Experimental Approaches for Improving the Assessment of Partner Violence

Abstract

Intimate partner violence (IPV) is a serious and widespread problem. Unfortunately, like many other private behaviors, self-report is one of the few practical means of assessing IPV. Despite its known limitations, few studies have examined methods for improving the assessment of IPV through self-report, particularly in research contexts. The current study examined implicit goal priming (IGP) and restricted response latencies (RRL), which have previously been used successfully to improve reporting of other types of sensitive information. Participants in the IGP condition engaged in a word-matching task intended to subliminally prime an honesty goal immediately before completing the IPV self-report measure. In the RRL condition, the amount of time participants were allotted to answer each question was restricted to 3 s. Using a pilot sample of 71 cohabiting/married couples, we currently provide experimental support for the utility of the IGP task for increasing the frequency of reported IPV, as well as agreement between self-reports and partner reports, particularly when used in conjunction with the RRL method. We found the strongest and most consistent effects for women’s reports of severe IPV perpetration and victimization, which is important given the extreme nature of these forms of IPV and because they are frequently underreported by women. We also found evidence to support the use of these methods to enhance reporting of other forms of IPV and provide additional directions for future research.

Keywords

intimate partner violence agreement implicit goal priming restricted response latencies

Studies of intimate partner violence (IPV) provide information critical to improving intervention and prevention efforts. However, our understanding of the causes, consequences, and treatment of IPV hinges on collecting accurate data on its prevalence and chronicity. Yet, for scientific, pragmatic, and ethical reasons, the measurement of IPV is not amenable to direct observation. Although there exist some other avenues for independent measurement (e.g., police reports), these indices may also contain bias (Erez & Belknap, 1998) or are otherwise unable to capture important facets of IPV, such as the chronicity of violence or legal forms of IPV (e.g., psychological; Erez, 2002; Ruiz-Pérez, Plazaola-Castaño, & Vives-Cases, 2007). Thus, for several reasons, self-report surveys are the measurement tools of choice for many applications.

Unfortunately, individuals are often tentative to answer questions about sensitive topics (e.g., elective abortions; Fu, Darroch, Henshaw, & Kolb, 1998) and may provide inaccurate information that leads to data containing systematic biases (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). It is therefore unsurprising that, in many circumstances, substantive instances of IPV are to varying degrees underreported by perpetrators and victims. Studies employing multimodal self-report assessment (i.e., written, questionnaire, interview; O’Leary, Vivian, & Malone, 1992), comparing self-reports with official police reports (Fleury, Sullivan, Bybee, & Davidson, 1998), using advanced statistical correction techniques for missing data (Emery, 2010), and examining social desirability (Riggs, Murphy, & O’Leary, 1989; Sugarman & Hotaling, 1997) provide converging evidence for underreporting. Correspondingly, similarity between partners’ reports is often unacceptably low vis-a-vis recognized standards for agreement (e.g., Fleiss, Levin, & Paik, 2003), which undermines their validity (Follingstad & Rogers, 2013). Although available research strongly indicates that self-reported IPV data are compromised by multiple sources of systematic reporting bias, which poses real problems for developing an accurate science around IPV, studies evaluating more viable alternatives are presently scarce.

To this backdrop, we applied to the assessment of IPV two experimental paradigms shown to improve the quality of sensitive self-report data, namely, implicit goal priming (IGP) and restricted response latencies (RRL). IGP tasks direct behaviors toward a particular outcome by subliminally activating a goal using word searches and sentence unscrambling, which are shown to significantly increase the reporting of socially undesirable behaviors. For example, in a study by Rasinski, Visser, Zagatsky, and Rickett (2005), participants admitted to more problematic alcohol use (e.g., binge drinking, losing consciousness) after an IGP manipulation than participants in the control condition.

The use of RRL is based on the notion that lying takes more time and cognitive effort than honest responding. Increasing the rate at which participants have to respond interferes with the processes required for dissimulation and thus can promote more honest responding for typically underreported phenomena (Zuckerman, DePaulo, & Rosenthal, 1981). For instance, Gerstein, Barke, and Johnson (1987) found that giving individuals an allotted amount of time to answer questions on a self-report measure of psychopathology effectively increased honest responding. Furthermore, and specifically supporting cognitive load as the mechanism driving this effect, Komar, Komar, Robie, and Taggar (2010) found that imposing a time restriction on a personality test decreased the level of socially desirable responding for those lower in general cognitive ability.

