Abstract
The Measure of Psychologically Abusive Behaviors (MPAB) was developed recently to improve on prior scales. Two nationally solicited samples of women were utilized to further validate the MPAB through the use of a criterion group, factor analysis, and designation of a threshold for accurate classification. MPAB scores were twice as high in the criterion group (M = 38.75 vs. M = 18.85; F = 22.17). Using a cutpoint of 1, the sensitivity and specificity for MPAB were .725 and .628, respectively. Cronbach’s alpha was .97 for both samples. These data provide additional evidence for the MPAB as valid and internally consistent.
The assessment of intimate partner violence (IPV) began with the introduction of the Conflict Tactics Scale (CTS) by Straus (1979), which was a checklist of primarily physically aggressive behaviors used against an intimate partner. Over the years, scales were subsequently developed for the same purpose that varied in terms of whether they included other forms of IPV in addition to physical abuse, and whether reliability and validity information were presented to support the use of the scales. Scales measuring only psychological abuse (ΨA) also emerged, indicating that researchers viewed this form of IPV as an independent field of inquiry in its own right (see O’Leary, 2001).
Although a special issue of Violence and Victims (1999) presented numerous articles addressing measurement issues of psychological abuse, Maiuro (2001) cited many problems that suggested we cannot consider this phenomenon to be effectively measured. Specifically, he mentioned a lack of reliable norms, no thresholds for determining someone is psychologically abused, the need for context, the need to differentiate psychological abuse from couple conflict, the lack of data on couples, and the need to assess intent and motivation. Historically, reliability estimates have often been used as the criterion for acceptability of a measure (see O’Leary, 2001) rather than establishment of construct validity. Follingstad (2007) continued the critique of the measurement of psychological abuse with an article that raised many definitional, conceptual, and methodological problems challenging the accuracy and validity of data collected on reports of psychological abuse. Some of the specific criticisms of psychological abuse scales included wide variation among scales as to the types of ΨA included or the relative emphasis within the scales for different forms of ΨA; inclusion of items representative of milder couple conflict strategies (e.g., “partner said something to spite you”) that greatly inflate estimates of ΨA; behaviors that could have been experienced as abusive or not depending on the context of the experience (such as the intent of the partner, how the behavior was enacted, and the meaning of the behavior for both members of a couple); and a typical focus on married heterosexual couples, which rendered many items inapplicable to other types of intimate relationships. Most scales of psychological abuse have yet to address the issues that have been raised in the literature.
The purpose of this study is to further investigate a psychological abuse measure that was developed to be distinctive as a scale representing psychological maltreatment at the severe end of the continuum. Specifically, this article reports findings from the exploration of its psychometric properties and some validity indicators. A factor analysis conducted on this scale is evaluated for content validity purposes. The scale’s utility for suggesting a threshold value for classification of individuals based on their scores is also explored.
Measure of Psychologically Abusive Behaviors (MPAB)
The MPAB was developed to improve on criticisms of ΨA measures. The scale intentionally designates more egregious behaviors representative of psychological abuse different from existing scales which cover a broader range of psychologically aggressive actions. The MPAB should therefore have utility in identifying more serious cases of psychological maltreatment distinguished from common couple conflict (see Johnson, 1995). Unique design factors of the MPAB include (a) the systematic development of 14 basically non-overlapping categories of ΨA in terms of rated severity (Follingstad, 2011; see Table 1 for category descriptions); (b) the development of three items within each category to represent a relatively milder, moderate, and severe form of that type of psychological maltreatment (see Table 2 for item descriptions); (c) phrasing of items to incorporate most variations of a particular type of ΨA, for example, “partner criticized and belittled you” rather than “partner said you were stupid,” and “partner said you weren’t a good parent”; (d) inclusion of the likely intent of each type of ΨA to ensure the recipient perceived the action as not only serious in nature but also as an intentional behavior by their partner to psychologically harm him or her; and (e) construction of items that apply to dating or married populations, cohabiting or non-cohabiting relationships, males or females, and heterosexual or homosexual couples.
Categories of the MPAB in Samples 1 and 2.
Note. MPAB = Measure of Psychologically Abusive Behaviors.
Adjusting for age, years in relationship, and race/ethnicity using binary logistic regression with MPAB behavior as the dependent variable.
