Intimate Partner Violence and Welfare Participation

Abstract

This longitudinal study examined the temporal-ordered causal relationship between intimate partner violence (IPV), five mental disorders (depression, generalized anxiety disorder, social phobia, panic attack, posttraumatic stress disorder [PTSD]), alcohol abuse/dependence, drug abuse/dependence, treatment seeking (from physician, counselor, and self-help group), employment, child support, and welfare participation. It was a secondary data analysis of records of 571 women; the records were extracted from the study “Violence Against Women and the Role of Welfare Reform” (VAWRWR). Results from generalized estimating equations (GEE) showed that experiencing controlling behaviors reduced likelihood of welfare participation whereas experiencing physical abuse increased it. Significant impact on welfare participation was wielded by panic attack, drug abuse/dependence, and employment; treatment seeking and child support made no significant impact. The study found no significant mediating effect wielded by panic attack, drug abuse/dependence, employment, or child support on welfare participation’s relationship to controlling behaviors or physically abusive behaviors experienced. Implications for intervention are discussed.

Keywords

battered women domestic violence mental health violence intervention/treatment

Introduction

Intimate partner violence (IPV) comprises physical, sexual, and psychological harm inflicted by a current or former, heterosexual or same-sex, partner or spouse (Centers for Disease Control and Prevention, 2010). Although IPV may manifest in numerous forms of abusive relationships, the present study focused on intimate terrorism, a relationship between a nonviolent, noncontrolling woman and her violent, controlling male partner (Johnson, 1995, 2006, 2011). IPV—which often reflects the abusive man’s fear of rejection, extreme jealousy, tolerance of violence, and general hostility toward women—involves a moderate-to-high level of violence (Holtzworth-Munroe, 2000; Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2000, 2003). When difference in partners’ physical size is not overwhelming, victims often defend themselves instinctively against their abusers (Johnson, 2011), but that type of mutual IPV was not the present study’s focus.

The study did focus on female targets of IPV who received benefits from Temporary Assistance to Needy Families (TANF). TANF began in 1996, charged with helping low-income families through cash payments. In 2008, more than 1.8 million families were enrolled, receiving on average (for a family with two children) US$390 monthly; the maximum monthly benefit provided by the states ranged from US$170 in Mississippi to US$923 in Alaska (U.S. Department of Health & Human Services, 2008). Research has shown IPV rates to be higher among welfare recipients than in the normative population (Tolman & Raphael, 2000), with 29% to 74% of TANF recipients reporting recent (within 12 months of interview) experience of IPV, versus 22% to 31% of the normative population (Lown, Schmidt, & Wiley, 2006; Yoshihama, Hammock, & Horrocks, 2006). Of TANF recipients reporting IPV, 4.5% to 12.3% described it as severe (Seefeldt & Orzol, 2005; Taylor & Barusch, 2004). In addition, among TANF recipients, lifetime prevalence of mental disorders is 53% and of substance use disorders is 29% (Cook et al., 2009). Such proportions suggest the importance of a study like ours concerning effects of substance use, mental disorders, and IPV on TANF participation.

Our study proposed that, because it fosters mental and substance use problems and jeopardizes treatment of these, IPV shapes women’s subsequent employment and child support, which in turn shapes eventual welfare use; that is, mental disorders, substance use, treatment seeking, employment, and child support mediate IPV’s impact on women’s welfare participation (see Figure 1).

Figure 1.

Proposed relationship between welfare participation and domestic violence and their mediating factors

Literature Review

Abusive and Controlling Behaviors

Victims of IPV may experience it in the three cyclical phases—tension-building phase, acute-battering phase, and loving-contrition phase—characteristic of abusive intimate relationships (Walker, 2000). The phases illustrate that the violence is not solely physical. The cycle also involves recurring controlling behaviors; these dominate the first and third phases, physical violence marking the second. Some IPV types are strongly positively correlated with controlling behaviors: physical assault, rape, stalking (Miller, 2006), and the psychological type of IPV (Prospero, 2008).

A woman experiencing IPV as controlling behaviors may be affected differently than one experiencing it as physical abuse. Walker (2000) suggested that victims of relatively more physical abuse than controlling behaviors were relatively likely to leave an abusive partner. Women subjected to controlling behaviors tend to become trapped in their abusive relationships, growing unlikely to participate in welfare or other services (McCloskey et al., 2007). In contrast, increasingly severe physical abuse is associated with increasing motivation to change (Chang et al., 2010) and to seek help (Cattaneo & DeLoveh, 2008; Cattaneo, Stuewig, Goodman, Kaltman, & Dutton, 2007). Women then may leave a relationship, positioning themselves to seek help, eventually, from welfare programs.

Welfare Participation

Whether it comprises primarily controlling behaviors or actual physical abuse, IPV impairs a woman’s ability to get and keep a job (Goodman, Smyth, Borges, & Singer, 2009). IPV is negatively associated with employment (Kimerling et al., 2009; Staggs, Long, Mason, Krishnan, & Riger, 2007). Moreover, employed women with IPV experiences have reported difficulty concentrating and excelling on the job (Swanberg & Macke, 2006). Up to 25% of unemployed low-income single mothers report a history of IPV (Blank, 2007). In light of employment (and resulting financial) difficulties, an IPV victim may apply for TANF.

Research on IPV’s direct relationship to welfare participation has yielded mixed findings. Two prior studies (Romero, Chavkin, Wise, Smith, & Wood, 2002; Seefeldt & Orzol, 2005) found no significant association between IPV and welfare participation, following cross-sectional analysis of more than 500 participants. (The employed IPV measure for one of these studies was unclear; the other presented respondents multiple items covering physical and sexual violence.) In contrast, a small longitudinal study, sampling just 40 TANF recipients, concluded IPV experience made welfare participation likelier (Yoshihama et al., 2006). The small study used temporal-order causal modeling and measured controlling and abusive behaviors with standardized scales; its reliance on a small, homogeneous sample limits its generalizability.

