Coercive Controlling Behaviors and Reporting Physical Intimate Partner Violence in Australian Women: An Exploration

Abstract

This study explores two approaches to measuring coercive controlling behaviors (CCBs)—counting how many different CCB types and examining the frequency of each CCB experienced—to examine their utility in explaining the relationship between CCBs and physical intimate partner violence (IPV). Australian women aged 18–68 years (n = 739; M_age = 31.58, SD_age = 11.76) completed an online survey. Count and frequency CCB approaches yielded similar significant associations with increased physical IPV. Both approaches suggest that frightening behaviors in particular are significantly indicative of also experiencing physical IPV; however, when you count CCB types, public name-calling becomes important, whereas when you examine the frequency of each CCB type, jealousy/possessiveness becomes important. These findings suggest differential utility between measures of CCBs, which examine the frequency of specific CCB types and which count CCB types, and that both approaches are useful in understanding how coercion and control relate to physical violence within intimate relationships.

Keywords

intimate partner violence coercive control coercive controlling behaviors

Coercive control was first conceptualized to better emphasize the ongoing patterned use of behaviors that are often referred to as battering (Stark, 2007). Instead of a focus purely on violence severity, coercive control emphasizes the multidimensionality of oppression, harms that are cumulative rather than incident-specific, and the ongoing rather than episodic nature of abuse. This differed from other family conflict theorists who concentrated more on discrete conflict with the lack of coercive controlling behavior (CCB) and was later expanded upon by Johnson (2008) in his work on typologies of intimate partner violence (IPV). Stark (2009) describes coercion as the use of force (i.e., explicit actions used to cause immediate pain or injury) or threats (i.e., mind games and intimidation to imply the threat of force) in response to some situation or event, while control is defined as deprivation or exploitation that ensures obedience indirectly by dominating and regulating vital resources, choices, behaviors, opinions, and independence. CCBs represent behaviors that are used to intimidate, isolate, degrade, and gain compliance over an intimate partner and can ultimately cause harm/injury beyond what can be physically seen (Stark, 2007, 2013). For example, non-physical CCBs such as threats, isolation, and coercion can lead to the fear, dependency, entrapment, and inferiority commonly experienced by victims of this abuse (Johnson, 2006; Kelly & Johnson, 2008; Stark, 2013). Victims of coercive control often report the psychological effects, such as fear, anxiety, stress, and lowered self-esteem, to outweigh any physical impacts of partner victimization (Johnson, 2008; Kelly & Johnson, 2008).

While CCBs have similarities with well-known concepts such as psychological abuse and physical IPV, it is acknowledged within the literature that coercive control is its own “form” of abuse (Hamberger et al., 2017; Kelly & Johnson, 2008), perhaps even a motivating force behind the broader concept of IPV (Langhinrichsen-Rohling et al., 2012). In particular, Stark (2007) describes CCBs as an underlying dynamic that is established and maintained by the use of violence, as well as other means, therefore suggesting that CCBs are important to explaining, but not synonymous with, IPV. However, coercive control has been inconsistently defined and measured throughout the years, with CCBs and psychological IPV (Hamberger et al., 2017) and CCBs and physical IPV (Crossman & Hardesty, 2018) frequently combined. For example, nationally representative studies have found that approximately 2 million (25%) Australian women (Australian Bureau of Statistics [ABS], 2013b) and approximately 50 million (41%) American women (Black et al., 2011) reported experiences of psychological IPV, such as CCBs, at some point in their lifetime. There are also studies that have emphasized the prevalence of CCBs individually (i.e., without measuring CCBs in combination with other forms of partner abuse), reporting that 41% (n = 9,086) of American females (Breiding et al., 2014), 41% (n = 947) of new American mothers (Charles & Perreira, 2007), and 22% (n = 81) of older women (Bonomi et al., 2007) reported experiencing CCBs in an intimate relationship. However, research on the effects of coercive control (measured uniquely rather in combination with other forms of IPV) in intimate relationships is scarce and is an area requiring further investigation (Crossman & Hardesty, 2018).

The inconsistency and confusion around defining coercive control as an independent form of victimization, rather than a factor co-occurring with other forms of IPV, likely stems from its strong association with other forms of violence (Aizpurua et al., 2017). In particular, Stark (2007) suggests that physical violence can assist a perpetrator’s efforts to control their partner, which is further supported by Johnson’s (2008) typology of intimate terrorism, later defined as coercive controlling violence (Kelly & Johnson, 2008), wherein perpetrators of coercive control are described as utilizing physical violence to establish that their subsequent control tactics are a legitimate threat (i.e., physical violence is used as a sanction to make non-physical control tactics within the relationship more effective). This differs from situationally motivated IPV, which is less likely to escalate and continue over time (Kelly & Johnson, 2008) and involves exhibiting physical violence in response to a specific situation or argument between partners rather than being influenced by a need to gain control within the relationship (Johnson, 2008; Kelly & Johnson, 2008). Research has found that the perpetration of CCBs within an intimate relationship is 6 (Johnson, 2008) to 8 (Policastro & Finn, 2015) times more likely to coincide with physical IPV perpetration when compared with those not engaging in CCBs within their intimate relationships, with CCBs commonly noted as a risk factor for more frequent and severe physical IPV perpetration (Beck & Raghavan, 2010; Graham-Kevan & Archer, 2003; Myhill, 2015; Próspero, 2008). So, while it is important to acknowledge that CCBs are a distinct type of partner victimization, it is also important to recognize the relationship that CCBs have with other forms of IPV and their ability to influence (i.e., precede, motivate, or increase the likelihood of) other forms of partner abuse (Crossman & Hardesty, 2018).

