Abstract
Using a sample of 1,190 married Spanish community couples in opposite-sex relationships, this study evaluated a path analytic model exploring the associations between individual and dyadic factors and partner aggression for both males and females. Specifically, the perpetrator’s report of their individual mental health symptomatology (borderline and antisocial traits, alcohol use), their report of relationship quality, and their report of perpetration were modeled to predict their partner’s victimization. The resultant model exhibited good fit as measured by multiple indices for both male-to-female and female-to-male perpetration. Furthermore, results demonstrated that antisocial and borderline traits were associated with each other, that alcohol use was associated with perpetration, and that one partner’s perpetration was associated with his or her partner’s victimization for both males and females. However, the pattern of significant pathways between individual pathology and relationship violence differed somewhat for male-to-female and female-to-male aggression. Given the results, policy makers concerned about prevention of partner aggression may consider interventions aimed at symptoms associated with antisocial and borderline traits and substance use, and may identify differential targets for intervention based on the perpetrator’s gender.
Intimate partner aggression has been associated with negative physical outcomes such as injury (e.g., Archer, 2000; Capaldi & Owen, 2001; Capaldi et al., 2009), poor physical health (Bonomi, Anderson, Rivara, & Thompson, 2007; Taft, Vogt, Mechanic, & Resick, 2007), and negative psychological outcomes (Goodman, Koss, & Russo, 1993). For instance, victims are more likely to engage in alcohol and drug use, to report symptoms of depression or PTSD, and to endorse suicidal symptoms (Golding, 1999). Because partner aggression leads to health consequences at the individual level as well as considerable financial cost for society at large (e.g., National Center for Injury Prevention and Control, 2003), it is important to understand the correlates and causes of perpetration in community samples to effectively identify targeted prevention strategies.
To date, a handful of multivariate models have explored aggression perpetration in the literature; most utilize samples from the United States. Although these models vary widely in the risk factors identified, in recent years, many have incorporated symptoms of personality pathology and substance use in predicting relationship aggression. For example, Stuart and colleagues found that antisocial traits predicted trait anger and alcohol problems, each of which independently predicted psychological aggression and, in turn, physical aggression (Stuart et al., 2006).
Although the study of multivariate perpetration risk factors is limited internationally, one notable exception is an epidemiological study conducted in the Ukraine, which found that, in addition to other factors, early onset alcohol abuse and psychiatric problems predicted men’s aggression perpetration (O’Leary, Tintle, Bromet, & Gluzman, 2008). Another study explored individual and dyadic risk factors for physical aggression in a Spanish sample of male batterers (Jose, O’Leary, Graña Gomez, & Foran, 2014). In this study, men’s borderline traits were associated with alcohol use, psychological perpetration, and physical perpetration, whereas, unexpectedly, antisocial traits were not significantly associated with other variables. (Here, the term “antisocial traits” is used to refer to psychopathy, which overlaps with, the behaviorally based Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnosis of antisocial personality disorder, while also including interpersonal and affective components; Ogloff, 2006.)
Given the limited research related to risk factors for intimate partner violence (IPV), particularly within an international context, as well as the significant consequences of IPV, it seems particularly important to better understand the individual and dyadic factors that lead to violence in relationships. In turn, this will ensure targeted interventions can be applied on the individual and dyadic levels, targeted at couples at risk of IPV. Within this framework, the present study attempts to both replicate and extend findings from this sample of Spanish male batterers to a community sample in Spain. This study tests the hypothesis that personality pathology and alcohol use, together and alone, lead directly to male perpetration, which, in turn, has a direct association with female victimization.
Because borderline and antisocial traits are both marked by long-standing interpersonal difficulties, these traits are hypothesized to be associated with each other in the model. A clear association between borderline traits and aggression exists across dozens of research studies, despite heterogeneity in the construct of borderline personality pathology (e.g., Dutton & Starzomski, 1993; Holtzworth-Munroe, Meehan, Herron, Rehman, & Stuart, 2000; Holtzworth-Munroe & Stuart, 1994; Jackson, Sippel, Mota, Whalen, & Schumacher, 2015). Research has also demonstrated an association between aggressive behavior and antisocial symptomatology (e.g., Holtzworth-Munroe & Stuart, 1994; Kim & Capaldi, 2004; Mauricio, Tein, & Lopez, 2007; Murphy, Meyer, & O’Leary, 1993). The previous batterer study identified significant paths between borderline pathology and aggression, but, unexpectedly, no significant paths between antisocial traits and other variables. Given the extensive research conducted to date, this was an unexpected result, and the present study attempts to explore whether this finding holds true in a large sample of community couples.