Health

Consequences of physically violent IPV include chronic health problems as well as acute injury. Multiple injuries to head, neck, arm, torso, and so on may result and may be severe (Btoush, Campbell, & Gebbie, 2009; Nurius & Macy, 2010). Victims may also see their physical functioning decrease or may develop debilitating chronic pain (Nurius & Macy, 2010; Wuest et al., 2009). Unsurprisingly, IPV victims’ health is often poor (Bonomi, Anderson, Rivara, & Thompson, 2007; Macy, Ferron, & Crosby, 2009), remaining so even years after an abusive relationship ends (Ford-Gilboe et al., 2009). Poor health figures strongly in longer-term TANF receipt (Alzate, Moxley, Bohon, & Nackerud, 2009; Romero et al., 2002); on average, a TANF participant has four chronic health conditions (Kneipp et al., 2011).

Mental Health

Unsurprisingly as well, IPV victims may have low self-esteem, depression, and learned helplessness (Walker, 2000). IPV may affect depression perhaps as long as 14 years after IPV occurs (Lindhorst & Oxford, 2008). In addition, since posttraumatic stress disorder (PTSD) is significantly associated with unemployment (Kimerling et al., 2009), it is plausible that some mental health disorders afflict IPV victims who are unemployed. At least one study with TANF recipients linked IPV victimization to problem drinking and heavy drug use (Lown et al., 2006). Other findings, though, suggest neither welfare participation nor wages are significantly affected by substance use, mood, or anxiety disorders (Cheng & McElderry, 2007; Cook et al., 2009; Montoya, Bell, Atkinson, Nagy, & Whitsett, 2002; Romero et al., 2002).

Treatment Seeking by IPV Victims Enrolled in TANF

Treatment seeking has been observed in as few as 7%, and as many as 41%, of those TANF recipients with mental or substance use disorders (Chandler, Meisel, Jordan, Rienzi, & Goodwin, 2004; Cook et al., 2009). Guilt and shame, unfamiliarity with mental health services, lack of access, language barriers, or, among IPV victims, partner intrusion may prevent treatment seeking (McCloskey et al., 2007; Rodriguez, Valentine, Son, & Muhammad, 2009). The relationship between treatment seeking for IPV and participation in welfare remains unexplored. However, TANF programs do feature a “Family Violence Option” under which work stipulations are waived for IPV victims (Pyles, 2006; Seefeldt & Orzol, 2005); through this opportunity, trained TANF staff help victims develop safety plans and refer them for other services (Saunders, Holter, Pahl, & Tolman, 2006). But only 9% of TANF recipients are screened for IPV (Lindhorst, Meyers, & Casey, 2008).

Human Capital and Child Support

Human capital factors including college education, marketable occupational skills, and work experience appear to assist welfare recipients in gaining employment and financial self-sufficiency (Cheng, 2002; Cheng & Lo, 2010; Hollenbeck & Kimmel, 2002; Kalil, Seefeldt, & Wang, 2002). Furthermore, receiving child support can be of significant help to mothers enrolled in welfare programs (Cheng, 2002; Cheng & Lo, 2010; Keng, Garasky, & Jensen, 2002; Mead, 2003). But most TANF recipients who have experienced IPV decline to pursue child support from an abuser, lest the petition provoke violence (Menard & Turetsky, 1999; Pearson, Griswold, & Thoennes, 2001).

The few studies on IPV’s relationship to welfare participation have been limited by their small samples or cross-sectional designs. And little attention has been paid to distinguishing physical abuse from controlling behaviors while studying victims’ welfare participation. The present study sought to avoid these limitations and, further, to control for human capital factors and health conditions. We hypothesize as follows:

Hypothesis 1: Experiencing controlling behaviors reduces the likelihood of welfare participation, whereas experiencing physical abuse increases it.

Hypothesis 2: Certain mental disorders, substance abuse, and treatment seeking increase welfare participation’s likelihood, whereas employment and child support reduce it.

Hypothesis 3: IPV’s impact on welfare participation is mediated by the presence of certain mental disorders, substance abuse, treatment seeking, employment, and child support.

Method

Sample

We extracted our sample from a data set, the “Violence Against Women and the Role of Welfare Reform” project (VAWRWR). In three waves of interviews between 1999 and 2002, VAWRWR had collected information from 632 mothers in two California counties (Inter-University Consortium for Political and Social Research, 2009). It had used stratification sampling to choose these respondents out of more than 4,500 TANF applicants and recipients. VAWRWR’s response rate was 71%. Interviewees provided comprehensive information about their IPV experiences, mental health problems, alcohol and drug problems, and treatment seeking. The research interviews occurred conjointly with routine appointments in TANF offices, safeguarding those respondents with abusive partners. The present secondary data analysis involved longitudinal records for 571 of the 632 women, those who answered all items necessary to address all variables in our study.

Measures

The present study comprised temporal-ordering causal analysis of the outcome variable subsequent welfare participation; explanatory variables included IPV. We investigated the impact of IPV reported in a given interview wave and on welfare participation in the next wave. We expected that seven of our explanatory variables (current welfare participation, number of controlling behaviors, number of physically abusive behaviors, physical health, specific mental disorder, specific substance use disorder, treatment seeking), measured in the first interview wave, would predict the outcome variable in the second wave, and measured in the second wave, would predict the outcome variable in the third wave. In other words, we expected these seven explanatory variables to exert lagged effects on subsequent welfare participation (Finkel, 1995). We expected four other explanatory variables (being employed, number of dependent children, received child support, married) to show relatively instantaneous impacts on subsequent welfare participation (Finkel, 1995). That is, speaking theoretically, these four variables’ causal impacts on the outcome variable should exhibit time lags shorter than the period elapsing between two VAWRWR interview waves. The explanatory variables assumed to have instantaneous impacts were measured in the same interview wave as the outcome variable.

We deemed some variables time varying, others time invariant (including racial and ethnic background, educational level in the first wave, number of controlling behaviors prior to the survey, number of physically abusive behaviors prior to the survey, and age 24 or younger in the first wave). A time-invariant variable’s measure or value was constant across interview waves.