The relationship between CCBs and physical IPV can be further explained by Wagers’ (2015) Internal Power Theory. The Internal Power Theory describes power and control as intrinsic to the individual, wherein control is obtained by an individual’s recognition that they can direct their own thoughts, feelings, and behaviors rather than relying on external influences/resources (Wagers, 2015). This theory outlines five components which contribute to an individual’s ability to gain power and control over their own life and outcomes—self-concept clarity, self-esteem/self-worth, self-efficacy, self-determination, and mastery over one’s life—and highlights that low internal power is associated with the use of CCBs and physical IPV (Wagers et al., 2019). Specifically, individuals with low internal power are more likely to use CCBs and physical IPV to regain power and control over their negative feelings (Wagers et al., 2019).

A review of the literature suggests that how CCBs are measured may influence observed associations between coercive control and physical partner violence. Specifically, a large proportion of research examining coercive control and physical IPV has measured CCBs using a count approach, which involves counting how many different CCB types are used (Hardesty et al., 2015); however, this approach may be problematic given perpetrators often only utilize as many, or as few, control tactics that are necessary to assert and maintain dominance within their relationship (Kelly & Johnson, 2008; Stark, 2007). Consequently, a count approach on its own may not illustrate the true nature of coercive control victimization. Other researchers have suggested that a frequency approach, which involves examining how often CCBs are used, is more appropriate (Hardesty et al., 2015) and they have used such frequency measures to define “high” and “low” levels of control within relationships (Graham-Kevan & Archer, 2003, 2008; Hardesty et al., 2015; Leone et al., 2004). These classifications of “high” and “low” control have been predominantly used to further differentiate between violence typologies such as intimate terrorism and situational violence. This study aims to build on previous research by exploring these two different approaches to measuring CCBs (count and frequency) to determine how both approaches explain the relationship between overall experiences of coercive control, as well as experiences of each specific CCB type, with physical IPV.

The Current Study

The Count Approach

When considering the count approach (i.e., examining the specific number of CCB types experienced), it is predicted that increases in the cumulative amount of unique self-reported experiences of CCB types will be associated with increased levels of self-reported physical IPV victimization. Furthermore, the number of CCB types above which people are most likely to report experiencing physical IPV will be examined; however, given that no similar research has been conducted, the threshold number of CCB types experienced for individuals to also report physical IPV will be explored as an extension to this count approach. Finally, to address the notion that perpetrators of coercive control may not need to use multiple CCB types to maintain/enforce control in the relationship, each CCB type will also be individually examined to determine their unique relationship with physical IPV. Past research has rarely reported on specific coercive control types when examining the relationship between CCBs and other forms of partner victimization; however, one study suggests that emotional-type (e.g., put-downs, humiliation, name-calling) and isolation-type (e.g., restricting activities or social outings, jealousy, suspiciousness) CCBs perpetrated by men are significant predictors of physical aggression in relationships defined as intimate terrorism (Graham-Kevan & Archer, 2008). This study therefore will investigate whether each specific CCB type is associated with increases in physical IPV, whereby it is predicted that emotional-type CCBs, such as calling one’s partner names in public, and isolation-type CCBs, such as limiting contact with family/friends, insisting on knowing who one’s partner is with, and being jealous or possessive, will be most strongly associated with self-reported physical IPV.

The Frequency Approach

When considering the frequency approach (i.e., the frequency of experiencing CCBs within an intimate relationship), it is predicted that increases in the frequency of self-reported experiences of CCBs will be associated with increased levels of self-reported physical IPV victimization. The frequency of experiencing each individual CCB type will also be examined to determine their unique relationship with physical IPV. As emotional-type and isolation-type CCBs have been shown to be important in predicting physical aggression in intimate relationships in past research (Graham-Kevan & Archer, 2008), it is predicted that increases in the frequency of experiencing emotional-type CCBs (e.g., calling one’s partner names in public) and isolation-type CCBs (e.g., limiting contact with family/friends, insisting on knowing who one’s partner is with, and being jealous or possessive) will be most strongly associated with self-reported physical IPV.