It is expected that alcohol use problems are higher in those with antisocial and borderline traits and that alcohol use increases the risk of relationship conflict and physical violence. Previous research suggests borderline traits longitudinally predict alcohol problems, even when controlling for parental alcohol use (Stepp, Trull, & Sher, 2005). Studies have also linked antisocial traits and alcohol use (Stuart et al., 2006; Stuart et al., 2008). Finally, a meta-analysis demonstrated the role of problematic alcohol use in aggression perpetration (Foran & O’Leary, 2008). Given the lack of impulse control often experienced by individuals with borderline or antisocial traits at baseline, and the role of alcohol in disinhibition, it seems plausible that these factors together would place couples at higher risk of aggression. Although the batterer sample found significant associations between alcohol use and aggression, it was restricted by the exclusion of participants with significant substance use. The current sample is not similarly restricted and allows further exploration of this association.
This study also explores relationship quality as a predictor of aggression. The association between relationship quality and aggression is worth further study, as one meta-analysis found an effect size of r = –.27 (Stith, Smith, Penn, Ward, & Tritt, 2004). Despite some disagreement as to which variable predicts the others (e.g., Murphy & O’Leary, 1989; Pan, Neidig, & O’Leary, 1994), the preponderance of research seems to suggest that poor relationship quality is a risk factor for psychological perpetration, which, in turn, is a risk for physical perpetration (e.g., Gavazzi, McKenry, Jacobson, Julian, & Lohman, 2000; O’Leary, Malone, & Tyree, 1994; Stuart & Holtzworth-Munroe, 2005). As the batterer study did not find any significant paths associated with relationship quality, the current model explores the possibility of a bidirectional association between relationship quality and psychological victimization.
Finally, the batterer study only included male reports of aggression perpetration and victimization; the present study utilizes a sample of couples who are currently married and living together. In addition, this study explores two parallel models to assess male-to-female and female-to-male violence.
In summary, the present study is a replication and extension of Jose and colleagues’ (2014) batterer study in Spain. The purpose of this study is to determine whether the findings replicate in a general community sample comprised of currently married couples. Moreover, changes in the design, such as the use of both partner’s reports of aggression, and the exploration of a bidirectional association between marital quality and aggression, were incorporated.
Method
Participants
Data were collected from a community sample of 1,190 married couples who were recruited from the greater Madrid area of Spain. Both members of the couple were invited to complete self-report measures assessing individual and dyadic functioning independently of their partner.
Measures
CAGE Questionnaire (CAGE; Ewing, 1984)
This four-item assessment is widely used as a brief screening tool to assess alcohol problems, particularly in health care settings. This measure focuses on behaviors and problems associated with alcohol use. Scores in this sample ranged from 0 to 4 and at least one item was endorsed by 20.6% of males and 9.1% of females. Problem drinking is reflected by a score of two or greater on the CAGE; 8.7% of men and 2.9% of females reported problem drinking on this measure. Cronbach’s alpha in this sample was .65 for males and .63 for females, likely due to the brevity of the scale.
Conflict Tactics Scale, Revised (CTS2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996)
The CTS2 is a 78-item self-report questionnaire assessing behaviors during relationship conflict. Thirty-nine items ask about perpetration and another 39 items ask about victimization within the past year, across five subscales (Negotiation, Psychological Aggression, Physical Assault, Injury, Sexual Coercion). According to Straus et al. (1996), Cronbach’s alpha ranged from .79 to .95 for subscales of the CTS2. In the present study, full scale alpha was .86 for males and .85 for females.
Levenson Primary and Secondary Psychopathy (LPSP) Scale (Levenson, Kiehl, & Fitzpatrick, 1995)
This 26-item self-report scale measures the domains of manipulation and callousness (primary psychopathy) and impulsivity (secondary psychopathy), and has adequate psychometric qualities (Lynam, Whiteside, & Jones, 1999). This measure is viewed as an acceptable and appropriate self-report measure of psychopathic traits (Lilienfeld & Fowler, 2006). The present study includes an abbreviated 10-item version of the LPSP (scores ranged from 0 to 30), for which alpha was suitable at .84 for males (primary α = .84, secondary α = .75) and .83 for females (primary α = .82, secondary α = .76).