As both the IGP and RRL methods are shown to improve self-reports of sensitive data, their ability to enhance the accuracy of self-reported IPV perpetration and victimization in controlled research studies appears worthy of preliminary investigation. Consistent with prior research in this area, we used indirect methods of evaluating whether the IGP and RRL approaches improved the accuracy of self-reported IPV. Specifically, we examined if these methods, as compared with standard administration, would (a) yield an increase in the reported frequency of IPV and (b) produce significantly reliable partner reports. With respect to the former, IPV is a socially stigmatized behavior and data suggest that it is consistently underreported (e.g., Simpson & Christensen, 2005). We therefore assumed that increasing the accuracy of the reports would correspond to an increase in the frequency at which these behaviors are reported. Notably, there is evidence that certain factors may lead IPV to be over-reported on survey measures, such as when participants include playful or mock violence in their responses (e.g., Lehrner & Allen, 2014) or when contextual factors specific to the individual are in play (e.g., custody proceedings; Gadd, Farrall, Dallimore, & Lombard, 2002). Although such influences may presently create statistical noise, we reasoned that they would be distributed equally across conditions and unlikely to account for any systematic differences in reported frequencies of IPV observed therein. Nevertheless, we looked for convergence of effects with our other criteria of increased agreement across partner reports. If the IGP and RRL approaches indeed elicit more accurate responses, then partner reports should be significantly reliable using these approaches, and significantly more so than reports from couples under standard administration.

Method

Participants

We recruited 71 couples from two counties in Alabama using Christopher, Pflieger, Canary, Guerrero, & Holtzworth-Munroe’s (2008) targeted sampling approach. Specifically, participants from neighborhoods that experienced higher levels of domestic abuse incidents, as indicated by local law enforcement data, were more heavily sampled using flyers and targeted mailing. The flyers, which provided broad details about the study and participant remuneration, were addressed to the “Current Resident” and directed potential participants to call the lab or visit the website to sign up for the study. In either case, interested couples were screened for eligibility via phone by an undergraduate research assistant. Couples were required to be at least 19 years of age; heterosexual; in a committed relationship (6 months), married, or engaged; and cohabitating with their partner. Couples in treatment or in the process of divorce or separation were ineligible for the study. The sample was predominantly Caucasian (81.7% and 77.5% of women and men). The mean age of men was 39.7 (SD = 14.5) and 37.9 (SD = 14.3) for women. The majority of couples were married (70.4%) and without children (56.9%). About 60.5% of men and 66.1% of women reported having at least a bachelor’s degree. The average monthly income for women and men was US$2,418 (SD = US$5,582) and US$3,120 (SD = US$2,871), respectively.

Procedure

Interested couples who met eligibility requirements were scheduled for a laboratory visit and provided instructions on how to participate. Upon arrival, couples were randomly assigned to the IGP condition (n = 17), the RRL condition (n = 19), the combined IGP and RRL condition (n = 19), or the control group (n = 16). Research assistants led partners to separate rooms to complete the informed consent forms and study procedures. During the informed consent, the research assistants emphasized that the study was completely voluntary and that participants could discontinue at any point without penalty. Furthermore, participants were informed about confidentiality in verbal and written formats. Participants were told that the study rooms were soundproofed and that the information provided during the study would be de-identified and would not be available to their partner. Consistent with Rasinski et al. (2005), participants in the IGP condition were presented with six target words—four of which were related to honesty (e.g., genuine). They were then presented with three synonyms of the target word (e.g., sincere, open, truthful) and were asked to choose the one that most closely matched their definition of the target. Participants in the control condition engaged in a similar task but were presented with six target words and synonyms unrelated to the goal of honesty. In both the RRL and control conditions, the respondents were instructed to answer the questions as quickly and accurately as possible, but participants in the RRL condition were given only 3 s to respond. All participants assessed their own and their partner’s physically and psychologically violent behaviors using the 12-item Physical Assault and eight-item Psychological Aggression subscales of a computer-adapted version of the Revised Conflict Tactics Scale (CTS2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). Participants were presented with items one at a time and were asked to indicate whether each violent event had occurred in the past month using the designated keys for True (Q) and False (P), which were randomized across items to prevent drift. If participants indicated that a specific violent behavior occurred, they were then asked to indicate how often it occurred in the past month on a 4-point scale (1 = once or twice, 2 = 3-5 times, 3 = 6-10 times, 4 = more than 10 times). We constructed prevalence and chronicity indices using these data. The prevalence index was constructed from the responses to the true or false question and indicates whether or not the participant endorsed that a specific violent behavior occurred at least once in the past month, whereas the chronicity variable was constructed from the responses to the question about how often a specific violent behavior had occurred and indicates the frequency of that violent behavior. In the latter case, the CTS2 was scored by summing the midpoints of each response (e.g., 3 to 5 times = 4). Consistent with CTS2 scoring (Straus et al., 1996), we used 15 as the midpoint score when participants endorsed “more than 10 times.”