Method

Sample and Participants

The sample was obtained by combining three separate datasets collected in 2015, which formed part of the Alcohol/Drug-Involved Family Violence in Australia project (Miller et al., 2016). This collation was justified as each sample used identical surveys that were administered in the same online manner. In addition, no significant variations in the proportion of experiencing physical IPV, the outcome variable utilized in this study, between the samples was observed (χ² = 1.79, df = 2, p > .05). The majority of participants (n = 669) were recruited from two universities in Victoria, Australia. One university recruited participants from the Health Faculty via email (n = 556), and the other recruited participants from the School of Psychology via flyers and in-class discussions (n = 113). The remainder of the sample was obtained through social media (Facebook; n = 221). Combined, 890 female participants were involved in the study; however, after cases with large amounts of missing data (i.e., more than 80% missingness on key variables) were excluded (n = 151), the final sample comprised 739 females aged 18–68 years (M = 31.58, SD = 11.76). In total, 74% (n = 546) of the final sample were from a university sample, while 26% (n = 193) were recruited through social media. The sample size for this study exceeded the required number (n = 123) to achieve power of .80 and an estimated moderate effect size.

Measures

Demographics

Participants reported their age, highest education level, employment status, marital status, household income, and postcode (utilized to calculate socio-economic status [SES] using the Socio-Economic Indexes for Areas [SEIFA] Index of Relative Socio-Economic Disadvantage [IRSD] guidelines; ABS, 2013a). Over three quarters of participants in the study (76.2%; n = 563) were in a current relationship. Participants reported the length of their current/most recent relationship, which ranged from less than 1 year to 50 years. Over half (52.5%; n = 388) of the participants reported their current/most recent relationship as 5 years in length or less, with only a small proportion of participants reporting their relationship to be 20 years or more in length (10.3%; n = 76). Relevant (significant on a bivariate level) participant demographics were controlled for throughout data analysis.

CCBs

Participants’ experiences of CCBs at any time by their current/most recent partner were assessed using the nine-item Coercive Control Scale (Johnson et al., 2014). Although there is no gold standard measure of CCBs to date, many validated scales measuring coercive control exist with a common focus on measuring behaviors/tactics which intend to gain control over another person (Hamberger et al., 2017). The Coercive Control Scale was utilized in this study as it specifically measures non-physical CCBs, therefore allowing for a more distinct examination of controlling behaviors and physical violence as independent constructs. This is of particular interest as coercive control has been commonly incorporated into broader IPV measures in the past (i.e., measured in conjunction with physical IPV or psychological IPV measures) (Crossman & Hardesty, 2018; Hamberger et al., 2017), therefore limiting the ability of such scales to differentiate the relationship between violent and non violent control tactics with IPV. Example items of this scale include “my partner provokes arguments” and “my partner frightens me.” Items were rated on a 4-point Likert-type scale ranging from 1 (never) to 4 (almost always). Responses to all CCB items were summed to calculate the frequency with which CCBs were used, for a maximum possible score of 36. The frequency of experiences of each specific CCB item was also examined. Dichotomous versions of the CCB items were created by changing responses so that 0 = never and 1 = sometimes, often or almost always. A count of control tactics was calculated by summing the total number of CCB types experienced, for a maximum possible score of 9. The Coercive Control Scale demonstrated good internal consistency in previous studies (α = .70–.91; Johnson et al., 2014) and in this study (α = .89).

Physical IPV

Experiences of physical IPV within the last 12 months (or the last 12 months of participants’ most recent relationship if there is no current relationship) were identified using the Interaction Tactics Scale (ITS; Winstok, 2013). The ITS identifies how much aggression participants have shown in response to physical IPV and is used to capture static, behavioral incidences of IPV victimization as well as dynamic, interactional forms of IPV (Winstok, 2013). Despite the twofold use of this scale, this study utilized the ITS to solely determine experiences (or lack thereof) of physical IPV. Frequency of responses to IPV were ranked along a 5-point Likert-type scale, ranging from 1 (NA/no attack occurred) to 5 (occurred in most cases), and were then converted to a dichotomous variable to represent experiencing or not experiencing physical IPV. While the ITS has not specifically been examined to show construct validity as a measure of physical IPV, it is a valid measure of whether or not a respondent has experienced physical IPV. The negative response option (NA/no attack occurred) is a clear indication of no violence experienced; therefore, respondents who selected “NA/no attack occurred” were categorized as experiencing no physical IPV. In addition, the positive response options (e.g., responding to physical IPV in any way) explicitly indicate that a physical attack occurred, irrespective of how they responded to such, therefore any response other than “NA/no attack occurred” was used to indicate having experienced physical IPV. The ITS demonstrated good internal consistency in this study (α = .90).

Procedure

All participants were provided a link to access the online version of the survey, which was accessed through Survey Monkey, via email (university sample one), flyers and in-class discussions (university sample two), or Facebook (social media sample). Participants recruited from university sample one were contacted by the university (on behalf of the research team), whereby a bulk email invite was sent to all student and staff email accounts within the Faculty of Health inviting participation in the study. University sample two utilized flyers, which were presented on noticeboards throughout the university, and word-of-mouth (discussions in second- and third-year psychology classes) to recruit participants. The social media sample used a small paid Facebook advertisement and a snowball sampling technique to recruit participants on Facebook. The participants took approximately 20 min to complete the survey.