McLean Screening Instrument for Borderline Personality Disorder (MSI-BPD; Zanarini et al.,. 2003)
This 10-item self-report scale measures BPD based on its DSM (4th ed.; DSM-IV; American Psychiatric Association, 1994) diagnostic criteria. Both internal consistency (α = .74) and test–retest reliability (ρ = .79) were high when assessed by the researchers. This study used an abbreviated version of the MSI-BPD (scores ranged from 0 to 5); Cronbach’s alpha was .64 for males and .68 for females; this may reflect the brevity of the scale.
Quality of marriage index (QMI; Norton, 1983)
This six-item measure of global relationship quality ranges in score from 7 to 45, with higher scores reflecting greater relationship quality. In the current study, alpha was excellent for male (α = .92) and female (α = .91) participants.
Procedure
Translation of measures
For a previous study using the same self-report questionnaires, English-language versions of the questionnaires were translated into Spanish by a bilingual psychologist at the Complutense University of Madrid. The translating psychologist had extensive professional experience converting research papers from English into Spanish. After the initial translation, measures were back-translated by a professor from the Department of English. A subset of the measures was then applied to a pilot sample of 15 Spanish men who participated in a batterer intervention program, to determine item comprehension. Pilot participants reported understanding the meaning of all items.
Study procedure
This study was approved by the Ethics Committee of the Faculty of Psychology at the Complutense University of Madrid. Informed consent statements were included with study questionnaires; participants were advised that their mailed return of the questionnaire form indicated their consent to participate.
The procedure used a quota sampling method to recruit married couples in the Madrid region. Specifically, research assistants were provided with training related to subject recruitment, as well as the provision of resources in case relationship violence was identified. Research assistants approached couples within the community and requested their voluntary participation (e.g., speaking to people on the street and handing out study advertisements). Participants were advised that both members of the couple would complete self-report measures independently, thereby maintaining response confidentiality. Each research assistant was assigned to a census area in the Madrid region, and was required to recruit at least eight couples.
Participants meeting all inclusion criteria (individuals aged 18+, who self-identified as being involved in an opposite-sex romantic relationship, and whose partners were also willing to participate) were invited to complete the battery of self-report measures. Participants were specifically instructed to respond to questionnaires measuring dyadic functioning by considering their current relationship, and to mail back their responses independently of their partner.
Results
Descriptive Statistics
Between 2010 and 2012, 5,440 participants were recruited, with a questionnaire return rate of 78.8% (4,284 returned questionnaires). Of the surveys returned, 220 were deemed invalid (e.g., due to excessive missing values). This study analyzed only data from married respondents, so data from unmarried, divorced, or widowed participants were not included. These steps yielded data from 2,380 individuals (1,190 couples).
Participating couples were married and had lived together for an average of 19.7 years (SD = 12.2 years). Males reported an average age of 46.5 years (SD = 12.3 years), whereas females were 44.4 years old (SD = 11.9 years) on average. The vast majority of participants reported being Spanish (92.3% of males, 97.6% of females). In addition, most male participants were employed, with 6.0% of the sample reporting unemployment (and 0.2% of the sample indicating they were students). Comparatively higher proportions of female participants indicated they were not currently employed (22.6%) or indicated they were students (7.4%).
Prevalence of Aggression
Table 1 documents the annual prevalence rates for psychological aggression, physical assault, and injury using males’ and females’ own reports of perpetration and victimization, and representing the entire sample (n = 1,190 couples). Women reported a higher prevalence of all types of aggression besides injury, compared with men. About half of men and women surveyed reported psychological aggression in the year prior. Females (7.1 %) reported perpetrating physical aggression more frequently than males (5.5 %). Less than 1% of men and women reported injury perpetration or victimization. Finally, it should also be noted that in every area except female-to-male injury, females consistently reported higher rates of perpetration and victimization than their partners (e.g., higher rates of female physical perpetration compared with male physical victimization).
Prevalence of Self-Reported Perpetration and Victimization (1 Year).