Based on evidence that violence severity (e.g., Cunradi, Bersamin, & Ames, 2009) and gender of the perpetrator and victim (e.g., Chan, 2011) are relevant to reporting biases, we decided to analyze reports separately across female-to-male and male-to-female perpetrated severe and minor acts of violence. All items on both the Physical and Psychological Aggression subscales were classified as minor (e.g., shouted or yelled at my partner, pushed or shoved my partner) or severe aggression (e.g., called my partner fat or ugly, choked my partner) in accordance with Straus et al.’s (1996) recommendations. In the present sample, there were 120 occurrences of male-to-female minor violence and 121 occurrences of female-to-male minor violence. The chronicity of male-to-female minor violence was 257, and 303 for female-to-male violence. There were 27 and 30 occurrences of male-to-female and female-to-male severe violence, respectively. The chronicity of male-to-female severe violence was 38, and 56 for female-to-male severe violence.

Following completion of the CTS2, participants responded to computer-administered self-report measures of dyadic constructs unrelated to the current analyses and a demographic assessment. After completing the questionnaires, each participant was given a list of referrals for couples counseling and contact information for the principle investigator for the purposes of debriefing, as well as US$40 in remuneration for their participation in the study. This study was approved by the University’s institutional review board (IRB) prior to execution.

All data were collected through E-prime and exported into SPSS for data analysis. E-prime is a computer platform for experimental design and data collection. A majority of participant data was complete and no participant was missing more than two items on the CTS2. Missing CTS2 items were quantified with 0s. This approach was predicated on the rationale that, if the violence in question was intentionally omitted, then the omission was most akin to responding that the act did not occur (i.e., a response of 0).

Results

Did Frequencies of Reported IPV Differ Across Conditions?

We first tested whether respondents reported higher prevalence and chronicity of psychological and physical violence in the experimental conditions when compared with the control conditions using regression models appropriate for count data. Fit indices suggested a Poisson distribution was the best fit for all models save one, for which a negative binomial regression was used to account for overdispersion. We used three dummy coded variables to identify membership in one of the three experimental groups (IGP condition = 1, otherwise 0; RRL condition = 1, otherwise 0; IGP/RRL condition = 1, otherwise 0), entering in these variables as predictors of IPV reports. The regression weights for the dummy variables test whether the expected frequency of reports in the IGP, RRL, or combined group differ significantly from the intercept value, which in this case represents the expected frequency of reported IPV in the control group. As regression weights are in natural log units, we also report exp^(b) values, which translate the effects back into the expected change in the number of reported events relative to the control condition and thus represent the extent to which the experimental manipulations yielded a change in reported frequencies of IPV. All assumptions of the models were currently met (e.g., independence of errors). Residual plots and the likelihood ratio χ² test implied that the models were overall adequate (e.g., see Coxe, West, & Aiken, 2009).

As shown in Table 1, we found that women in the combined IGP/RRL group reported significantly higher IPV with respect to severe perpetration and victimization prevalence and chronicity and minor perpetration chronicity, with the size of these effects being rather large. We also found that—relative to women in the control condition—those in the RRL group reported higher counts of chronic severe and minor perpetration and minor victimization, whereas those in the IGP condition reported greater frequencies of chronic severe and minor IPV perpetration and minor victimization. For men’s reports, shown in Table 2, we found only that the experimental conditions produced lower reports for minor victimization chronicity and the combined IGP/RRL condition produced lower chronicity reports for minor perpetration, contrary to hypotheses. Considered overall, these findings provide some support for the use of the IGP and RRL methods—both alone or in tandem—to increase the frequency at which IPV is self-reported, but these effects varied depending on the form of IPV being reported and by whom. Specifically, the results appeared more particular to women’s reports of both perpetration and victimization, with the strongest effects being observed for severe acts of IPV in the combined IGP/RRL condition.

Table 1.

Poisson Regression Estimates Predicting Severe or Minor IPV Prevalence or Chronicity Reports by Experimental Group for Women (N = 71).