Data Analytic Strategy

Bivariate correlations were first conducted between the CCB items to identify the magnitude and direction of associations between the various coercive control types. Multiple Pearson point-biserial correlations and chi-square analyses were then conducted to examine whether participant demographics were positively related to self-reported experiences of physical IPV. Bivariately significant demographics were then controlled for in subsequent analyses. Hierarchical logistic regression analyses were conducted to determine the extent that CCBs were associated with the likelihood of self-reporting physical IPV using both count and frequency approaches to measuring coercive control. Receiver operating characteristic (ROC) analyses and area under the curve (AUC; Hosmer & Lemeshow, 2000) values were also calculated to investigate whether there is a specific number of CCB types experienced at which the association with physical IPV is strongest. Sensitivity (true positive rate) and specificity (avoiding false positives) values were calculated and interpreted according to conventional guidelines (Hosmer & Lemeshow, 2000).

Results

Bivariate Associations

Multicollinearity between the nine CCB items was not observed (Table 1), with significant associations ranging from small positive to large positive associations.

Table 1.

Correlations Between the Nine CCB Items.

CCB type	CCB 1	CCB 2	CCB 3	CCB 4	CCB 5	CCB 6	CCB 7	CCB 8	CCB 9
CCB 1	1
CCB 2	.57**	1
CCB 3	.61**	.58**	1
CCB 4	.69**	.53**	.59**	1
CCB 5	.42**	.52**	.47**	.40**	1
CCB 6	.50**	.60**	.60**	.49**	.57**	1
CCB 7	.42**	.61**	.46**	.43**	.54**	.61**	1
CCB 8	.54**	.64**	.64**	.49**	.55**	.66**	.65**	1
CCB 9	.26**	.30**	.41**	.32**	.40**	.44**	.33**	.41**	1

Note. CCB = coercive controlling behavior; CCB 1 = my partner is jealous or possessive; CCB 2 = my partner provokes arguments; CCB 3 = my partner limits contact with family or friends; CCB 4 = my partner insists on knowing who I am with; CCB 5 = my partner calls me names in public; CCB 6 = my partner makes me feel inadequate; CCB 7 = my partner shouts or swears; CCB 8 = my partner frightens me; CCB 9 = my partner prevents knowledge of or access to income.

p < .01.

Demographics and Physical IPV

Pearson point-biserial correlations revealed a significant weak positive relationship between age and self-reporting physical IPV (r = .11, n = 739, p < .001). Chi-square analyses revealed a small but significant relationship between marital status and reporting physical IPV (χ² = 21.81, df = 2, ϕ_c = .17, p < .001). Examination of the standardized residuals indicated that those who are no longer married have higher proportions of reporting physical IPV than those who have never been married or are currently married/in a de facto relationship. No significant relationships were found between SES, income, employment, or education and reporting physical IPV.

Count Approach to Examining CCBs and Physical IPV

Overall number of CCB types experienced. A hierarchical binary logistic regression analysis was conducted to test whether, after controlling for significant demographics (age and marital status), the cumulative amount of unique self-reported experiences of CCB types were positively related to self-reported physical IPV (Table 2). Age, marital status, and total CCB score were significantly associated with physical IPV; χ² = 76.12, df = 4, p < .001. The final model accounted for approximately 11% (R²_LL = .11) of the variance explained in reported physical IPV, representing a medium effect (Cohen, 1988). A one-unit increase in age was associated with a 3% increase in the odds of reporting physical IPV, and each additional CCB type reported corresponds to a 1.34 times the odds of reporting physical IPV.

Table 2.

Hierarchical Logistic Regression Model of the Relationship Between the Number of CCBs Experienced and Reporting Physical IPV.

Predictor variables	Model 1			Model 2
Predictor variables	b	OR	95% CI	b	OR	95% CI
Age	0.02	1.02*	[1.01, 1.04]	0.03	1.03**	[1.01, 1.05]
Married/de-facto		1.00			1.00
Never married	0.48	1.62	[0.99, 2.65]	0.41	1.50	[0.90, 2.51]
No longer married	1.19	3.28***	[1.73, 6.21]	0.41	1.50	[0.73, 3.07]
Total CCB score				0.29	1.34***	[1.24, 1.46]

Note. CCB = coercive controlling behavior; IPV = intimate partner violence; b = unstandardized b-weight; OR = odds ratio; CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Exploratory CCB threshold

The AUC statistic is an indicator of overall predictive value, whereby a value of 1 represents a perfect predictor of reporting physical IPV, and a value of .50 is no better than chance (Hosmer & Lemeshow, 2000). Of the nine CCB types, eight produced AUC values between .61 and .65, representing a fair chance of predicting physical IPV (Table 3).

Table 3.

ROC/AUC Analysis to Determine a Cut-Point for the Number of CCBs Experienced and the Reporting of Physical IPV.