Table 2 shows the frequency of aggressive incidents separately for men and women, once again representing the entire sample. Due to nonnormality, Wilcoxon’s signed rank test was used to compare male and female reports of psychological aggression, physical assault, and injury separately. Men reported they were more frequently the victim, rather than the perpetrator for psychological and physical aggression (there was no significant difference for injury). Women reported that they more often perpetrated psychological aggression, but there was no significant difference in terms of physical aggression or injury.
Frequency of Aggressive Incidents (1 Year).
Bivariate Correlations
To separately assess the impact of male and female aggression, the sample was randomly split into two independent files, with one file used to analyze male-to-female aggression (n = 591), and the second file used to analyze female-to-male aggression (n = 599). Tables 3 and 4 include additional descriptive data for male-to-female and female-to-male aggression separately. These tables contain the mean and standard deviation of each variable on the diagonal. Above the diagonal are the bivariate correlations between path variables, using Spearman’s rho to account for nonnormality of some variables. For male-to-female aggression, patterns of correlation generally reflect paths hypothesized in the tested models, with the major exception being that relationship quality did not demonstrate significant first-order correlations with any variable. For female-to-male aggression, both relationship quality and alcohol use demonstrated few first-order associations with other variables of interest.
Male-to-Female Aggression: Means, SDs, and Correlations.
Note. CTS2 = Conflict Tactics Scale, Revised; LPSP = Levenson Primary and Secondary Psychopathy Scale; MSI-BPD = McLean Screening Instrument for Borderline Personality Disorder; CAGE = CAGE Questionnaire; QMI = Quality of Marriage Index.
p < .05. **p < .01.
Female-to-Male Aggression: Means, SDs, and Correlations.
Note. CTS2 = Conflict Tactics Scale, Revised; LPSP = Levenson Primary and Secondary Psychopathy Scale; MSI-BPD = McLean Screening Instrument for Borderline Personality Disorder; CAGE = CAGE Questionnaire; QMI = Quality of Marriage Index.
p < .05. **p < .01.
Path Analytic Strategy and Model Fit
Data from the full sample of 1,190 couples were utilized. 1 As described above, this sample was randomly split into two independent files, with one file used to analyze male-to-female aggression (n = 591), and the second file used to analyze female-to-male aggression (n = 599). Path analyses were conducted to test the model illustrated in Figure 1, using maximum likelihood estimation with robust standard errors (MLR) with Mplus statistical software (Muthén & Muthén, 2015). Following Hu and Bentler’s (1998) guidelines, overall fit was evaluated using multiple indices: the comparative fit index (CFI; Bentler, 1990), root mean square error of approximation (RMSEA; Steiger & Lind, 1980), and standardized root mean square residual (SRMR; Bentler, 1995). For these analyses, CFI > 0.95, RMSEA < 0.05, and SRMR < 0.05, each signifies good fit.
Hypothesized model.
Path Analytic Results: Male-to-Female Aggression
The measured indices indicated very good fit: CFI = 1.00, RMSEA = 0.00 (90% confidence interval [CI] = [0.00, 0.06]), and SRMR = 0.02. However, a number of the hypothesized paths were demonstrated to be nonsignificant (Figure 2). Modification indices (M.I.s) were examined to determine whether to trim nonsignificant paths; no M.I.s greater than 10 were identified so all paths were retained. (For visual clarity, nonsignificant paths are not depicted.)
Final model: Male-to-female aggression.
As hypothesized, antisocial and borderline traits were associated with each other, with alcohol use, and with male perpetration; all associations were found to be in the expected direction. Although relationship quality was hypothesized to be associated with individual and dyadic variables, this was not borne out in the model (or in the examination of first-order correlations). Theoretically, this is unexpected; however, it is consistent with prior results (Jose et al., 2014). Finally, male perpetration and female victimization were strongly and significantly associated in the path model. These combined aggression rates are likely driven by the presence of psychological aggression (as reflected by the Spearman’s rho values between .50 and .99 when correlations between combined aggression rates and psychological aggression rates are calculated).
Path Analytic Results: Female-to-Male Aggression
Similar to the results for male-to-female aggression, this model also demonstrated excellent fit: CFI = 1.00, RMSEA = 0.00 (90% CI = [0.00, 0.05]), and SRMR = 0.01. Once again, a review of M.I.s suggested all paths could be retained in the model (though they are not depicted in the figure).