Factor Analysis of MPAB Items for Normative Population (Sample 1) and Women in Distressed Relationships (Sample 2).
Note. Item 14a was missing for Sample 1. MPAB = Measure of Psychologically Abusive Behaviors.
In a prior study, a sample of adults scored all of the items as constituting a “strong” violation (or more severe descriptor) of acceptable behavior within intimate relationships (Follingstad, 2011). Some normative data on the MPAB were reported on this national sample of adult males and females regarding psychological maltreatment in their “worst relationship,” and preliminary validity information was suggested by ratings of the emotional and behavioral impacts of these behaviors as quite negative.
These initial attempts at establishing validity are not sufficient. Nunnally and Bernstein (1994) reviewed a variety of validity indicators and stated that “validation is an unending process . . . and most psychological measures need to be constantly evaluated and reevaluated to see if they are behaving as they should” (p. 84). They also indicated the importance of factor analysis for explicating the structure and composition of a scale, which in turn affects scale development and modification. And finally, many studies of IPV often classify individuals according to the presence or absence of abuse. Using summary statistics to suggest a threshold value for the likelihood of serious psychological abuse within one’s intimate relationship based on accurate assignment of individuals should add another form of validity.
Research Objectives for the Current Assessment of the MPAB
To further assess the psychometric properties and explore the validity of the MPAB, the following objectives were proposed:
Describe psychometric properties of the MPAB in terms of reliability (i.e., internal consistency) and mean scores of specific tactics, the overall scale and the three subscales varying in severity of psychological abuse for the normative sample of young women aged 18 to 29 (Sample 1) and women in distressed relationships (Sample 2).
For concurrent and criterion validity, compare the sample of young women with the sample of women in distressed relationships on chronicity of abuse, factor structure of the MPAB, and categories of psychological abuse. Results will be determined after adjusting for age, length of relationship, and race/ethnicity.
Explore the predictive validity of the MPAB for classification purposes as measured by sensitivity and specificity when comparing cutpoints of the MPAB determined in the distressed population and applied to the normative sample.
Method
Participants
Data from two national samples were used to assess the validity and reliability of the MPAB. Sample 1 included 3,313 young women 18 to 29 years of age who were recruited to represent a normative national sample of women in an intimate relationship in the past 12 months (i.e., a younger general population group). Sample 2 included a nationally recruited group of 361 women self-identified as currently in a distressed or conflictual relationship. Women in Sample 2 demonstrated significant amounts of receipt of ΨA, which defined this sample as a criterion group of psychologically abused women. In contrast, Sample 1 represents the general population of young women at risk for psychological abuse. Both samples consisted of women volunteering to take an online survey (see “Procedure” section of this article) and therefore did not constitute random samples.
Assessment of psychological abuse in younger women
The experience with intimate partner violence (IPV) among younger women (aged 18 to 29) has typically only been assessed using college samples. Younger women in samples of battered women in a shelter or in a community sample are rarely assessed separately from older women in these identified groups. Battered women from shelters or the community are quite different from college women, who typically are better educated with higher income, are younger, and are more likely to be in dating relationships.
Normative data are needed from a national sample of young women to provide estimates of psychological abuse (Jackson, 1999) because age alone is an important predictor of victimization for women. Young women are more likely to be victimized in their intimate relationships than older women (e.g., Bachman & Saltzman, 1995; Markward, 1996; O’Donnell, Smith, & Madison, 2002). Age is the demographic variable that most consistently demonstrates an association for women with IPV, sexual assault, and stalking (Jordan, Campbell, & Follingstad, 2010). When young women have been assessed for IPV, the focus has typically been on physical abuse with little data regarding other types (e.g., sexual abuse, psychological maltreatment, stalking). Assessing ΨA among young women through educational settings (high school and college) has typically included measurement of mildly aggressive behaviors that may be more representative of conflict than abuse, yet those behaviors have been labeled “abuse.” Developing measurement that can identify serious cases of psychologically maltreated young women for which intervention or treatment might be needed is important.
For a normative sample of young females in intimate relationships, solicitation occurred through an online website (www.zoompanel.com) where they were provided a link to the questionnaire that was posted on a survey platform (www.Zoomerang.com). A total of 3,313 females who met screening criteria of U.S. citizenship, aged between 18 and 29, and having been in a dating or romantic relationship in the past 12 months were available for analysis. Of these, 332 participants (10%) were dropped from the analysis because they did not complete all MPAB items.