CCB type	ROC curve			Sensitivity	1 − specificity
CCB type	AUC	95% CI	SE	Sensitivity	1 − specificity
Jealous or possessive	.61***	[.56, .67]	0.03	.56	.34
Provokes arguments	.62***	[.56, .67]	0.03	.57	.34
Limits contact with family or friends	.61***	[.55, .66]	0.03	.32	.11
Insists on knowing who I am with	.62***	[.56, .67]	0.03	.52	.28
Calls me names in public	.62***	[.56, .67]	0.03	.30	.07
Makes me feel inadequate	.63***	[.58, .69]	0.03	.48	.22
Shouts or swears	.61***	[.55, .66]	0.03	.59	.37
Frightens me	.65***	[.59, .71]	0.03	.41	.12
Prevents knowledge of or access to income	.54	[.49, .60]	0.03	.14	.05

Note. ROC = receiver operating characteristic; AUC = area under the curve; CCB = coercive controlling behavior; IPV = intimate partner violence; CI = confidence interval.

***

p < .001.

A series of dichotomous composite variables were then calculated, grouping participants into those who had, or had not, experienced increasing numbers of CCB types (Table 4). This was done with the intention of outlining a specific criterion above which people are most likely to report experiencing physical IPV. Given the similarity of resultant AUC values across each CCB type, the choice of an appropriate cut-point was based on a combination of sensitivity and specificity scores. Reporting three or more (3+) CCB types produced the strongest sensitivity value and an AUC value of .65, which is considered fair. This cut-point represents the prediction that an individual will report experiences of physical IPV with 60% accuracy and with a false positive rate of 30%.

Table 4.

ROC/AUC Analysis Using Composite Variables to Determine a Cut-Point for the Number of CCBs Experienced and the Reporting of Physical IPV.

Composite CCB variable	ROC curve			Sensitivity	1 − specificity
Composite CCB variable	AUC	95% CI	SE	Sensitivity	1 − specificity
2+ CCB types	.59***	[.54, .65]	0.03	.66	.47
3+ CCB types	.65***	[.60, .71]	0.03	.60	.30
4+ CCB types	.64***	[.50, .70]	0.03	.48	.20
5+ CCB types	.66***	[.61, .72]	0.03	.43	.10
6+ CCB types	.65***	[.59, .71]	0.03	.36	.06
7+ CCB types	.62***	[.56, .68]	0.03	.28	.04

Note. This analysis excludes the CCB item my partner prevents knowledge of or access to income. ROC = receiver operating characteristic; AUC = area under the curve; CCB = coercive controlling behavior; IPV = intimate partner violence; CI = confidence interval.

***

p < .001.

Individual CCB types

A hierarchical binary logistic regression analysis was conducted to test whether, after controlling for significant demographics (age and marital status), each individual CCB type was positively related to self-reported physical IPV (Table 5). Age and two CCB types (“my partner calls me names in public,” and “my partner frightens me”) were significantly associated with physical IPV; χ² = 90.09, df = 12, p < .001. The final model accounted for approximately 13% (R²_LL= .13) of the variance explained in reported physical IPV, representing a medium effect (Cohen, 1988). A one-unit increase in age was associated with a 3% increase in the odds of reporting physical IPV, while indicating that your partner calls you names in public and frightens you corresponds to a 2.50 times and a 2.77 times the odds, respectively, of reporting physical IPV.

Table 5.

Hierarchical Logistic Regression Model of the Relationship Between Each CCB Type and Reporting Physical IPV.

Predictor variables	Model 1			Model 2
Predictor variables	b	OR	95% CI	b	OR	95% CI
Age	0.02	1.02*	[1.01, 1.04]	−0.03	1.03**	[1.01, 1.06]
Married/de-facto		1.00			1.00
Never married	0.48	1.62	[0.99, 2.65]	−0.39	1.48	[0.87, 2.52]
No longer married	1.19	3.28***	[1.73, 6.22]	−0.48	1.62	[0.77, 3.43]
Jealous or possessive				−0.14	1.15	[0.67, 1.95]
Provokes arguments				−0.10	0.90	[0.52, 1.56]
Limits contact with family or friends				−0.08	1.08	[0.57, 2.07]
Insists on knowing who I am with				−0.51	1.66	[0.97, 2.83]
Calls me names in public				−0.92	2.50**	[1.33, 4.72]
Makes me feel inadequate				−0.24	1.34	[0.77, 2.35]
Shouts or swears				0.00	1.00	[0.59, 1.71]
Frightens me				−10.2	2.77***	[1.54, 5.00]
Prevents knowledge of or access to income				−0.47	0.63	[0.27, 1.45]

Note. CCB = coercive controlling behavior; IPV = intimate partner violence; b = unstandardized b-weight; OR = odds ratio; CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Frequency Approach to Examining CCBs and Physical IPV

Overall frequency of CCBs

A hierarchical logistic regression analysis was conducted to test whether, after controlling for significant demographics (age and marital status), the frequency of self-reported experiences of CCB were positively related to self-reported physical IPV (Table 6). Age and overall frequency of CCBs were significantly associated with physical IPV, χ² = 80.86, df = 4, p < .001. The final model accounted for approximately 12% (R²_LL = .12) of the variance explained in reported physical IPV, representing a medium effect (Cohen, 1988). A one-unit increase in age was associated with a 3% increase in the odds of reporting physical IPV, and a one-unit increase in the frequency of CCB experiences was associated with a 17% increase in the odds of reporting physical IPV.