	Prevalence				Chronicity
	Minor		Severe		Minor		Severe
	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}
Victimization
Control (Int.)	0.17 (0.23)	1.19 [0.76, 1.86]	−2.77 (1.00)**	0.06 [0.01, 0.44]	0.37 (0.20)^†	1.44 [0.98, 2.12]	−2.20 (0.71)**	0.11 [0.03, 0.44]
IGP	−0.06 (0.32)	.94 [0.50, 1.78]	0.63 (1.22)	1.88 [0.17, 20.76]	0.79 (0.24)**	2.20 [1.38, 3.51]	0.97 (0.84)	2.65 [0.49, 12.89]
RRL	0.03 (0.31)	1.03 [0.56, 1.90]	1.49 (1.10)	4.44 [0.52, 38.04]	0.77 (0.24)**	2.15 [1.35, 3.43]	0.92 (0.84)	2.50 [0.51, 13.64]
IGP/RRL	0.04 (0.32)	1.04 [0.56, 1.93]	2.34 (1.04)*	10.35 [1.34, 80.18]	0.35 (0.26)	1.42 [0.86, 2.35]	1.87 (0.76)*	6.50 [1.47, 28.80]
LR χ²		0.12		11.96**		16.23**		9.90*
Perpetration
Control (Int.)	0.41 (0.20)*	1.50 [1.01, 2.24]	−2.77 (1.00)**	0.06 [0.01, 0.44]	0.51 (0.18)**	1.67 [1.17, 2.40]	−2.89 (1.00)**	0.06 [0.01, 0.39]
IGP	0.09 (0.28)	1.10 [0.64, 1.89]	1.33 (1.12)	3.77 [0.42, 33.68]	0.86 (0.22)**	2.37 [1.54, 3.64]	2.14 (1.05)*	8.47 [1.06, 67.73]
RRL	−0.12 (0.29)	0.89 [0.51, 1.57]	1.27 (1.12)	3.56 [0.40, 31.81]	0.77 (0.22)**	2.17 [1.41, 3.34]	2.20 (1.06)*	9.00 [1.14, 71.04]
IGP/RRL	−0.21 (0.30)	0.82 [0.46, 1.50]	2.19 (1.05)*	8.89 [1.14, 69.44]	0.61 (0.23)**	1.83 [1.18, 2.86]	2.57 (1.04)*	13.00 [1.71, 99.38]
LR χ²		1.30		8.10*		19.34**		12.60**

Note. exp^(b) value represents the expected number of reported instances of IPV by participants relative to the control condition. The 95% confidence interval for the exp^(b) value is reported. IPV = intimate partner violence; IGP = implicit goal priming; RRL = restricted response latencies; LR = likelihood ratio.

†

p < .10. *p < .05. **p < .01.

Table 2.

Poisson Regression Estimates Predicting Severe or Minor IPV Prevalence or Chronicity Reports by Experimental Group for Men (N = 71).

	Prevalence				Chronicity
	Minor		Severe		Minor		Severe^a
	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}	b (SE)	exp^{( b )}
Victimization
Control (Int.)	0.11 (0.22)	1.11 [0.72, 1.72]	−2.20 (0.71)**	0.11 [0.03, 0.44]	1.28 (0.12)**	3.61 [2.83, 4.61]	−1.50 (0.55)**	0.22 [0.08, 0.66]
IGP	−0.37 (0.36)	0.69 [0.34, 1.40]	0.46 (0.91)	1.60 [0.77, 17.8]	−1.07 (0.25)**	0.34 [0.21, 0.56]	0.75 (0.70)	2.12 [0.54, 8.34]
RRL	0.01 (0.32)	1.01 [0.54, 1.89]	1.31 (0.80)	3.71 [0.27, 9.5]	−0.77 (0.22)**	0.46 [0.30, 0.71]	0.81 (0.69)	2.25 [0.58, 8.70]
IGP/RRL	−0.31 (0.36)	0.73 [0.36, 1.50]	1.22 (0.82)	3.40 [0.68, 16.72]	−0.88 (0.23)**	0.42 [0.27, 0.65]	0.92 (0.68)	2.50 [0.67, 9.50]
LR χ²		1.93		4.46		28.10**		2.21
Perpetration
Control (Int.)	0.00 (0.24)	1.00 [0.63, 1.60]	−1.79 (0.57)**	0.17 [0.05, 0.51]	0.87 (0.15)**	2.39 [1.78, 3.22]	−0.81 (0.35)*	0.44 [0.22, 0.89]
IGP	0.46 (0.30)	1.59 [0.88, 2.89]	0.35 (0.76)	1.41 [0.32, 6.31]	0.06 (0.22)	1.06 [0.69, 1.61]	0.06 (0.50)	1.06 [0.40, 2.82]
RRL	0.16 (0.32)	1.18 [0.62, 2.22]	−0.35 (0.91)	0.71 [0.12, 4.22]	−0.12 (0.22)	0.88 [0.57, 1.40]	−1.39 (0.79)^†	0.25 [0.05, 1.18]
IGP/RRL	0.32 (0.32)	1.38 [0.88, 2.89]	0.81 (0.71)	2.25 [0.56, 9.00]	−0.58 (0.25)*	0.56 [0.34, 0.92]	−0.47 (0.57)	0.63 [0.20, 1.91]
LR χ²		2.62		2.69		7.87*		5.22

The negative binomial regression.