Women in distressed relationships as the criterion group for validity
A national sample of women was recently solicited for a path model study using the MPAB (Follingstad & Rogers, 2012; Follingstad, Rogers, & Duvall, 2012). For that project, women were screened as being U.S. citizens and in a cohabiting relationship of at least 2 years duration that the women themselves classified as problematic or conflictual. This screening process was designed to obtain a sample of women who were more likely to have experienced serious ΨA. As expected, of the 361 women who participated in the survey, the mean prevalence for experiencing the 42 behaviors was 16.08 (SD = 11.93). Only 9% of the women did not report any of these serious abusive behaviors occurring in their problematic relationship. Quartiles were calculated and Quartile 1 scores ranged from 0 to 5, Quartile 2 scores ranged from 6 to 14, Quartile 3 scores ranged from 15 to 23, and Quartile 4 scores ranged from 24 to 42. Thus, this sample reported clinically significant amounts of ΨA.
Procedure
The two samples of women were solicited approximately 1 year apart, but both samples were solicited to complete an online survey utilizing www.MarketTools.com to provide a national sample. This company uses census information to direct surveys to individuals on the www.zoompanel.com site of volunteers who fit the profile of any requested sample. For these two studies, the authors requested a national sample that was basically representative of census data on education and race/ethnicity. For the criterion sample, we requested women ranging in age similar to census data, but for the large normative sample, we requested that the age range be limited between 18 and 29. Individuals who comprised the samples were volunteers who sign on with www.zoompanel.com to take surveys through this medium in exchange for incentive points that can be redeemed for a wide range of products (e.g., donations to charity, electronics and appliances, entertainment, health and personal care products, jewelry and accessories, sports equipment, tools). Any person with access to a computer can volunteer at this site if they are a U.S. or U.K. citizen. Informed consent was obtained online by requiring them to designate whether they consented to participate or not after individuals read the institutional review board (IRB)–approved consent document. Those not consenting were directed out of the survey site. To promote anonymity, no identifying information was required for consenting to participate, and the IRB waived the requirement of a signature. Individuals could refuse to answer any item, and they could exit the study at any point by closing down the website. The surveys ended with debriefing information that included potential referral sources. Participants were also provided with the information necessary to erase the browser history of their involvement in the study, thus increasing confidence for the women that their reports were anonymous and that no one would be able to know of their involvement. The procedures for the study were approved by the University of Kentucky IRB (#10-0412-P2H).
Measures
MPAB
In addition to the information above describing the MPAB (Follingstad, 2011) as a measure of more severe psychological maltreatment, reliability indicators have been assessed (internal consistency = .98 for the entire scale, and .94 for all three severity subscales; split-half = .96; individual items correlated .53-.81 with the overall score). All 42 items were rated as “serious” violations of intimate partner relationships or stronger descriptors than “serious,” such as “abusive” or “worst possible” (i.e., 6 and above on a scale from 1 to 10), and relatively milder, moderate, and severe items within all 14 categories were rated significantly differently from each other as to severity. [Note: Only 41 items are being used in this study because 1 item was inadvertently omitted in the survey for Sample 1. However, it was an item from the least severe category, Making the Partner’s Personal Choices.] The categories themselves demonstrated statistically significant differences in severity ratings among 10 of the 14 categories at the upper and lower ends of the ratings, with only 4 categories at the midrange not being significantly distinguished from each other, although they were distinct from categories at the two poles. Attachment and social desirability, although mildly correlated with this scale, demonstrated discriminant validity in that they only accounted for a slight amount of the variance of reported ΨA.
A total score of psychological abuse incorporated both the number of behaviors women experienced and the frequency each occurred in a 12-month period. The Likert-type options ranged from 0 = never to 3 = almost daily. The average frequency of each behavior (range = 0-3), prevalence scores of the number of categories experienced (range = 0-14), prevalence scores of the items experienced (range = 0-41; highest score is not 42 due to a Milder item missing from the young women’s survey), and prevalence of the Milder, Moderate, and Severe subscales including frequency scores (range for Milder subscale = 0-39; range for Moderate and Severe subscales = 0-42) were also devised. Because the total score correlated highly with all of the other ΨA scoring dimensions, that variable was used in analyses unless otherwise specified. In addition, dichotomous variables were created to measure the frequency of each behavior occurring at least monthly (yes/no).