Table 6.

Hierarchical Logistic Regression Model of the Relationship Between the Frequency of Experiencing CCBs and Reporting Physical IPV.

Predictor variables	Model 1			Model 2
Predictor variables	b	OR	95% CI	b	OR	95% CI
Age	0.02	1.02*	[1.01, 1.04]	0.03	1.03**	[1.01, 1.05]
Married/de-facto		1.00			1.00
Never married	0.48	1.62	[0.99, 2.65]	0.38	1.46	[0.87, 2.44]
No longer married	1.19	3.28***	[1.73, 6.21]	0.15	1.16	[0.54, 2.51]
Frequency CCB experiences				0.16	1.17***	[1.12, 1.23]

Note. Frequency of CCB experiences was scored on a scale of 9–36, with a score of 9 representing no CCBs experienced. CCB = coercive controlling behavior; IPV = intimate partner violence; b = unstandardized b-weight; OR = odds ratio; CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Frequency of experiencing each individual CCB type

A hierarchical logistic regression analysis was conducted to test whether, after controlling for significant demographics (age and marital status), the frequency of experiencing each individual CCB type was positively related to self-reported physical IPV (Table 7). Age and the frequency of experiencing two CCB types (“my partner is jealous or possessive” and “my partner frightens me”) were significantly associated with physical IPV, χ² = 90.09, df = 12, p < .001, whereby those experiencing jealous or possessive control tactics almost all of the time, and frightening control tactics sometimes or often, were significantly more likely to also report experiencing physical IPV. The final model accounted for approximately 17% (R²_LL = .17) of the variance explained in reported physical IPV, representing a medium effect (Cohen, 1988). A one-unit increase in age was associated with a 3% increase in the odds of reporting physical IPV, while indicating that your partner is jealous or possessive all of the time, frightens you sometimes, or frightens you often, corresponds to a 6.06, 2.20, and 6.83 times the odds, respectively, of reporting physical IPV.

Table 7.

Hierarchical Logistic Regression Model of the Relationship Between the Frequency of Experiencing Each CCB Type and Reporting Physical IPV.

Predictor variables	Model 1			Model 2
Predictor variables	b	OR	95% CI	b	OR	95% CI
Age	0.02	1.02*	[1.01, 1.04]	0.03	1.03**	[1.01, 1.06]
Married/de-facto		1.00			1.00
Never married	0.48	1.62	[0.99, 2.65]	0.40	1.49	[0.86, 2.59]
No longer married	1.19	3.28***	[1.73, 6.22]	0.36	1.43	[0.62, 3.28]
Jealous or possessive				0.11^a 0.09^b 1.80^c	1.11^a 1.10^b 6.06^c*	[0.64, 1.93]^a [0.39, 3.06]^b [1.23, 29.77]^c
Provokes arguments				−0.18^a 0.44^b −0.24^c	0.84^a 1.55^b 0.79^c	[0.47, 1.49]^a [0.56, 4.32]^b [0.14, 4.27]^c
Limits contact with family or friends				−0.30^a −0.74^b −0.41^c	0.74^a 0.48^b 0.66^c	[0.33, 1.68]^a [0.11, 2.10]^b [0.07, 5.97]^c
Insists on knowing who I am with				0.48^a 0.85^b 0.03^c	1.61^a 2.35^b 1.03^c	[0.91, 2.87]^a [0.79, 6.94]^b [0.24, 4.44]^c
Calls me names in public				0.69^a 1.22^b −0.98^c	2.00^a 3.39^b 0.38^c	[0.92, 4.37]^a [0.63, 18.08]^b [0.02, 6.70]^c
Makes me feel inadequate				0.29^a −0.05^b 0.04^c	1.34^a 0.95^b 1.04^c	[0.73, 2.46]^a [0.33, 2.75]^b [0.22, 4.90]^c
Shouts or swears				0.05^a 0.11^b −0.18^c	1.05^a 1.12^b 0.84^c	[0.60, 1.83]^a [0.42, 2.99]^b [0.17, 4.04]^c
Frightens me				0.79^a 1.92^b 1.99^c	2.20^a* 6.83^b** 7.32^c	[1.12, 4.32]^a [1.65, 28.26]^b [0.56, 96.25]^c
Prevents knowledge of or access to income				−0.94^a 1.45^b −0.78^c	0.39^a 4.26^b 0.46^c	[0.12, 1.28]^a [0.58, 31.43]^b [0.06, 3.56]^c

Note. Each CCB (entered in at Model 2) included four response options; the reference category was “Never,” while the other three categories and their respective statistics are indicated by the following superscript. CCB = coercive controlling behavior; IPV = intimate partner violence; b = unstandardized b-weight; OR = odds ratio; CI = confidence interval.