The outcome variable, subsequent welfare participation, indicated whether a respondent received TANF benefits in the wave immediately following a given interview wave. Current welfare participation, an explanatory variable serving as a control, indicated whether TANF benefits were already being received at the time of a given interview wave. The remaining explanatory variables were assigned to five groups: IPV experience, specific mental disorder, specific substance use disorder, human capital, child support, and other demographic characteristics.

The variables number of controlling behaviors and number of controlling behaviors prior to the survey measured the sample’s experience of IPV. To estimate how many controlling behaviors had been experienced from a spouse or partner in the past 12 months, VAWRWR interviewers had queried respondents about 11 behaviors: name calling, jealousy, insisting on knowing whereabouts, limiting or denying access to family income, limiting contact with family and friends, stalking, threatening with hitting, threatening with abuser’s own death, coercing sexual activity, threatening to kidnap dependent children, and physically harming dependent children. Queries VAWRWR used to measure controlling behaviors resemble items used in an earlier study of controlling behaviors (Johnson, 1995). For the 11 items, we computed a Cronbach’s alpha of .86 for the first interview wave and .87 for the second. In addition, number of controlling behaviors prior to the survey was used to indicate how many of the 11 behaviors the spouse or partner had exhibited prior to the survey. This variable had a Cronbach’s alpha of .89 and provided a control on each respondent’s history of controlling behaviors.

Also included in the IPV experience group of explanatory variables was number of physically abusive behaviors, measured by items covering 7 spousal/partner behaviors during the past 12 months: throwing objects, pushing or grabbing or shoving, slapping, kicking or biting, hitting, beating, and choking. VAWRWR surveys drew on the widely used Conflict Tactics Scales (or CTS), borrowing seven of its nine items gauging intimate partner physical violence (Straus, Hamby, BoneyMcCoy, & Sugarman, 1996). For the seven items, we found a Cronbach’s alpha of .90 for the first interview wave and .91 for the second. The variable number of physically abusive behaviors prior to the survey was used to indicate how many of the seven behaviors the spouse or partner had exhibited prior to the survey. This variable had a Cronbach’s alpha of .92 and served to control each respondent’s history of physical abuse. Another variable reflecting IPV experience was physical health. Its scores were based on the Short-Form Health Survey’s (or SF-12’s) Physical Component Summary (PCS), a weighted score (Ware, Kosinski, & Keller, 1996). The SF-12 comprises 12 items; its PCS includes 6 items related to physical functioning, general health, and bodily pain, with higher scores indicating better health.

We measured respondents’ specific mental disorders and substance abuse/dependence with seven dummy variables: depression, generalized anxiety disorder, social phobia, panic attack, PTSD, alcohol abuse/dependence, and drug abuse/dependence. For each, a reference group comprised respondents free of the disorder. VAWRWR researchers had used the Composite International Diagnostic Interview—Short Form (CIDI-SF) to survey respondents about these disorders. CIDI-SF is a highly accurate classification system that in 77% to 100% of cases correctly identifies these disorders (Kessler, Andrews, Mroczek, Ustun, & Wittchen, 1998). Our variable treatment seeking indicated consulting a professional (physician, counselor, or self-help group with facilitator), during the past 12 months, for help with an emotional, mental, or substance use problem.

The human capital group of explanatory variables included being employed (yes/no) and education level at the first wave; three responses were offered for the latter, at least some college, high school diploma or GED, and no high school (the reference group). The final variable in this group, received child support, indicated whether a respondent’s child had gotten any support payments from the noncustodial parent.

Demographic characteristics included racial and ethnic background (White, African American, Hispanic, other ethnic minority), married (wed to a spouse or living with a common-law partner), age 24 or younger in the first wave (yes/no), and number of dependent children. VAWRWR measured age in three broad categories (24 or below, 25 to 44, 45 to 64); with more-detailed data unavailable, we chose to subsume these in a dichotomous variable.

The variables first wave and second wave indicated that interview wave at which respondents’ current welfare participation was measured. These time indicators served as a control during data analysis. The first wave provided the reference category.

Data Analysis

During data analysis, each of the 571 longitudinal records was initially split into a first record, consisting of the first and second interview waves, and a second record, covering the second and the third waves. The final analytical sample comprised 1,103 split records. Since the outcome variable was dichotomous, we employed STATA generalized estimating equations (GEE) with binomial family and logit Link options. We chose a first-order autoregressive correlation structure to estimate autocorrelations among repeat observations and among longitudinal records of unequal length, assuming the correlations of repeated measurements within waves to be time dependent (Hardin & Hilbe, 2003; Rabe-Hesketh & Skrondal, 2008).

Multicollinearity diagnostic procedures indicated that only the tolerance statistics (not shown) for IPV variables fell below .4 (.35 to .39). The correlation between number of controlling behaviors and number of physically abusive behaviors was .79 and that between number of controlling behaviors prior to the survey and number of physically abusive behaviors prior to the survey was .77. These correlations echo the findings of a prior study (Miller, 2006). Nevertheless, we opted for separate measures of controlling behaviors and of physically abusive behaviors during analysis, since our focus was the differential impact of each behavior type.

We developed three analytical models to test hypothesized mediating effects of mental disorders, substance use disorders, treatment seeking, employment, and child support on the relationship between IPV and TANF participation. The first model included the current welfare participation, IPV, and demographic variables only. To create the second model, we added to these the mental disorder, substance use disorder, and treatment seeking variables. To create the third model, we added the employment and child support variables to the second model. As part of evaluating Models 2 and 3, we identified statistically significant “hypothesized mediators” in our results and gauged their mediating effects using Sobel–Goodman tests (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002; UCLA Academic Technology Services, 2010).

Results

Of the 571 respondents, 40.3% were White, 15.4% were African American, 37.1% were Hispanic, and 7.2% were from some other racial and ethnic group; 23.6% of the women were 24 years old or younger. Of the respondents, 24.5% had at least some college education, 34.5% had graduated from high school or earned a GED, and the rest (41%) had left school before high school graduation (with no GED) at the first wave. On average, the women reported experiencing 2.73 (of 11 possible) controlling behaviors prior to the survey and 2.15 (of 7 possible) physically abusive behaviors prior to survey; 60% reported experiencing controlling behaviors sometime prior to survey, with 49.4% reporting prior experience of physically abusive behaviors (data not shown).