Current physical and sexual abuse
The presence of physical and/or sexual abuse in a woman’s current relationship was assessed with six items devised by one of the authors. Measurement of physical abuse consisted of three summary items corresponding to milder, moderate, or severe physical violence. The three items gave a general description of physical force at either the milder, moderate, or severe level, and was followed by examples. Response options ranged from 0 to 5 and corresponded to frequency of occurrence (0 = never; 1 = once a year; 2 = few times a year; 3 = monthly; 4 = weekly; 5 = almost daily). In a prior study (Follingstad, 2011), each of these summary items was found to correlate highly (r = .66-.79) with subscales of the Revised Conflict Tactics Scale (CTS-2) corresponding to milder, moderate, and severe physical abuse, respectively (Straus, Hamby, Boney-McCoy, & Sugarman, 1996). In this study, women’s average yearly frequency of the three items measuring milder, moderate, and severe physical abuse was summed to create a total score (Cronbach’s α = .96).
Parallel to the summary items for physical abuse, three items of increasing severity measured sexual abuse in the women’s current intimate relationship. The three levels of sexual abuse corresponded to obtaining sexual activity through harassment/insistence, coerced sexuality, and physically forced sexuality. Response options ranged from 0 to 5 and corresponded to frequency of occurrence (0 = never; 1 = once a year; 2 = few times a year; 3 = monthly; 4 = weekly; 5 = almost daily). Follingstad (2011) also found these three summary items of sexual abuse to be highly correlated with subscales of CTS-2 items representative of mild, moderate, and severe levels of sexual abuse in a prior study, thus supporting their utility as a shorter form. The total score was the women’s average yearly frequency of sexual abuse summed across the three levels of severity (Cronbach’s α = .81).
Demographics
For this study, the demographics that had been collected from both samples were used (see Table 3). Variables included participants’ ages; relationship status (i.e., never dated [screened as ineligible]; not currently dating, going out on dates but not in a dating, sexual, or romantic relationship; in a dating, sexual, or romantic relationship, but not living together; or currently married or living with partner); duration of the current or most recent relationship in the past 12 months (i.e., less than 6 months, 6 months but less than 1 year, 1 year but less than 2 years, 2-5 years, and 5 years or more); education (less than high school, some high school, high school graduate or received general education development [GED], some college, vocational or technical certificate or degree, associate’s degree, bachelor’s degree, and post-graduate degree); current employment status (i.e., working, student, or “other” to include not working, disabled, or a homemaker); and race/ethnicity (i.e., African American, Asian American; Bi- or Multi-Racial; Caucasian; Hispanic; and Other [please specify]). Given the small numbers of respondents identifying as Asian, Hispanic, or Other, these respondents were all grouped as Other.
Demographics of Two Samples With Statistical Comparisons.
Note. SE = standard error; GED = general education development.
Statistical Analyses
All analyses were conducted using SAS 9.3. Continuous variables (age, scores based on the number of items experienced) were summarized with descriptive statistics (n, M, standard error), and categorical measures were described with counts and percentages. To identify factors related to membership in Samples 1 and 2, two-sample t tests of means and chi-square test of proportions were provided (Table 3). Comparisons of continuous outcomes (MPAB scores) between Samples 1 and 2 were made using ANCOVA (Tables 3 and 4), allowing for the comparison of Samples 1 and 2 means, adjusted for demographic differences in the two samples (i.e., age, years in the relationship, and race/ethnicity). Item response proportions from Samples 1 and 2 are presented with regression coefficient estimates from binary logistic regression (Table 1), χ2(df), and p value from adjusted analyses. Research Objective 1 was addressed by calculating Cronbach’s alpha for the total MPAB and the three subscales by severity (Table 4). Mean scores and standard errors were also calculated for those scales as well as for the 14 categories of psychological abuse (Table 1). To address concurrent and criterion validity, the frequency of the MPAB items was calculated and compared for the two samples (Table 4). Comparisons were also conducted for the two samples on the categories of psychological abuse (Table 1). Factor analysis (with varimax rotation) was conducted for the 41-item MPAB in both samples to determine whether the factor loading was similar in the two samples (Table 2). Receiver operator curves (ROCs) with estimates of the area under the curve (AUC) were used to address Research Objective 3 to investigate the predictive power of the MPAB using the total number of items as a numerical score. ROCs were also used to identify optimal cutpoints for the MPAB index for distinguishing between Samples 1 and 2. Optimality was defined by Youden’s index and by maximizing sensitivity and specificity. Figure 1 provides the ROC curve for the total MPAB, adjusted for age and years in relationship.