Sometimes. ^bOften. ^cAlmost Always.

p < .05. **p < .01. ***p < .001.

Discussion

This study investigates two different approaches to measuring CCBs (count and frequency) to better understand the usefulness of both approaches for explaining the relationship between coercive control and experiences of physical IPV in a sample of Australian women. This extends previous research by looking at the utility of using different approaches to measuring coercive control with the goal of understanding the most important questions to ask victims of partner abuse.

The hypotheses that increases in the cumulative amount of unique CCB types (overall count approach) and increases in the frequency of CCB experiences (overall frequency approach) will be positively associated with increased levels of physical IPV were supported, with both approaches resulting in moderate associations with physical IPV. These findings are consistent with past research showing positive associations between CCBs and physical IPV experiences (Johnson, 2008). Previous research has concluded that when classifying “high” and “low” control within relationships, a frequency approach to measuring CCBs is more appropriate than a count approach (Hardesty et al., 2015); however, the current findings suggest similar utility in terms of variance explained in physical IPV when examining count and frequency approaches to measuring CCBs. This suggests that the choice between measuring CCBs using an overall count or an overall frequency approach likely depends on the goal of the research. It should be noted, however, that a frequency approach to measuring coercive control, wherein you ask about the experience of CCB types and their frequency, necessarily contains more information than a count of CCB types experienced and therefore should be prioritized when possible. Furthermore, it should be acknowledged that coercive control-type behaviors can occur to some degree in healthy relationships (e.g., a partner shouting or swearing on occasion out of anger/frustration) which may affect researchers’ ability to assess when such behaviors reach an abusive level. In particular, this is especially problematic when utilizing a count approach as respondents may answer “yes” to experiencing some CCBs due to how these items are worded, even though these behaviors could reflect non abusive scenarios. The frequency approach is therefore likely to be more effective in differentiating cases that are genuinely ongoing/abusive.

Further exploration of the count approach via investigation of determining a threshold number of unique CCB types experienced for predicting physical IPV experiences was also conducted, revealing that experiencing three or more (3+) CCB types was a fair cut-point (Hosmer & Lemeshow, 2000). The AUC value associated with this cut-point is modest, but it is predictive at a level better than chance. This investigation of a threshold above which people are likely to report physical IPV is unique to IPV literature and provides an exploratory first step in the development of a possible new technique for assisting in predicting whether individuals reporting on CCBs would also likely report experiencing physical IPV. The use of CCBs in this way is a potentially less confrontational and less biased approach (i.e., less likely to be influenced by stigma concerning violence as being physical) than directly asking about physical abuse. Given the sensitive nature of researching IPV, the proposed 3+ cut-point should be considered in future research where directly asking about physical IPV is not appropriate.

This study further explores the association between CCBs and physical IPV via examination of each individual CCB type. It was hypothesized that each specific CCB type would be positively associated with physical IPV, with emotional-type and isolation-type CCBs predicted to be most strongly associated. This hypothesis was partially supported, wherein two of the nine CCB types (“my partner calls me names in public” and “my partner frightens me”) were significantly associated with increases in reporting physical IPV. The significant relationship between public name-calling and physical IPV is consistent with past research which suggests that emotional-type control tactics are more strongly related to physical IPV (Graham-Kevan & Archer, 2008); however, contrary to past findings, isolation-type CCBs, such as limiting contact with family/friends, insisting on knowing who you are with, and being jealous or possessive, were not significant predictors of physical IPV. However, due to a lack of research concerning specific CCB types and their individual association with physical IPV experiences, this study may not be directly comparable to the study conducted by Graham-Kevan and Archer (2008) due to using differing measures of coercive control. The association between frightening behaviors and physical IPV suggests that experiencing control tactics that imply more escalatory behavior within the relationship, or a level of aggression that could be perceived as visible to other people (including the victim), more frequently co-occur with physical IPV.

The hypothesis that the frequency of experiencing each CCB type would be positively associated with also reporting physical IPV, with emotional and isolation CCB types being most strongly associated, was partially supported, whereby experiencing jealousy or possessiveness “almost always” and having a partner that frightens you “sometimes” or “often” were significantly associated with increases in reported physical IPV experiences. The relationship between experiencing jealous or possessive controlling behaviors and experiencing physical IPV is consistent with past research (Graham-Kevan & Archer, 2008); however, the importance of frightening behaviors is unexpected. Frightening behaviors are often considered an important element of coercive control in the literature (Myhill, 2015), but have rarely been examined individually in terms of its impact on other forms of partner victimization (Graham-Kevan & Archer, 2008). Specifically, the current finding suggests that being frightened by your partner “sometimes” or “often” is associated with physical IPV, but experiencing this frightening behavior “almost always” is not. This may indicate that perpetrators who use frightening behaviors very frequently in their relationships are able to maintain a sense of control without having to escalate this behavior to physical forms of violence.