In this model, a strong association between borderline and antisocial traits, and between perpetration and victimization reports remained (Figure 3). In addition, inconsistent with the original hypothesis, but consistent with the findings for male-to-female aggression (Figure 2) as well as previous research (e.g., [Jose et al., 2014]), no significant association was found between relationship quality and any other variables. Unexpectedly, two paths that were significant for male-to-female aggression were nonsignificant when female-to-male aggression was assessed. Specifically, there was no direct association between female borderline traits and female-to-male perpetration. In addition, there was no direct association between female psychopathic traits and female alcohol use (the first-order correlation for this association was also nonsignificant).
Final model: Female-to-male aggression.
Discussion
Individual Pathology
As expected, borderline traits were significantly associated with alcohol use and aggression perpetration in males, but only the association with alcohol use remained in female-perpetrated aggression. Clearly, borderline traits seem to have a detrimental effect on individual psychopathology (i.e., alcohol problems); it is curious that borderline traits were not a risk factor for female perpetration; this finding deserves additional exploration. Further analysis of the specific MSI-BPD (Zanarini et al., 2003) items endorsed by participants suggests that the most frequently endorsed items are about emotional dysregulation, including moodiness (endorsed by 11.2% of males and 20.9% of females in the full sample) and anger (12.5% of males and 20.0% of females), whereas the least frequently endorsed items were about self-harm behaviors (1.3% of males and 2.7% of females). The low endorsement of self-harm behaviors is unsurprising in this community sample. These results suggest the affective aspects of borderline traits may drive the relationships between the variables. Strong negative affect that is not well-regulated can lead to insults, threats, and other types of verbal or physical conflict, and may also lead to a desire to self-medicate. Possibly, self-medication directly leads to conflict through difficulty suppressing these behaviors (e.g., Abroms, Fillmore, & Marczinski, 2003).
Antisocial traits were significantly associated with both aggression and substance use in males, which differs from the earlier batterer study (Jose et al., 2014). Given the batterer sample’s involvement in the justice system, underreporting, ceiling effects, or a somewhat restricted range may have precluded significant results. However, although antisocial traits were associated with aggression in female-to-male aggression, there was no significant association between antisocial traits and alcohol use for female perpetrators. The most frequently endorsed antisocial trait items were affective dysregulation, particularly “blowing [one’s] top” in response to frustration (43.0% of males and 48.4% of females in the full sample) and boredom (42.8% of males and 49.4% of females), which may reflect the high sensation-seeking needs that are often found in antisociality.
One well-known tripartite typology suggests the severity, frequency, and generality of violence as well as personality pathology can predict men’s membership into one of three groups of perpetrators (Holtzworth-Munroe & Stuart, 1994): generally violent/antisocial (GVA), dysphoric/borderline (DB), and family only (FO). Although the current study does not address typology, it provides additional evidence that antisocial and borderline traits are independently associated with psychological aggression. However, the finding that borderline and antisocial traits are significantly associated with each other also supports prior research and clinical conceptualizations that suggest a relationship between borderline and antisocial symptomatology. Moreover, a review of the most frequently endorsed items on each measure suggests that shared associations between antisocial and borderline traits may be linked to affective dysregulation. In fact, research continues to explore the ways antisocial and borderline traits independently and additively lead to aggression and other relationship problems.
Clinically, these results suggest that difficulty managing uncomfortable emotions, or a desire to experience sensation, may link personality pathology and substance use with aggression. If this is the case, the clinician may wish to make a detailed assessment of antisocial and borderline traits, particularly those reflecting affect (e.g., emotional dysregulation, boredom), and explore the use of alcohol (and related triggers) when conceptualizing dyadic functioning, especially in a couple’s management of relationship conflict. Furthermore, these findings suggest the potential usefulness of mindfulness or emotion regulation strategies in decreasing aggression in relationship conflict.
Relationship Quality
Consistent with findings in the 2014 Spanish batterer sample, this measure was not significantly associated with any other variable as hypothesized in the final model, and had very few significant associations when first-order correlations were examined. However, this result remains inconsistent with the plethora of research in the field that suggests a strong association between marital quality and relationship aggression (although the directionality of this association seems to vary across studies). Because relationship quality was uniformly high in this sample, the restricted range may have suppressed correlations. The unidimensional nature of the measure used in this study may also not capture the aspects of relationship quality that are associated with conflict and violence.