Adjusted Means and Standard Error and Cronbach’s Alphas for MPAB Dimensions of Two Samples With Statistical Comparisons a .
Note. MPAB = Measure of Psychologically Abusive Behaviors; SE = standard error.
Adjusted for age, years in relationship, and race/ethnicity using ANCOVA.
Used 41-item questionnaire (14a—milder item of Personal Control is missing) for Samples 1 and 2; response option frequency ranged from never (= 0), monthly (= 1), weekly (= 2), to almost daily (= 3).
Standardized Cronbach’s alpha coefficient.
Used 13 items for mild MPAB for Samples 1 and 2.
Receiver operator curve for total MPAB scores adjusting for age and years in the relationship (AUC = 0.97).
Results
Demographic Comparisons for Two Samples
Demographics for the two samples of women are presented in Table 3. As expected, women in Sample 1 were significantly younger and had been in their relationships for shorter periods than those in Sample 2. Sample 1 also had significantly more non-Caucasian participants than Sample 2. Although initially significant, all other demographics were no longer significantly different for the two samples when age, years in the relationship, and race/ethnicity were also in the statistical model. Thus, analyses were all conducted while adjusting for those three demographic variables.
Psychometric Properties of MPAB in Sample 1 and Sample 2
Internal reliability
Internal consistency of the total MPAB score was similar for women in Sample 1 (α = .968 total) and Sample 2 (α = .973 total; see Table 4). Cronbach’s alpha coefficients for the three subscales constituting Milder, Moderate, and Severe items were also similar for the two samples (Milder: Sample 1 = .907, Sample 2 = .905; Moderate: Sample 1 = .917, Sample 2 = .924; and Severe: Sample 1 = .899, Sample 2 = .931). Thus, reliability in terms of internal consistency appears excellent for the total scale and subscales of the MPAB.
Means and quartiles
MPAB scores were significantly higher among women in distressed relationships (Sample 2) than among women in the normative sample (Sample 1; see Table 4). Adjusting for age, years in the relationship, and race/ethnicity, all comparisons of the means for Samples 1 and 2 resulted in highly significant differences (p < .0001). All comparisons indicated that women in Sample 2 reported much more psychological abuse than women in Sample 1. This finding was anticipated as Sample 2 was included as the case group of women experiencing psychological abuse based on their self-identification as being in distressed relationships.
Quartiles for women in Sample 1 who experienced any of the 41 behaviors of the MPAB revealed that Quartile 1 = 0, Quartile 2 = 1-4, Quartile 3 = 5-12, and Quartile 4 = 13-41. Compared with the quartiles for Sample 2 (see data in “Participants” section of this article), the data of the young women were skewed toward few of the participants experiencing more than 12 of the 41 behaviors, and one fourth of the women reporting no psychologically abusive behaviors directed toward them.
Comparisons Between the Normative (Sample 1) and the Criterion (Sample 2) Groups
Chronicity of psychological abuse
To assess only women who were exposed to the psychological form of IPV, all women experiencing moderate or severe physical abuse or forced sex were eliminated from an analysis designed to investigate chronicity of psychological abuse, that is, women who experienced MPAB behaviors at least monthly during the prior year. Chi-square analyses were conducted to compare the two samples after adjusting for age, years in the relationship, and race/ethnicity (see Table 4). Thirty-eight percent of women in the large normative sample (Sample 1) reported at least one MPAB behavior occurring at least monthly compared with 65% of women in Sample 2. Young women in Sample 1 reported fewer of the Milder items occurring at least monthly compared with women in Sample 2 (31% vs. 61%). Young women reported fewer of the Moderate items (21% vs. 53%) and Severe items (19% vs. 51%) occurring at this rate of chronicity as well. All chi-square comparisons between the two groups resulted in significant differences (p < .0001). Thus, women in distressed relationships not only reported experiencing more types of abusive behaviors, but they also reported higher frequency levels of experiencing them.