†

p < .10. *p < .05. **p < .01.

Does Partner Agreement Differ Across Conditions?

We next examined agreement across partner reports in each condition using one-way random effects intraclass correlations (ICCs), which were estimated separately for the control group and for each of the three experimental groups. We first tested each of these values against the null hypothesis that agreement across partners within each group was zero, and then applied the F test developed by Feldt, Woodruff, and Salih (1987) to compare ICC estimates across groups. Based on prior research (e.g., Armstrong, Wernke, Medina, & Schafer, 2002), we did not expect that estimates of partner agreement would be significantly different than chance (i.e., 0) in the control condition. However, we expected that partner agreement from couples in the experimental conditions would be significantly different than chance and that there would be significantly higher agreement in comparison with reporting from control participants. Table 3 provides the ICCs for male-to-female and female-to-male violence reports.

Table 3.

Intraclass Correlations and 95% Confidence Intervals for Female-to-Male (Top) and Male-to-Female (Bottom) Violence (N = 71).

	Control	IGP	RRL	RRL/IGP
Female-to-male IPV
Severe violence
Prevalence	−0.07 [–0.53, 0.42]_d	0.07 [–0.40, 0.52]	0.09 [–0.39, 0.53]	0.85 [0.64, 0.95]_a**
Chronicity	−0.04 [–0.51, 0.44]	0.04 [–0.43, 0.49]	−0.15 [–0.55, 0.31]	0.49 [0.07, 0.77]*
Minor violence
Prevalence	0.34 [–0.16, 0.70]^†	0.38 [–0.10, 0.72]^†	0.53 [0.09, 0.80]*	0.71 [0.36, 0.89]**
Chronicity	0.06 [–0.43, 0.52]	0.25 [–0.24, 0.64]	0.20 [–0.26, 0.59]	0.34 [–0.11, 0.68]^†
Male-to-female IPV
Minor violence
Prevalence	0.30 [–0.20, 0.68]_b	0.85 [0.65, 0.94]_ac**	0.10 [–0.37, 0.54]_b	0.51 [0.05, 0.80]*
Chronicity	0.05 [–0.43, 0.51]_b	0.84 [0.63, 0.94]_ac**	0.09 [–0.36, 0.51]_b	0.42 [–0.02, 0.73]*
Severe violence
Prevalence	0.14 [–0.36, 0.58]_d	0.50 [0.05, 0.78]_d*	0.26 [–0.23, 0.65]_d	0.92 [0.78, 0.97]_abc**
Chronicity	0.16 [–0.34, 0.59]	0.37 [–0.11, 0.71]^†	−0.04 [–0.46, 0.41]_d	0.72 [0.42, 0.88]_c**

Note. Subscripts indicate which reliabilities are significantly different from each other at p < .05. a = control; b = implicit goal priming (IGP); c = restricted response latency (RRL); and d = combined IGP/RRL. IPV = intimate partner violence.

†

p < .10. *p < .05. **p < .01.

As anticipated, reliabilities were not significant in the control condition for male-to-female and female-to-male severe and minor IPV. In contrast, the results for the IGP and combined IGP/RRL conditions were generally consistent with our expectations. In particular, the estimates for male-to-female minor and severe violence prevalence and chronicity were significantly reliable in the IGP and combined IGP/RRL conditions (except for the minor chronicity reports in the IGP condition). For female-to-male violence, the severe and minor prevalence and severe chronicity reports were significantly reliable in the combined IGP/RRL condition, whereas the RRL alone condition produced significantly reliable minor prevalence reports. In addition, male-to-female minor and severe violence chronicity in the combined condition was higher than in the control condition, F(15, 18) = 3.04, p < .05, and F(15, 16) = 6.01, ps < .01. Furthermore, ICCs for male-to-female and female-to-male severe violence prevalence were higher than in the control condition for the IGP and combined IGP/RRL condition, respectively, F(15, 16) = 4.78 and F(15, 15) = 7.29, ps < .01. Considered as a set, these results suggest that the IGP task may be particularly effective in bolstering reliabilities for a variety of IPV domains, especially when used in conjunction with RRL.

