-HTTLPR and BDNF to predict lifetime depression diagnosis. Journal of affective disorders, 132, 89-93.

16.

Caspi

Sugden

Moffitt

T. E.

Taylor

Craig

I. W.

Harrington

. . . Poulton

(2003). Influence of life stress on depression: Moderation by a polymorphism in the 5-HTT gene. Science, 301, 386-389.

17.

Cerel

Fristad

M. A.

Weller

E. B.

Weller

R. A.

(2000). Suicide-bereaved children and adolescents: II. Parental and family functioning. Journal of the American Academy of Child & Adolescent Psychiatry, 39, 437-444.

18.

Cloward

R. A.

Ohlin

L. E.

(1960). Delinquency and opportunity: A theory of delinquent gangs. New York, NY: The Free Press.

19.

Cohen

A. K.

(1955). Delinquent boys. New York, NY: MacMillan.

20.

Eley

T. C.

Sugden

Corsico

Gregory

A. M.

Sham

McGuffin

. . . Craig

I. W.

(2004). Gene-environment interaction analysis of serotonin system markers with adolescent depression. Molecular Psychiatry, 9, 908-915.

21.

Gardner

Mulvey

E. P.

Shaw

E. C.

(1995). Regression analyses of counts and rates: Poisson, overdispersed Poisson, and negative binomial models. Psychological Bulletin, 118, 392-404.

22.

Goldman

Glei

D. A.

Lin

Y. H.

Weinstein

(2010). The serotonin transporter polymorphism (5-HTTLPR): Allelic variation and links with depressive symptoms. Depression & Anxiety, 27, 260-269.

23.

Guo

Roettger

M. E.

Cai

(2008). The integration of genetic propensities into social-control models of delinquency and violence among male youths. American Sociological Review, 73, 543-568.

24.

Guo

Tong

Cai

(2008). Gene by social context interactions for number of sexual partners among white male youths: Genetics-informed sociology. American Journal of Sociology, 114(Suppl. 1), S36-S66.

25.

Hagan

Foster

(2003). S/He’s a rebel: Toward a sequential stress theory of delinquency and gendered pathways to disadvantage in emerging adulthood. Social Forces, 82, 53-86.

26.

Harris

K. M.

Florey

Tabor

Bearman

P. S.

Jones

Udry

J. R.

(2003). The national longitudinal study of adolescent health: Research design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design

27.

Haynie

D. L.

(2001). Delinquent peers revisited: Does network structure matter? American Journal of Sociology, 106, 1013-1057.

28.

Haynie

D. L.

(2003). Contexts of risks? Explaining the link between girls’ pubertal development and their delinquency involvement. Social Forces, 82, 355-397.

29.

Hindelang

M. J.

Travis

Weis

J. G.

(1981). Measuring delinquency. Beverly Hills, CA: Sage.

30.

Jang

S. J.

(2007). Gender differences in strain, negative emotions, and coping behaviors: A general strain theory approach. Justice Quarterly, 24, 523-553.

31.

Jang

S. J.

Johnson

B. R.

(2003). Strain, negative emotions, and deviant coping among African Americans. Journal of Quantitative Criminology, 19, 79-105.

32.

Jang

S. J.

Johnson

B. R.

(2005). Gender, religiosity, and reactions to strain among African Americans. The Sociological Quarterly, 46, 323-357.

33.

Kaufman

J. M.

(2009). Gendered responses to serious strain: The argument for a general strain theory of deviance. Justice Quarterly, 26, 410-444.

34.

Langton

Piquero

N. L.

(2007). Can general strain theory explain white-collar crime? A preliminary investigation of the relationship between strain and select white-collar offenses. Journal of Criminal Justice, 35, 1-15.

35.

Manuck

S. B.

Flory

J. D.

Ferrell

R. E.

Mulddon

M. F.

(2004). Socio-economic status covaries with central nervous system serotonergic responsivity as a function of allelic variation in the serotonin transporter gene-linked polymorphism region. Psychoneuroendocrinology, 29, 651-668.

36.

Mazerolle

Piquero

A. R.

Capowich

G. E.

(2003). Examining the links between strain, situational and dispositional anger, and crime. Youth & Society, 35, 131-157.