In 63% of the obtained 1,103 split records, the respondent reported participating in TANF currently; in 54%, the respondent reported subsequent TANF participation (see Table 1, column labeled “Total Sample”). Across the split records, the average number of controlling behaviors experienced was 1.37 (of 11 possible) and of physically abusive behaviors experienced was 0.61 (of 7 possible). The average physical health score was 48.5 (of 67.1 possible). Depression was exhibited in 30.1% of the split records, general anxiety disorder in 95.0%, social phobia in 40.6%, panic attack in 12.7%, PTSD in 13.1%, alcohol abuse/dependence in 6.8%, and drug abuse/dependence in 5.3%. In 17.6% of the split records, a respondent reported seeking treatment. In 47.5%, a respondent reported being employed, and in 33.5% she reported being married. On average across the split records, a woman reported having two children; in 27.3% of the records, child support was reportedly received. Nearly 52% of the split records were generated from the first and second interview waves, the remainder from the second and third waves. Descriptive statistics contrasting TANF recipients and nonrecipients appear in Table 1, in the two columns after “Total Sample.”

Table 1.

Descriptive Statistics of Time-Varying Outcome and Explanatory Variables (N = 1,103 Split Records)

	Total Sample (N = 1,103)		TANF Recipients (n = 596)		Nonrecipients (n = 507)
Variable	%	M	%	M	%	M
Outcome variable
Subsequent welfare participation
Yes	54.0	—	100	—	0	—
No	46.0	—	0	—	100	—
Explanatory variables
Current welfare participation
Yes	63.0	—	73.7	—	50.5	—
No	37.0	—	26.3	—	49.5	—
Number of controlling behaviors	—	1.36	—	1.30	—	1.45
Number of physically abusive behaviors	—	0.61	—	0.63	—	0.58
Physical health	—	48.50	—	48.39	—	48.63
Depression	30.1	—	30.7	—	29.4	—
Generalized anxiety disorder	95.0	—	95.1	—	94.9	—
Social phobia	40.6	—	41.3	—	39.8	—
Panic attack	12.7	—	14.1	—	11.0	—
Posttraumatic stress disorder	13.1	—	13.8	—	12.2	—
Alcohol abuse/dependence	6.8	—	7.6	—	5.9	—
Drug abuse/dependence	5.3	—	5.9	—	4.5	—
No disorder or abuse/dependence	3.8	—	3.9	—	3.7	—
Treatment seeking
Yes	17.6	—	19.0	—	16.0	—
No	82.4	—	81.0	—	84.0	—
Being employed
Yes	47.5	—	37.6	—	59.2	—
No	52.5	—	62.4	—	40.8	—
Received child support
Yes	27.3	—	29.7	—	24.5	—
No	72.7	—	70.3	—	75.5	—
Married
Yes	33.5	—	28.4	—	39.6	—
No	66.5	—	71.6	—	60.4	—
Number of dependent children	—	2.23	—	2.44	—	1.98
First wave	51.8	—	54.4	—	51.3	—
Second wave	48.2	—	45.6	—	48.7	—

The most reported controlling behaviors were name calling (in 21.9% of the split records), jealousy (in 25.9%), and insisting on knowing whereabouts (in 23.1%); least reported were coercing sexual activity (in 2.3%) and physically harming dependent children (in 1.5%; see Table 2). The most reported physically abusive behaviors were pushing/grabbing/shoving (in 17.6% of the split records), throwing objects (in 9.8%), and slapping (in 9.0%); least reported was beating (in 4.7%). In almost 62% of the split records, a woman reported she experienced no controlling behaviors; in almost 81.1%, no physical abuse was cited.

Table 2.

Descriptive Statistics of Specific Controlling and Physically Abusive Behaviors (N = 1,103 Split Records)

Controlling Behaviors	%	Physically Abusive Behaviors	%
Name calling	21.9	Throwing objects	9.8
Jealousy	25.9	Pushing, grabbing, or shoving	17.6
Insisting on knowing whereabouts	23.1	Slapping	9.0
Limiting/denying access to family income	4.5	Kicking or biting	7.7
Limiting contact with family and friends	13.6	Hitting	6.9
Stalking	9.9	Beating	4.7
Threatening with hitting	12.8	Choking	5.2
Threatening with abuser’s own death	8.6	None	81.1
Coercing sexual activity	2.3
Threatening to kidnap dependent children	7.2
Physically harming dependent children	1.5
None	61.9

Multivariate analysis showed all of our hypothesized models to differ significantly from the null model (Wald’s χ² ranged from 118.62 to 165.87, with p < .01; see Table 3). In Model 1, with current welfare participation and demographic variables included, higher numbers of controlling behaviors significantly reduced likelihood of subsequent welfare participation (OR = 0.90, p < .05), and higher numbers of physically abusive behaviors increased it (OR = 1.17, p < .05). Multivariate results also showed that presurvey experiences of controlling or physically abusive behaviors did not significantly affect subsequent welfare participation and that physical health did significantly affect participation (OR = 0.99, p < .05).

Table 3.