Discussion

IPV survey data are compromised by systematic reporting biases that can result in attenuated partner agreement about IPV prevalence and chronicity and thus place inferential limitations on findings derived therefrom (Caetano, Field, Ramisetty-Mikler, & Lipsky, 2009). Given the importance of data on IPV, there is a critical need to explore methods that have the potential to improve the accuracy of self-reported violence, particularly in research contexts. Although some evidence supports more elaborate data collection methods (e.g., daily diaries; Derrick, Testa, & Leonard, 2014), our findings provide preliminary support that goal priming, in particular, may provide an expeditious and cost-effective alternative to improve the assessment of IPV through self-reports. In addition, the pattern of effects indicated that the experimental procedures may have more of an impact on reports of violence frequency.

We also found unique reporting effects when we combined the RRL with goal priming tasks. This latter finding is also important by itself, as this is the first study to our knowledge that evaluated the use of these methods in conjunction. Although it is possible that these manipulations operated independently, contributing only additive effects, we speculate that restricting response times within and across trials may have increased cognitive load, which taxed higher order processes needed to screen out extraneous influences on responding (e.g., working memory; Lavie, Hirst, De Fockert, & Viding, 2004). This may have generated a higher vulnerability to unconscious influences for those in the combined condition, relative to those receiving only the honesty goal primes.

It is also notable that the goal priming conditions had the most pronounced effects for women’s reported frequency of severe victimization and perpetration, which appeared to be driving the significantly higher reliabilities found at the couple level, as compared with controls. This may reflect that women’s reports of severe IPV are more highly subject to reporting biases, whereas the effects for more openly reported forms of IPV were presently harder to detect. Indeed, the underreporting of severe female victimization is consistent with research that suggests serious forms of cyclical and premeditated violence (e.g., “intimate terrorism”) are often denied by women participating in quantitative research studies (Emery, 2010; Johnson, 1995). Moreover, women may also underreport their severe perpetration because women who violate prescribed gender roles by engaging in violent criminal behaviors may be punished more harshly by the criminal justice system than men (Crew, 1991; Jenkins & Davidson, 1990; Kingsnorth & MacIntosh, 2007). In any case, these results provide converging evidence across criteria that the IGP/RRL combination successfully enhanced the reporting accuracy of highly deleterious acts of IPV that are, in particular, frequently denied by women in quantitative research studies.

Our ICC analyses also showed that, relative to controls, the experimental manipulations bolstered agreement across a wide range of behaviors. This implies that the experimental conditions led at least some couples to report more acts of IPV than they otherwise may have under standard administration. However, it deserves mention that these effects did not correspond to higher frequencies of reported IPV in our regression models, which was perhaps due to statistical power issues and/or resulted because the base rate of IPV for control participants may have been a priori higher for particular forms of violence. As such, these latter findings are encouraging, but warrant further investigation using larger samples with sufficient power to detect modest effects and wherein preexisting group differences will likely wash out through random assignment. Likewise, the coefficients for male reports in the IGP and combined IGP/RRL conditions indicated that there were fewer reports of IPV in the experimental conditions relative to control condition, which was contrary to our expectations. However, this did not correspond to increased agreement in the control condition relative to the experimental groups, suggesting that this difference may less likely reflect less accurate responding for the experimental groups, but rather preexisting group differences in the frequency of the form of IPV as compared with controls.

Future research can build on the current study in a number of ways. Although we recruited couples most likely to be at risk for IPV through a targeted sampling approach (Christopher et al., 2008), the sample was atypical in some respects (e.g., more than half of couples did not have children), relatively small, and somewhat homogeneous on several demographic variables. Studies that have included diverse samples have found significant differences in IPV rates and agreement based on ethnicity, sexuality, socioeconomic status, and education (Burke & Follingstad, 1999; Caetano et al., 2009; Cunradi et al., 2009; Heyman & Schlee, 1997), indicating that future studies are required before generalizing the present effects to diverse samples. Furthermore, a larger sample size would have made it possible to analyze agreement at the level of violence subtype (e.g., physical, psychological, sexual, and injury) and item across gender. Given evidence that men and women underreport differentially based on the nature of the behavior in question (e.g., sexual perpetration; O’Leary & Williams, 2006), it is likely that these factors will influence reporting and may correspondingly modulate the effects of goal priming and RRL. Future studies may also examine how IGP and RRL would perform in other research samples and contexts (e.g., evaluation of batterer intervention programs) or modify the IGP and RRL tasks to optimize their effects, such as through multiple administrations of the IGP throughout IPV reporting. Finally, our conclusions rest on an assumption that increased reported frequencies and ICC coefficients equate to the more accurate reporting of meaningful interpersonal violence. Although reasonable, additional research in this area may benefit from a multi-method approach to assessment (e.g., individual interviews), which could provide additional reference points against which to verify the instances of IPV being reported on surveys.