Exploratory factor analysis
The factor loading for the 41 items in the MPAB were similar for the two samples. Table 2 provides the results of the factor loading analysis for each of the 41 items of the MPAB included in both samples. Two varimax rotation factor analyses were conducted on the items—one for the large normative sample and the other for the women in distressed relationships. The scree plot indicated for both samples that three factors captured most of the variance (93% for Sample 1 and 86% for Sample 2). Therefore, the factor analysis was rerun specifying three factors to be extracted and the varimax rotation method was again used. Factors 1, 2, and 3 appear quite comparable for the large normative sample of young women (with the amount of variance explained by each factor as 43%, 35%, and 23%, respectively) and the sample of women in distressed relationships (with the amount of variance explained by each factor as 39%, 36%, and 24%, respectively). The items that loaded highest on each factor (and items that loaded second highest on that factor within .04 of the loading on the highest factor) by both groups of women were used to determine the thematic content of the factors.
Factor 1 had 16 items in common that loaded highest on that factor. The items included intentionally showing a temper, manipulating through emotional upset and threats, engaging in verbal abuse, treating the woman as inferior, creating a hostile environment, refusing to speak to, punishing, and controlling personal choices and major decisions.
Factor 2 had 13 items in common that loaded highest on that factor. These items appeared even more serious in nature reflecting sadistic behavior, severe threats to harm or kill, threatening suicide to manipulate, wounding the partner through sexuality, wounding the partner through threats to fidelity, and monitoring the woman by following her.
Factor 3 had fewer items loading jointly on this factor with only six items having their highest loadings on this factor. This factor includes items consistent with a very jealous partner who attempts to control the partner. In addition to accusations of infidelity, the partner tries to prevent the woman from socializing with others or even from speaking with or looking at others. Invasions of privacy appear to be a result of the jealousy in that the partner demands details of daily events, monitors the woman’s communications, and goes through her belongings.
Categories of psychological abuse
Table 1 provides a comparison of the two samples of women based on the proportion of women disclosing at least monthly receipt of the 14 different categories of psychological abuse on the MPAB. For every category of abuse, the proportion of women experiencing that category at least monthly was consistently greater (twice as high) among those in Sample 2 than among those in Sample 1. Among those in Sample 1, the proportions experiencing the 14 categories of abuse ranged from 3.0% to 28.6%, whereas among women in distressed relationships (Sample 2), the range was 9.4% to 51.8%. Thus, more women in distressed relationships have experienced each category of psychological abuse more than the sample of young women.
The top four categories most frequently reported in Sample 1 were Monitoring, Verbal Abuse, Creating a Hostile Environment, and Treating as an Inferior. Verbal Abuse, Treating as an Inferior, and Creating a Hostile Environment were similarly ranked in the top four most frequently reported categories among Sample 2 as well. However, Manipulation was of the fourth category reported most frequently by women in distressed relationships in contrast to Monitoring being the fourth category reported by the large sample of young women.
Predictive Validity
ROCs were used to identify the predictive power of the MPAB as well as determine an appropriate cutpoint of the MPAB for accurate classification of participants. ROC curves were drawn using the total MPAB score, and Youden’s index was used to identify the optimal cutpoint for total MPAB scores as 1. When additionally using the intersection of sensitivity and specificity as a criterion for optimal cutoffs, the sensitivity of the total MPAB score was .725, and the specificity was .628 when a cutpoint of 1 was used. When adjusting for age and years in relationship (Figure 1), the ROC curve provides a further demonstration of how well the combination of variables predict psychological abuse status when adjustments are made for age and years in the relationship (AUC = 0.97). Thus, classification of women as being exposed to serious forms of psychological abuse was determined to be a score of 1 when considering the best score (sensitivity) that classifies psychologically abused women as such and the best score (specificity) that classifies women who have not been psychologically abused as such.
Discussion
A large national sample of women 18 to 29 years of age provided normative data for the MPAB, and a separate national sample of women in distressed relationships was used for comparison to provide validity information. For both samples, the MPAB was highly reliable in terms of internal consistency for the total score as well as for the three subscales consisting of relatively milder, moderate, and severe items on the scale.