Results of Generalized Estimating Equations for Subsequent Welfare Participation (N = 1,103 Split Records)

	Model 1		Model 2		Model 3
Variables	OR	SE	OR	SE	OR	SE
Current welfare participation	2.58**	.34	2.77**	.38	2.93**	.41
Number of controlling behaviors	0.90*	.04	0.87**	.04	0.87**	.04
Number of controlling behaviors prior to the survey	0.96	.03	0.95	.03	0.95	.03
Number of physically abusive behaviors	1.17*	.08	1.16*	.08	1.14*	.08
Number of physically abusive behaviors prior to the survey	1.05	.04	1.04	.04	1.03	.04
Physical health	0.99*	.00	0.99	.00	1.00	.01
Depression	—	—	1.14	.19	1.13	.19
Generalized anxiety disorder	—	—	0.84	.21	0.84	.21
Social phobia	—	—	0.93	.13	0.87	.13
Panic attack	—	—	1.53*	.34	1.55*	.36
Posttraumatic stress disorder	—	—	1.08	.24	1.11	.26
Alcohol abuse/dependence	—	—	1.10	.32	1.13	.34
Drug abuse/dependence	—	—	1.94*	.66	1.77*	.61
Treatment seeking	—	—	1.31	.26	1.23	.25
Being employed	—	—	—	—	0.40**	.06
Received child support	—	—	—	—	1.09	.17
At least some college	0.67**	.11	0.64**	.11	0.77	.13
High school diploma or GED	0.73*	.11	0.71*	.11	0.82	.13
Married	0.50**	.07	0.49**	.07	0.46**	.07
Age 24 or younger in the first wave	1.31**	.07	1.33**	.07	1.31**	.07
African American	0.83	.17	0.89	.19	0.94	.20
Hispanic	0.99	.15	1.04	.16	1.07	.17
Other ethnic minority	0.87	.23	0.87	.23	0.91	.25
Number of dependent children	1.09	.18	1.12	.19	1.08	.18
Second wave	0.87	.11	0.87	.12	0.90	.12
Wald’s χ²	118.62**		134.17**		165.87**

Note: OR = odds ratio; SE = standard error.

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

In Model 2, which brought in the mental disorder and substance use disorder variables, panic attack (OR = 1.53, p < .05) and drug abuse/dependence (OR = 1.94, p < .05) exhibited statistically significant associations with subsequent welfare participation. We observed no such association, however, for other mental disorders, either for alcohol abuse/dependence or for treatment seeking. In Model 2, the number of controlling or physically abusive behaviors continued to be significantly associated with subsequent welfare participation, although odds ratios fell slightly (less than 0.04) compared to Model 1. When we employed Sobel–Goodman tests, however, to confirm the mediation effect of panic attack and of drug abuse/dependence, results fell short of statistical significance (not shown). Finally, in Model 2, although physical health no longer showed significant association with subsequent welfare participation, two education levels did. College education (OR = 0.64, p < .01) and high school graduation/GED (OR = 0.71, p < .05) remained linked to reduced likelihood of subsequent welfare participation.

Model 3 included our employment and child support variables as well as all other explanatory variables. The observed odds ratios between subsequent welfare participation and number of controlling and abusive behaviors remained significant, despite decreasing slightly. Of the mental disorder and substance use disorder variables, panic attack and drug abuse/dependence alone remained significant predictors of subsequent welfare participation, echoing Model 2’s results. In Model 3, being employed (OR = 0.40, p < .01) significantly reduced likelihood of subsequent welfare participation, whereas child support showed no significant link to it. Moreover, Sobel–Goodman test indicated no significant mediation effect for employment or for child support (not shown). In Model 3, other explanatory variables demonstrating significant association with subsequent welfare participation essentially resembled our Model 2 results. Whereas marriage (OR = 0.46, p < .01) was associated, in Model 3, with reduced likelihood of subsequent welfare participation, having more children (OR = 1.31, p < .01) was associated with increased likelihood of subsequent participation. Our multivariate analysis showed that demographic characteristics other than marriage and children (e.g., education level, ethnicity) did not significantly affect the outcome variable. Similarly, our time indicators showed no significant effect on the outcome.

Discussion

When interviewed for VAWRWR, more than half our respondents had formerly participated, or were still participating, in TANF. A recent experience of physically abusive behavior was reported by 19% of the participants and, of controlling behavior, by 38%; these rates echo findings of some prior studies (Lown et al., 2006; Yoshihama et al., 2006). Our participants reported suffering specific mental disorders at much higher rates than an earlier study found, the rate of panic attack being 21 times higher and the rate of depression being 1.75 times higher (Cook et al., 2009). Interestingly, though, that earlier study reported substance use disorder rates to be very similar to those we observed (Cook et al., 2009). In light of the very high rates of mental disorders we observed, it is especially important to note that less than 18% of the affected women had sought help with mental health or substance use. Plausible reasons for declining to seek treatment include unawareness of treatment availability, lack of access to treatment, and fear.

Able to control for impacts of prior welfare participation and several other variables, our multivariate results supported our first hypothesis, that experiencing controlling behaviors would decrease likelihood of welfare participation whereas experiencing physically abusive behaviors would increase it. Subsequent welfare participation’s positive association with physically abusive behaviors was consistent across the three models tested and is furthermore consistent with prior findings linking physical abuse and treatment seeking by victims (Cattaneo & DeLoveh, 2008; Cattaneo et al., 2007). Our multivariate results imply that IPV victims who have experienced controlling behaviors will be relatively unlikely to apply for TANF, and those who have experienced physically abusive behaviors will be relatively likely to apply for TANF, upon ending the abusive relationship. The significant negative association we observed between subsequent welfare participation and marriage supports this line of thought. (Marital status, of course, also figures in eligibility for TANF.)

Recent controlling behaviors and recent physically abusive behaviors significantly affected our respondents’ likelihood of subsequent welfare participation, while their pre-VAWRWR experiences with intimate partners showed no impact. This supports the idea that abusive intimate relationships are cyclical. Thus, a decision to pursue TANF might spring from an abuse-related crisis occurring in a relationship, despite earlier IPV’s failure to trigger help seeking such as welfare application. Welfare application stemming from crisis in IPV-marred relationships is an opportunity for social work professionals to intervene for the abused.

Our findings partially supported our second hypothesis, that specific mental disorders, substance use disorders, and treatment seeking would increase welfare participation’s likelihood, whereas employment and child support would decrease it. Consistent with prior research (Cheng & Lo, 2010; Cook et al., 2009; Romero et al., 2002), our study did find that neither depression nor alcohol abuse/dependence significantly affected subsequent welfare participation. At the same time—and contrary to another prior study reporting no significant relationship between anxiety disorder and welfare participation (Cook et al., 2009)—our study showed panic attack alone, of the tested mental disorders, to increase subsequent welfare participation’s likelihood. One plausible explanation is that severe anxiety characterizing panic attack severely impairs employability and job performance (Erickson et al., 2009). The significant association we observed between drug abuse/dependence and subsequent welfare participation also contradicted findings of an earlier study (Cheng & McElderry, 2007).