Considered overall, the present study highlights two experimental manipulations that may be used to improve the quality of IPV self-reports in research settings. It is important to make such efforts because IPV data are most commonly collected by survey and are used to develop and test theories about IPV perpetration, victimization, and its consequences and treatment. To our knowledge, this is the first study to examine goal priming and RRL as methods for increasing the accuracy of IPV reports. Using an experimental design and targeted sampling, we found that the IGP and RRL may be particularly beneficial in improving the quality of sensitive self-report data when used in combination, which was presently the case with respect to women’s reports of IPV, especially for more extreme forms of violence that are likely to be underreported. More research is necessary to ascertain whether IGP and RRL tasks can be effectively and reliably used to increase the accuracy of IPV reports. However, our results provide some indication that IPV reporting was generally improved and, overall, support efforts to continue exploration of these methods for improving the quality of IPV self-report data in larger and more diverse samples.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a Competitive Research Grant (CRG) sponsored by Auburn University, wherein the research was carried out.

Author Biographies

Edwin G. Ortiz, MS, is a graduate student in the clinical psychology program at Binghamton University. His research focuses on improving assessment of intimate partner violence (IPV) and examining perceptions of perpetrators and victims of IPV. He has a particular interest in examining couples who identify with a socially stigmatized identity. For example, his dissertation will focus on the role of stereotypes and implicit bias on judgments of Black and interracial couples in an IPV vignette. His work will continue to examine the unique influence of stigma on diverse couples experiencing IPV to tailor interventions to these populations.

Richard E. Mattson’s, PhD, research aims are to (a) improve the conceptualization and assessment of satisfaction in relationships and (b) explore the causes and consequences of functional versus dysfunctional relationship communication, including social support and IPV, respectively. With respect to the latter, his foci are on (a) the interplay of IPV and substance and alcohol use, (b) examining ways to improve the self-report validity of IPV measures, and (c) situational and attitudinal factors that influence perceptions of IPV. He is also focused on how genetic, neuroendocrinological, and early environmental factors moderate the role of partner social support in buffering the impact of stress on health.

References

Armstrong

T. G.

Wernke

J. Y.

Medina

K. L.

Schafer

(2002). Do partners agree about the occurrence of intimate partner violence? A review of the current literature. Trauma, Violence, & Abuse, 3, 181-193.

Burke

L. K.

Follingstad

D. R.

(1999). Violence in lesbian and gay relationships: Theory, prevalence, and correlational factors. Clinical Psychology Review, 19, 487-512.

Caetano

Field

Ramisetty-Mikler

Lipsky

(2009). Agreement on reporting of physical, psychological, and sexual violence among White, Black, and Hispanic couples in the United States. Journal of Interpersonal Violence, 24, 1318-1337.

Chan

K. L.

(2011). Gender differences in self-reports of intimate partner violence: A review. Aggression and Violent Behavior, 16, 167-175.

Christopher

F. S.

Pflieger

J. C.

Canary

D. J.

Guerrero

L. K.

Holtzworth-Munroe

(2008). Targeted neighborhood sampling: A new approach for recruiting abusive couples. Journal of Family Violence, 23, 89-100.

Coxe

West

S. G.

Aiken

L. S.

(2009). The analysis of count data: A gentle introduction to Poisson regression and its alternatives. Journal of Personality Assessment, 91, 121-136.

Crew

B. K.

(1991). Sex differences in criminal sentencing: Chivalry or patriarchy? Justice Quarterly, 8, 59-83.

Cunradi

C. B.

Bersamin

Ames

(2009). Agreement on intimate partner violence among a sample of blue-collar couples. Journal of Interpersonal Violence, 24, 551-568.

Derrick

J. L.

Testa

Leonard

K. E.

(2014). Daily reports of intimate partner verbal aggression by self and partner: Short-term consequences and implications for measurement. Psychology of Violence, 4, 416-431.

10.

Emery

C. R.

(2010). Examining an extension of Johnson’s hypothesis: Is male perpetrated intimate partner violence more underreported than female violence? Journal of Family Violence, 25, 173-181.

11.

Erez

(2002). Domestic violence and the criminal justice system: An overview. Online Journal of Issues in Nursing, 7(1), 1-24.