Relationship Aggression
Given the low rate of physical aggression in this sample, a combined measure of physical and psychological aggression was calculated for perpetration, and the same was done for victimization. As expected, individual psychopathology was significantly associated with male perpetration, which was, in turn, significantly linked to female victimization; this finding supports the role of individual psychopathology in relationship violence. Interestingly, the picture was somewhat less clear when female-to-male aggression was explored: Although female psychopathy and alcohol use were associated with aggression perpetration, there was no significant association between female borderline traits and aggression.
The extremely high first-order association between psychological aggression and the combined male-to-female and female-to-male scores suggests that overall aggression as measured here is driven by psychological conflict. Prior research demonstrates that if psychological aggression can be prevented or significantly reduced, the physical aggression against a partner will be reduced (e.g., O’Leary & Maiuro, 2001). Despite the significant consequences described for physical aggression, perhaps additional focus on psychological aggression can allow for targeted secondary prevention strategies to prevent physical aggression by intervening to address psychological aggression.
Notably, the prevalence of aggression in this sample was lower than rates typically found in representative U.S. samples. However, in comparison with a representative Spanish community sample (Ruiz-Pérez, Escribà-Agüir, Montero-Piñar, Vives-Cases, & Rodríguez-Barranco, 2017), female reports of physical aggression were similar though slightly lower in the present study (8.0% vs. 8.9%). The prevalence for physical aggression in this study was also similar though slightly lower than physical aggression rates in the United States, which are typically around 10% (Jose & O’Leary, 2009). Given the overall similarities between both cultures, while also taking into account possible stigma due to the relatively more recent identification of relationship violence as a serious public health concern in Spain compared with the United States, it is expected that aggression rates in Spain would be similar or perhaps slightly lower than rates in the United States.
As expected, rates for psychological victimization as reported by women were somewhat lower than U.S. community and representative samples (Jose & O’Leary, 2009). Although this rate was much higher than the psychological aggression rate found in the representative Spanish study (12.9%), Ruiz-Pérez and colleagues (Ruiz-Pérez et al., 2017) found that education and family income were protective against psychological aggression; the high rates of employment in our sample may, at least, partially explain the difference in findings. Furthermore, clients recruited in a health care setting may not have been primed to consider psychological aggression, which tends to be less concrete in nature than physical aggression, in the same way they did in the current study, which was focused on couple relationships.
Limitations and Implications
Limitations include the use of a volunteer, convenience sample, and the exclusive use of self-report to assess sensitive topics such as relationship aggression and substance use. Furthermore, data were collected cross sectionally and some of the measures (specifically the MSI-BPD and LPSP) were provided to participants in abbreviated form. This may partially explain the relatively low internal consistency of some scales. Finally, given the low base rate of physical aggression in this sample, physical aggression and psychological aggression were combined for these analyses.
However, despite these limitations, this study generally replicates findings from a Spanish batterer sample to a community sample, particularly in terms of male-to-female aggression. The findings strongly suggest a greater emphasis on the assessment and treatment of individual psychopathology, particularly in terms of affective symptoms such as dysregulation, and strategies to self-soothe (or alleviate boredom) such as substance use. Of interest is the role of differential predictors for aggression in male and female perpetrators. The results described herein suggest that antisocial traits and alcohol use may affect perpetration for both males and females, but that borderline traits may affect perpetration only for males. Future research elucidating the association between individual and dyadic risk factors for aggression in women would be valuable to further understand the current findings.
Additional clinical research on the impact of third-wave cognitive-behavioral therapy interventions in affecting both individual psychopathology symptoms and dyadic functioning would be extremely meaningful and informative. In addition, because these interventions can be conducted individually if needed, typical concerns related to confidentiality and safety in couples therapy to address IPV may be circumvented. Furthermore, these interventions may also be helpful in a psychoeducational or skills-training format for couples who are experiencing psychological aggression, presuming safety is not a concern. Overall, these results provide fertile ground for clinical research consistent with the spirit of today’s cutting edge psychotherapy.
Footnotes
Acknowledgements
The authors would like to thank Virginia Navascues for her translations; and Rini Jose, Andrew McElroy, and three anonymous reviewers for their helpful comments.
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.