It is interesting that the incorporation of employment in our third model saw education’s significance for welfare participation disappear. The finding may reflect TANF’s emphasis on job placement over education. Not surprisingly, and in line with prior studies (Cheng, 2002; Cheng & Lo, 2010; Hollenbeck & Kimmel, 2002; Kalil et al., 2002), our results indicate that being employed reduces the likelihood of welfare participation. Also supporting earlier research, we observed no significant relationship between child support and welfare participation by IPV victims; Menard and Turetsky (1999) along with Pearson and colleagues (2001) have argued that TANF-receiving IPV victims often view pursuing child support as compromising safety vis-à-vis abusive fathers.

Unsupported by our present findings was our third hypothesis, that specific mental disorders, substance use disorders, treatment seeking, employment, and child support would mediate IPV’s impact on welfare participation. When panic attack and drug abuse/dependence were considered, in Model 2, and when employment and child support were included, in Model 3, we observed slight reductions in odds ratios for experiencing controlling behaviors and physically abusive behaviors. However, further tests failed to confirm any of the hypothesized mediating impacts of variables on IPV’s relationship to welfare: Mediating impacts of these variables were minimal at best. What the absence of support for the hypothesis implies is that IPV—both controlling behaviors and physically abusive behaviors—affects welfare participation fairly directly.

The present study was limited by not including variables describing childhood abuse experienced by the women; such abuse has been linked to increased likelihood of IPV in adulthood (Halpern, Spriggs, Martin, & Kupper, 2009; Miller, 2006). The trauma of childhood abuse might impair victims mentally and physically, hindering financial self-sufficiency and welfare participation alike. Our study was also limited by the fact that enrollment in TANF following application is generally determined according to eligibility/compliance regulations.

Conclusion

Our study demonstrated the temporal-ordered, potentially direct relationship between experience of IPV and subsequent welfare participation. Specifically, it illustrated distinct impacts for controlling behaviors versus physically abusive behaviors on victims’ subsequent welfare participation. Its results imply the cyclical character of IPV. In addition, its finding that women experiencing relatively few controlling behaviors tend to pursue welfare suggests an important need for social workers to inform women of IPV’s cyclical character and foster early recognition of controlling behaviors, enhancing interruption of the cycle. Social workers might also invite women experiencing physical abuse to apply for TANF and other services (e.g., treatment for panic disorder and drug use), helping them end abusive relationships.

The present study’s finding that an unemployed mother is likely to seek TANF benefits implies the importance of helping unemployed mothers find work promptly. However, according to some prior studies, those who leave TANF possessing limited occupational skills, work experience, and education are likely to return to TANF and in some cases to become totally dependent on the program (Anderson & Gryzlak, 2002; Cheng, 2003, 2005, 2010; Kalil et al., 2002; Loprest, 2002). It appears to be important, then, to ensure sufficient skills and employment training of TANF participants before they seek to independently support their families via a paycheck.

Although our findings suggest the relationship between IPV (both types) and subsequent welfare participation is fairly direct, future research might nevertheless productively explore whether other factors perhaps mediate the relationship. National samples should be analyzed; the present sample was relatively small and too geographically specific. Finally, although many women in our sample apparently had mental disorders, few had sought help. Certainly one goal of future research, then, will be to understand barriers to treatment seeking, including the roles that controlling behaviors and physically abusive behaviors play in treatment seeking.

Footnotes

Declaration of Conflicting Interests

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

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Bio

Tyrone C. Cheng received the MSW and PhD in social work from the University of Alabama in 1993. He had taught in Southern Illinois University at Carbondale (1993-1994) and University of Alabama at Birmingham (2002-2008). Since 2008, he has been an associate professor and the MSW program chair for the School of Social Work. He had been a social worker, child therapist, and community organizer in Hong Kong, Chicago, Michigan, and Ohio. His research areas include welfare policies, substance abuse, and child welfare. Recent publications on welfare participation include articles published in Journal of Social Policy, International Journal of Social Welfare, Journal of Social Service Research, Journal of Health Care for the Poor and Underserved, Health & Social Work, and Social Service Review.

References

Alzate

M. M.

Moxley

D. P.

Bohon

S. A.

Nackerud

(2009). The influence of perceived health status on poor women’s confidence in leaving welfare: Implications for social work. Social Work in Health Care, 48(1), 57-75.

Anderson

S. G.

Gryzlak

B. M.

(2002). Social work advocacy in the post-TANF environment: Lessons from early TANF research studies. Social Work, 47(3), 301-314.

Blank

R. M.

(2007). Improving the safety net for single mothers who face serious barriers to work. Future of Children, 17(2), 183-197.

Bonomi

A. E.

Anderson

M. L.

Rivara

F. P.

Thompson

R. S.

(2007). Health outcomes in women with physical and sexual intimate partner violence exposure. Journal of Womens Health, 16(7), 987-997.

Btoush

Campbell

J. C.

Gebbie

K. M.

(2009). Care provided in visits coded for intimate partner violence in a National Survey of Emergency Departments. Womens Health Issues, 19(4), 253-262.

Cattaneo

L. B.

DeLoveh

H. L. M.

(2008). The role of socioeconomic status in helpseeking from hotlines, shelters, and police among a national sample of women experiencing intimate partner violence. American Journal of Orthopsychiatry, 78(4), 413-422.

Cattaneo

L. B.

Stuewig

Goodman

L. A.

Kaltman

Dutton

M. A.

(2007). Longitudinal helpseeking patterns among victims of intimate partner violence: The relationship between legal and extralegal services. American Journal of Orthopsychiatry, 77(3), 467-477.