12.

Erez

Belknap

(1998). In their own words: Battered women’s assessment of the criminal processing system’s responses. Violence and Victims, 13, 251-268.

13.

Feldt

L. S.

Woodruff

D. J.

Salih

F. A.

(1987). Statistical inference for coefficient alpha. Applied Psychological Measurement, 11, 93-103.

14.

Fleiss

J. L.

Levin

Paik

M. C.

(2003). Statistical methods for rates and proportions (3rd ed.). Hoboken, NJ: John Wiley & Sons.

15.

Fleury

R. E.

Sullivan

C. M.

Bybee

D. I.

Davidson

W. S.

II . (1998). “Why don’t they just call the cops?” Reasons for differential police contact among women with abusive partners. Violence and Victims, 13, 333-346.

16.

Follingstad

D. R.

Rogers

M. J.

(2013). Validity concerns in the measurement of women’s and men’s report of intimate partner violence. Sex Roles, 69, 149-167.

17.

Darroch

J. E.

Henshaw

S. K.

Kolb

(1998). Measuring the extent of abortion underreporting in the 1995 National Survey of Family Growth. Family Planning Perspectives, 30, 128-138.

18.

Gadd

Farrall

Dallimore

Lombard

(2003). Male victims of domestic violence. Criminal Justice Matters, 53, 16-17.

19.

Gerstein

L. H.

Barke

C. R.

Johnson

S. D.

(1987, August). The integrity interview: A valid and reliable alternative to other honesty devices. Paper presented at the 95th Annual Convention of the American Psychological Association, New York, NY.

20.

Heyman

R. E.

Schlee

K. A.

(1997). Toward a better estimate of the prevalence of partner abuse: Adjusting rates based on the sensitivity of the Conflict Tactics Scale. Journal of Family Psychology, 11, 332-338.

21.

Jenkins

Davidson

(1990). Battered women in the criminal justice system: An analysis of gender stereotypes. Behavioral Sciences & the Law, 8, 161-170.

22.

Johnson

M. P.

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

23.

Kingsnorth

R. F.

MacIntosh

R. C.

(2007). Intimate partner violence: The role of suspect gender in prosecutorial decision-making. Justice Quarterly, 24, 460-495.

24.

Komar

J. A.

Robie

Taggar

(2010). Speeding personality measures to reduce faking. Journal of Personnel Psychology, 9, 126-137.

25.

Lavie

Hirst

De Fockert

J. W.

Viding

(2004). Load theory of selective attention and cognitive control. Journal of Experimental Psychology: General, 133, 339-354.

26.

Lehrner

Allen

N. E.

(2014). Construct validity of the Conflict Tactics Scales: A mixed-method investigation of women’s intimate partner violence. Psychology of Violence, 4, 477-490.

27.

O’Leary

K. D.

Vivian

Malone

(1992). Assessment of physical aggression against women in marriage: The need for multimodal assessment. Behavioral Assessment, 14, 5-14.

28.

O’Leary

K. D.

Williams

M. C.

(2006). Agreement about acts of aggression in marriage. Journal of Family Psychology, 20, 656-662.

29.

Podsakoff

P. M.

MacKenzie

S. B.

Lee

J. Y.

Podsakoff

N. P.

(2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879-903.

30.

Rasinski

K. A.

Visser

P. S.

Zagatsky

Rickett

E. M.

(2005). Using implicit goal priming to improve the quality of self-report data. Journal of Experimental Social Psychology, 41, 321-327.

31.

Riggs

D. S.

Murphy

C. M.

O’Leary

K. D.

(1989). Intentional falsification in reports of interpartner aggression. Journal of Interpersonal Violence, 4, 220-232.

32.

Ruiz-Pérez

Plazaola-Castaño

Vives-Cases

(2007). Methodological issues in the study of violence against women. Journal of Epidemiology and Community Health, 61(Suppl. 2), ii26-ii31.

33.

Simpson

L. E.

Christensen

(2005). Spousal agreement regarding relationship aggression on the Conflict Tactics Scale-2. Psychological Assessment, 17, 423-432.

34.

Straus

M. A.

Hamby

S. L.

Boney-McCoy

Sugarman

D. B.

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

35.

Sugarman

D. B.

Hotaling

G. T.

(1997). Intimate violence and social desirability: A meta-analytic review. Journal of Interpersonal Violence, 12, 275-290.

36.

Zuckerman

DePaulo

B. M.

Rosenthal

(1981). Verbal and nonverbal communication of deception. Advances in Experimental Social Psychology, 14, 1-59.