One fourth of the young women in the normative sample reported that none of the psychologically abusive behaviors listed on the MPAB had been directed toward them in the prior year, but three-fourths of them experienced at least one during the past year. However, mean scores of experiencing psychological abuse in the normative group of young women were relatively low. The MPAB was deliberately designed as a measure that identifies serious violations of intimate relationships for which recipients believe their partners deliberately intended psychological harm, and so it is a relief that mean scores were low. There is still cause for concern that such a large proportion still reported at least one behavior directed at them because experiencing even one psychologically abusive action on the MPAB classified a respondent as belonging in the “exposed” group. And all categories of maltreatment showed up in fairly high percentages of young women’s relationships within the past year when the analysis identified participants who at least monthly experienced at least one of the behaviors in a category. For example, the category that was present in the most relationships of these young women was Monitoring. However, even the most egregious category of sadistic behaviors was present in 3% of their relationships, and although 50% of the sample experienced 4 or fewer of the abusive behaviors in the last year, 25% reported experiencing 13 to 41 behaviors. A significant proportion (38%) of young women reported fairly high chronicity of experiencing ΨA, defined for this study as the monthly occurrence of at least 1 of the 41 behaviors. Using this more stringent measure and requiring a higher frequency of psychological abuse for analysis, it appears that young women are no strangers to being on the receiving end of psychologically abusive tactics.
Comparisons of the large national sample’s scores on the MPAB with those of the women in distressed relationships (even while adjusting for age, years in the relationship, and race/ethnicity) demonstrated highly significant differences. Women in distressed relationships were much more likely to experience a greater number of psychologically abusive behaviors, they experienced them more frequently, and much larger proportions of them reported experiencing the various categories of ΨA. Thus, the data from the women in distressed relationships are useful for supporting the validity of the MPAB. The likelihood is very high that a woman being in a “problematic or conflictual” relationship has experienced some serious form or forms of ΨA.
The most frequently experienced types of ΨA that young women in the normative sample overlapped to a substantial degree with the most frequently experienced types by women in distressed relationships. Of the four categories of ΨA that showed up most frequently in women’s relationships, the two groups commonly experienced Verbal Abuse, Creation of a Hostile Environment, and Treatment as an Inferior. Monitoring was experienced most frequently by the young women, whereas the women in distressed relationships were likely to experience Manipulation as a major category. Possibly, there is more checking up on one’s partner that arises out of jealousy in the young population still characterized by dating and cohabiting. Serious acts of manipulation do seem consistent with distressed relationships because they represent negative tactics deliberately aimed at maneuvering one’s partner.
Conducting this study allowed for an analysis of the sensitivity and specificity of the MPAB to be used for classifying women as experiencing serious ΨA or not. The threshold score on the MPAB to classify women was designated as 1, which suggests that experiencing even one of these more serious violations of intimate partner relationships places a woman into the “exposed” category. Some prior research has suggested that women experiencing very few of these forms of maltreatment seem more similar to women in distressed relationships with no ΨA than to women experiencing many abusive behaviors (Follingstad & Rogers, 2012). However, that data also indicated that women experiencing only a few forms of ΨA still reported anxiety, depression, and somatization at a diagnosable level that suggested the need for services. Therefore, future research is needed to determine whether young women in a national sample who experience even one type of serious psychological maltreatment also experience mental health consequences that may require intervention.
The three-factor structure of the MPAB for both groups of women yielded consistent patterns of ΨA by the women’s partners, allowing for an analysis of the structure of the scale. The content appears descriptive of three themes of received psychological abuse. There have been different typologies of physically abusive men advanced in the literature, and the factors of psychological abuse derived from these analyses have some similarities to previously identified typologies. The first factor seems to fit Tweed and Dutton’s (1998) impulsive and overcontrolled type or Hamberger and Hastings’ (1986) dependent and compulsive type. This pattern of psychological abuse is characterized by an interpersonal style of controlling the woman through treating her as inferior, treating her with hostility, engaging in tactics to make her feel bad about herself, and keeping her on edge through emotionally labile eruptions. The frequent use of tactics to reduce the woman’s self-view may arise from counterdependency and insecurity within the partner. Thus, the controlling strategies seem designed to make sure she perceives him as superior and someone to be placated, while perceiving herself as not having value, in order that she not abandon him.