Centers for Disease Control and Prevention. (2010). Intimate partner violence: Definitions. Atlanta, GA: Author

Chandler

Meisel

Jordan

Rienzi

B. M.

Goodwin

S. N.

(2004). Substance abuse, employment, and welfare tenure. Social Service Review, 78(4), 628-651.

10.

Chang

J. C.

Dado

Hawker

Cluss

P. A.

Buranosky

Slagel

McNeil

Scholle

S. H.

(2010). Understanding turning points in intimate partner violence: Factors and circumstances leading women victims toward change. Journal of Womens Health, 19(2), 251-259.

11.

Cheng

T. C.

(2002). Welfare recipients: How do they become independent? Social Work Research, 26(3), 159-170.

12.

Cheng

T. C.

(2003). Welfare “recidivism” among former welfare recipients. Families in Society, 84(1), 63-74.

13.

Cheng

T. C.

(2005). Recidivism among former recipients who remain unemployed: The vicious circle created by welfare reform. Families in Society, 86(1), 134-142.

14.

Cheng

T. C.

(2010). Financial self-sufficiency or return to welfare? A longitudinal study of mothers among the working poor. International Journal of Social Welfare, 19(2), 162-172.

15.

Cheng

T. C.

C. C.

(2010). Heavy alcohol use, alcohol/drug screening and their relationship to mothers’ welfare participation: A temporal-ordered causal analysis. Journal of Social Policy, 39(4), 543-559.

16.

Cheng

T. C.

McElderry

C. G.

(2007). How do drug use and social relations affect welfare participation? Social Service Review, 81(1), 155-165.

17.

Cook

J. A.

Mock

L. O.

Jonikas

J. A.

Burke-Miller

J. K.

Carter

T. M.

Taylor

Petersen

C. A.

Grey

D. D.

Gruenenfelder

(2009). Prevalence of psychiatric and substance use disorders among single mothers nearing lifetime welfare eligibility limits. Archives of General Psychiatry, 66(3), 249-258.

18.

Erickson

S. R.

Guthrie

VanEtten-Lee

Himle

Hoffman

Santos

S. F.

Janeck

A. S.

Zivin

Abelson

J. L.

(2009). Severity of anxiety and work-related outcomes of patients with anxiety disorders. Depression and Anxiety, 26(12), 1165-1171.

19.

Finkel

S. E.

(1995). Causal analysis with panel data. Thousand Oaks, CA: Sage.

20.

Ford-Gilboe

Wuest

Varcoe

Davies

Merritt-Gray

Campbell

Wilk

(2009). Modelling the effects of intimate partner violence and access to resources on women’s health in the early years after leaving an abusive partner. Social Science & Medicine, 68(6), 1021-1029.

21.

Goodman

L. A.

Smyth

K. F.

Borges

A. M.

Singer

(2009). When crises collide how intimate partner violence and poverty intersect to shape women’s mental health and coping? Trauma, Violence, & Abuse, 10(4), 306-329.

22.

Halpern

C. T.

Spriggs

A. L.

Martin

S. L.

Kupper

L. L.

(2009). Patterns of intimate partner violence victimization from adolescence to young adulthood in a nationally representative sample. Journal of Adolescent Health, 45(5), 508-516.

23.

Hardin

J. W.

Hilbe

J. M.

(2003). Generalized estimating equations. Boca Baton, FL: Chapman & Hall/CRC.

24.

Hollenbeck

Kimmel

(2002). The role of postsecondary education in welfare reform—Ohio’s JOBS student retention program. Evaluation Review, 26(6), 618-644.

25.

Holtzworth-Munroe

(2000). A typology of men who are violent toward their female partners: Making sense of the heterogeneity in husband violence. Current Directions in Psychological Science, 9(4), 140-143.

26.

Holtzworth-Munroe

Meehan

J. C.

Herron

Rehman

Stuart

G. L.

(2000). Testing the Holtzworth-Munroe and Stuart (1994) batterer typology. Journal of Consulting and Clinical Psychology, 68(6), 1000-1019.

27.

Holtzworth-Munroe

Meehan

J. C.

Herron

Rehman

Stuart

G. L.

(2003). Do subtypes of maritally violent men continue to differ over time? Journal of Consulting and Clinical Psychology, 71(4), 728-740.

28.

Inter-University Consortium for Political and Social Research. (2009). Violence against women and the role of welfare reform in Stanislaus and Kern Counties, California, 1992-2002: Description. Ann Arbor, MI: Author.

29.

Johnson

M. P.

(1995). Patriarchal terrorism and common couple violence—2 forms of violence against women. Journal of Marriage and the Family, 57(2), 283-294.

30.

Johnson

M. P.

(2006). Conflict and control—Gender symmetry and asymmetry in domestic violence. Violence against Women, 12(11), 1003-1018.

31.

Johnson

M. P.

(2011). Gender and types of intimate partner violence: A response to an anti-feminist literature review. Aggression and Violent Behavior, 16(4), 289-296.

32.

Kalil

Seefeldt

K. S.

Wang

H. C.

(2002). Sanctions and material hardship under TANF. Social Service Review, 76(4), 642-662.

33.

Keng

S. H.

Garasky

Jensen

H. H.

(2002). Welfare dependence and recidivism in an era of welfare reform. Applied Economics, 34(18), 2311-2323.

34.

Kessler

R. C.

Andrews

Mroczek

Ustun

Wittchen

H.-U.

(1998). The World Health Organization Composite International Diagnostic Interview Short-Form (CIDI-SF). International Journal of Methods in Psychiatric Research, 7(4), 171-185.

35.

Kimerling

Alvarez

Pavao

Mack

K. P.

Smith

M. W.

Baumrind

(2009). Unemployment among women examining the relationship of physical and psychological intimate partner violence and posttraumatic stress disorder. Journal of Interpersonal Violence, 24(3), 450-463.

36.

Kneipp

S. M.

Beeber

Schwartz

Kairalla

J. A.

Lutz

B. J.

Pereira

Hall

A. G.

Flocks

Beeber

Schwartz

(2011). Public health nursing case management for women receiving Temporary Assistance for Needy Families: A randomized controlled trial using community-based participatory research. American Journal of Public Health, 101(9), 1759-1768.