37.

Merton

R. K.

(1938). Social structure and anomie. American Sociological Review, 3, 672-682.

38.

Moon

Morash

McCluskey

C. P.

Hwang

H.-W.

(2009). A comprehensive test of general strain theory: Key strains, situational- and trait-based negative emotions, conditioning factors, and delinquency. Journal of Research in Crime & Delinquency, 46, 182-212.

39.

Osgood

D. W.

McMorris

B. J.

Potenza

M. T.

(2002). Analyzing multiple-item measures of crime and deviance I: Item response theory scaling. Journal of Quantitative Criminology, 18, 267-296.

40.

Pfeffer

C. R.

Martins

Mann

Sunkenberg

Ice

A. M. Y.

Damore

J. P.

Jr. . . . Jiang

(1997). Child survivors of suicide: Psychosocial characteristics. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 65-74.

41.

Piquero

A. R.

Sealock

M. D.

(2004). Gender and general strain theory: A preliminary test of Broidy and Agnew’s gender/GST hypotheses. Justice Quarterly, 21, 125-158.

42.

Priess-Groben

H. A.

Hyde

J. S.

(2013). 5-HTTLPR X Stress in adolescent depression: Moderation by MAOA and gender. Journal of Abnormal Child Psychology, 41, 281-294.

43.

Reif

Rosler

Freitag

Schneider

Eujen

Kissling

. . . Retz

(2007). Nature and nurture predispose to violent behavior: Serotonergic genes and adverse childhood environment. Neuropsychopharmacology, 32, 2375-2383.

44.

Retz

Freitag

C. M.

Retz-Junginger

Wenzler

Schneider

Kissling

. . . Rosler

(2008). A functional serotonin transporter promoter gene polymorphism increases ADHD symptoms in delinquents: Interaction with adverse childhood environment. Psychiatry Research, 158, 123-131.

45.

Schofield

M. J.

Ratnarajah

(2007). Parental suicide and its aftermath: A review. Journal of Family Studies, 13, 78-93.

46.

Shepherd

D. M.

Barraclough

B. M.

(1976). The aftermath of parental suicide for children. The British Journal of Psychiatry: The Journal of Mental Science, 129, 267-276.

47.

Simone

(2008). Parental suicide: The long-term impact on children and young people. Bereavement Care, 27, 43-46.

48.

Simons

R. L.

Lei

M. K.

Beach

S. R. H.

Brody

G. H.

Philibert

R. A.

Gibbons

F. X.

(2011). Social environment, genes, and aggression: Evidence supporting the differential susceptibility perspective. American Sociological Review, 76, 883-912.

49.

Simons

R. L.

Lei

M. K.

Beach

S. R. H.

Brody

G. H.

Philibert

R. A.

Gibbons

F. X.

(2012). Social adversity, genetic variation, street code, and aggression: A genetically informed model of violent behavior. Youth Violence and Juvenile Justice, 10, 3-24.

50.

Taylor

S. E.

Way

B. M.

Welch

W. T.

Hilmert

C. J.

Lehman

B. J.

Eisenberger

N. I.

(2006). Early family environment, current adversity, the serotonin tranporter promoter polymorphism, and depressive symptomatology. Biological Psychiatry, 60, 671-676.

51.

Udry

J. R.

(2003). The national longitudinal study of adolescent health (Add Health). Chapel Hill: Carolina Population Center, University of North Carolina at Chapel Hill.

52.

Vaughn

M. G.

DeLisi

Beaver

K. M.

Wright

J. P.

(2009). DAT1 and 5HTT are associated with pathological criminal behavior in a nationally representative sample of youth. Criminal Justice and Behavior, 36, 1113-1124.

53.

Verona

Joiner

Johnson

Bender

(2006). Gender specific gene-environment interactions on laboratory-assessed aggression. Biological Psychiatry, 71, 33-41.

54.

Wilcox

H. C.

Kuramoto

S. J.

Lichtenstein

Långström

Brent

D. A.

Runeson

(2010). Psychiatric morbidity, violent crime, and suicide among children and adolescents exposed to parental death. Journal of the American Academy of Child & Adolescent Psychiatry, 49, 514-523.