The second factor represents a partner who fits Tweed and Dutton’s (1998) instrumental/undercontrolled type, Saunders’ (1992) generally violent type, Holtzworth-Munroe and Stuart’s (1994) antisocial type, and Hamberger and Hastings’ (1986) antisocial/narcissistic type. This psychologically abusive partner is willing, in a calculated sense, to terrorize his partner to promote compliance. In addition, there are also elements of borderline personality organization (see Dutton, 1994) in the form of impulsive destructiveness, willingness to wound the woman’s self-view in devastating ways (i.e., through sexuality and threats to fidelity), and intense, unstable anger.
The third factor represents a partner with pathological jealousy that appears to be encapsulated and that does not seem to develop or expand into other unrelated forms of psychological abuse. However, the jealousy is not only present in terms of accusing the partner of infidelity, but the pathology emerges as restrictions the jealous partner wants to impose on the partner and a willingness to invade the partner’s privacy. Because of the attempts to impose restrictions on the partner, the effects on the relationship can be far-reaching. For example, attempts to regulate the woman’s relationships with everyone in her life, from family, to friends, to work acquaintances, continuously invading the woman’s privacy while seeking evidence of infidelity, demanding information about her daily activities, or monitoring her communications, can infiltrate most of the relationship.
Thus, the factor analysis of the women’s report of the frequency and types of ΨA they were experiencing resulted in similar patterns across the two samples as to descriptions of psychologically abusive partners. Whereas the first two factors seemed to match descriptions in the literature of particular typologies of physically battering men, a third and seemingly unique type of psychologically abusive partner emerged, characterized by pathological jealousy with particular behavioral manifestations. There is no research that has identified typologies of partners based only on the forms of psychological abuse in which they engage. Thus, this is potentially a useful direction for research.
The fact that so little variance of psychological abuse was accounted for by demographics of the participants suggests that gross demographic variables are unlikely to predict or explain the presence of this phenomenon. To ensure that the most significant differences between the two groups of women in this study were accounted for, however, analyses were always adjusted for age, length of time in the relationship, and race/ethnicity.
Limitations of this study include the standard concerns with self-report and the fact that the sample was self-selected rather than random. In addition, this study provided a cross-section of the women’s lives rather than a longitudinal assessment, which would have allowed for statements regarding causality. The decision to limit both samples to U.S. citizens was designed to eliminate potential cross-cultural influences, but the unintended plurality of Caucasian women, women with educational levels averaging above the national average, and only 50% of the sample currently used limit generalizability. Ethnic diversity was not elaborated for these results other than being controlled for as a potential confounding variable. Further research with relevant subgroups of women is needed to clarify whether these data should be modified for particular subsets.
Although initially a concern, the smaller number of women in distressed relationships for comparison with the normative sample still produced significant and apparently clinical differences. This study attempted to maximize disclosure for reporting on an intimate partner’s behavior through anonymous data collection using an online survey.
This study’s findings provide us with reliability and validity data in support of the MPAB as a measure of serious psychological abuse, some normative data for understanding its psychometric properties, thematic factors illustrative of potential “types” of psychological abusers, and a window into the extent of psychological abuse within young women’s intimate relationships. This information adds to the reliability and validity data already in existence on the MPAB and suggests that this scale may have utility for screening for serious psychological abuse as well as utility as a research tool.
What we do not know from these data and what we still need to investigate are those factors that would result in a more complete picture of psychological abuse in young women’s relationships. For example, because the measure that was used in this study required a woman to believe her partner intended psychological harm through the use of that particular tactic to report that it occurred (e.g., intimidation, feel bad about oneself, increase insecurity, control), the actual frequency of the same behaviors occurring that the woman did not consider deliberately harmful might be much higher. Follingstad, Coyne, and Gambone (2005) found that, on average, approximately 10% of psychologically aggressive actions reported by respondents were considered by them to meet a threshold of “abuse.” We also do not know from these data whether psychologically abusive tactics were reciprocal within these relationships or within certain subgroups of relationships. And it is likely that prevalence and frequency counts would have been significantly higher had we collected data as to the women ever experiencing these psychological tactics above and beyond the current partner.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.