37.

Lindhorst

Meyers

Casey

(2008). Screening for domestic violence in public welfare offices—An analysis of case manager and client interactions. Violence against Women, 14(1), 5-28.

38.

Lindhorst

Oxford

(2008). The long-term effects of intimate partner violence on adolescent mothers’ depressive symptoms. Social Science & Medicine, 66(6), 1322-1333.

39.

Loprest

(2002). Who returns to welfare? (No. B-49). Washington, DC: Urban Institute.

40.

Lown

E. A.

Schmidt

L. A.

Wiley

(2006). Interpersonal violence among women seeking welfare: Unraveling lives. American Journal of Public Health, 96(8), 1409-1415.

41.

MacKinnon

D. P.

Lockwood

C. M.

Hoffman

J. M.

West

S. G.

Sheets

(2002). A comparison of methods to test the significance of the mediated effect. Psychological Methods, 7(1), 83-104.

42.

Macy

R. J.

Ferron

Crosby

(2009). Partner violence and survivors’ chronic health problems: Informing social work practice. Social Work, 54(1), 29-43.

43.

McCloskey

L. A.

Williams

C. M.

Lichter

Gerber

Ganz

M. L.

Sege

(2007). Abused women disclose partner interference with health care: An unrecognized form of battering. Journal of General Internal Medicine, 22(8), 1067-1072.

44.

Mead

L. M.

(2003). Welfare caseload change: An alternative approach. Policy Studies Journal, 31(2), 163-185.

45.

Menard

Turetsky

(1999). Child support enforcement and domestic violence. Juvenile and Family Court Journal, 50(2), 27-38.

46.

Miller

(2006). A specification of the types of intimate partner violence experienced by women in the general population. Violence Against Women, 12(12), 1105-1131.

47.

Montoya

I. D.

Bell

D. C.

Atkinson

J. S.

Nagy

C. W.

Whitsett

D. D.

(2002). Mental health, drug use, and the transition from welfare to work. Journal of Behavioral Health Services & Research, 29(2), 144-156.

48.

Nurius

P. S.

Macy

R. J.

(2010). Person-oriented methods in partner violence research: Distinct biopsychosocial profiles among battered women. Journal of Interpersonal Violence, 25(6), 1064-1093.

49.

Pearson

Griswold

E. A.

Thoennes

(2001). Balancing safety and self- sufficiency—Lessons on serving victims of domestic violence for child support and public assistance agencies. Violence Against Women, 7(2), 176-192.

50.

Prospero

(2008). Effects of masculinity, sex, and control on different types of intimate partner violence perpetration. Journal of Family Violence, 23(7), 639-645.

51.

Pyles

(2006). Toward safety for low-income battered women: Promoting economic justice strategies. Families in Society: The Journal of Contemporary Social Services, 87(1), 63-70.

52.

Rabe-Hesketh

Skrondal

(2008). Multilevel and longitudinal modeling using Stata. College Station, TX: Stata Press.

53.

Rodriguez

Valentine

J. M.

Son

J. B.

Muhammad

(2009). Intimate partner violence and barriers to mental health care for ethnically diverse populations of women. Trauma, Violence, & Abuse, 10(4), 358-374.

54.

Romero

Chavkin

Wise

P. H.

Smith

L. A.

Wood

P. R.

(2002). Welfare to work? Impact of maternal health on employment. American Journal of Public Health, 92(9), 1462-1468.

55.

Saunders

D. G.

Holter

M. C.

Pahl

L. C.

Tolman

R. M.

(2006). Welfare workers’ responses to domestic violence cases: The effects of training and worker characteristics. Families in Society: The Journal of Contemporary Social Services, 87(3), 329-338.

56.

Seefeldt

K. S.

Orzol

S. M.

(2005). Watching the clock tick: Factors associated with TANF accumulation. Social Work Research, 29(4), 215-229.

57.

Staggs

S. L.

Long

S. M.

Mason

G. E.

Krishnan

Riger

(2007). Intimate partner violence, social support and employment in the post–welfare reform era. Journal of Interpersonal Violence, 22(3), 345-367.

58.

Straus

M. A.

Hamby

S. L.

BoneyMcCoy

Sugarman

D. B.

(1996). The revised Conflict Tactics Scales (CTS2): Development and preliminary psychometric data. Journal of Family Issues, 17(3), 283-316.

59.

Swanberg

J. E.

Macke

(2006). Intimate partner violence and the workplace: Consequences and disclosure. Affilia: Journal of Women and Social Work, 21(4), 391-406.

60.

Taylor

M. J.

Barusch

A. S.

(2004). Personal, family, and multiple barriers of long-term welfare recipients. Social Work, 49(2), 175-183.

61.

Tolman

R. M.

Raphael

(2000). A review of research on welfare and domestic violence. Journal of Social Issues, 56(4), 655-681.

62.

U.S. Department of Health & Human Services. (2008). Temporary Assistance for Needy Families program (TANF): Eighth annual report to Congress. Washington, DC: Author.

63.

UCLA Academic Technology Services. (2010). How to perform Sobel-Goodman mediation tests in Stata? Retrieved from http://www.ats.ucla.edu/stat/fAQ/sgmediation.htm

64.

Walker

L. E.

(2000). The battered woman syndrome. New York, NY: Springer.

65.

Ware

J. E.

Kosinski

Keller

S. D.

(1996). A 12-item short-form health survey: Construction of scales and preliminary tests of reliability and validity. Medical Care, 34(3), 220-233.

66.

Wuest

Ford-Gilboe

Merritt-Gray

Varcoe

Lent

Wilk

Campbell

(2009). Abuse-related injury and symptoms of posttraumatic stress disorder as mechanisms of chronic pain in survivors of intimate partner violence. Pain Medicine, 10(4), 739-747.

67.

Yoshihama

Hammock

A. C.

Horrocks

(2006). Intimate partner violence, welfare receipt, and health status of low-income African American women: A lifecourse analysis. American Journal of Community Psychology, 37(1-2), 95-109.