Overall, this study was able to highlight the differences in using a count or frequency approach to measuring specific CCBs; some CCBs are more related to experiencing physical IPV than others, and count and frequency approaches emphasize the importance of different CCBs. Specifically, while both approaches highlight the importance of one particular CCB (frightening behavior), they also each point toward a different additional CCB type: public name-calling (count approach) and being jealous/possessive (frequency approach). This suggests that examining each CCB type uniquely is important, but also that the way in which specific CCB types are explored (i.e., experiences of each CCB type, or the frequency of each CCB type) is relevant to consider. This further highlights the complex nature of coercive control and suggests the need for multiple approaches, which measure more than just an overall CCB score, to be utilized when examining this construct. Furthermore, given the emphasis of different CCBs between approaches, future research into the relative associations between these CCB types and other negative outcomes (e.g., mental health) may further tease out the relative utility of count and frequency approaches to measuring coercive control.

Limitations

First, the limitations associated with the scales used to measure CCBs and physical IPV should be noted. This study chose to examine non-physical CCBs to remove the confound of physicality in their relationship with physical IPV. This choice means the current findings are only applicable to non-physical CCBs. In addition, this could be extended by investigating the broader CCB construct, which researchers have suggested to also include motivation or intentionality of CCBs and the ability of the perpetrator to make a credible threat (Hamberger et al., 2017). Furthermore, there are some items within the Coercive Control Scale that do not specify that the behavior in question is directed toward the victim (i.e., “my partner provokes arguments” or “my partner shouts or swears”); therefore, such items could be revised in future uses of this scale to ensure all items are answered regarding respondents’ violence-related experiences toward them. Moreover, the stem question for our measure of physical IPV, the ITS, does not explicitly ask individuals to answer if they have experienced physical IPV, but instead asks how they have responded to incidents of physical IPV, allowing a choice of “no attack occurred” or a response option describing their response to such an incident. While this approach should still validly elicit a response indicating experiencing IPV or not, it is possible that a simpler and more direct question may elicit a different response. In addition, this measure may somewhat mask some of the biases against reporting IPV that the CCB items were intended to capture; therefore, future researchers may wish to simplify their measure of physical IPV to reduce the degree of subjectivity associated with measuring such experiences.

Second, the AUC values identified for predicting physical IPV were only slightly beyond that of chance. This suggests that the accuracy of this threshold is fair, but should not be used in isolation. In addition, this study only focused on the association between CCBs and reporting physical IPV; however, other types of IPV (i.e., sexual IPV) could also be considered in future research to allow for a more comprehensive understanding of CCBs and differing types of partner victimization.

Finally, the current sample was drawn from two Australian universities and social media users; therefore, findings may not necessarily generalize to other populations. With approximately three quarters of participants being obtained from university-based samples, this may have also contributed to the differences seen in regard to demographic details in this study compared with those outlined in previous studies in this field. In particular, the differences in the characteristics of the current sample may explain some of our non-significant findings. For example, over 70% of the current sample earned over AUD50,000 annually (approx. USD34,000); however, research indicates that individuals with lower incomes (i.e., below AUD36,000/USD25,000) are more likely to report IPV than those with higher incomes (Cunradi et al., 2002; Sorenson et al., 1996). Furthermore, the predominantly university-based sample may have skewed our findings pertaining to education and income. Our findings therefore need to be considered in light of this somewhat more affluent and educated sample. In addition, this study only included females; therefore, it should be noted that different patterns of findings may emerge for men. Furthermore, the use of cross-sectional data limits the ability to determine time-order effects concerning the association between CCBs and physical IPV. Future studies may benefit from examining this association longitudinally.

Implications and Conclusion

The current exploratory study demonstrates that the association between coercive control and experiences of physical IPV may differ based on the approach taken to measure CCBs (i.e., count or frequency), especially when considering specific CCB types rather than composite scores or overall frequency scores. The current findings indicate that certain CCB types, including public name-calling, frightening behavior, and jealousy or possessiveness, are individually associated with victims’ reports of physical IPV, therefore alluding to particular types of non-physical control tactics which are more likely to co-occur with physical partner violence. Furthermore, the different associations with physical IPV seen when examining the experiences of each specific CCB type and the frequencies of experiencing each specific CCB type suggest that multiple approaches to measuring coercive control may be appropriate when exploring its relationship to other forms of partner victimization. From this perspective, it may be important for professionals to specifically screen for these particularly important behaviors and their frequency to target interventions to curtail or minimize their occurrence. However, further research may wish to build on this exploratory study by measuring all aspects of coercive control (i.e., motivation and intentionality of CCBs) for a more comprehensive overview of this relationship. Furthermore, it is important to note that the current findings do not specify which form of partner victimization occurred first (i.e., CCBs or physical IPV), or any such causal relationship; rather, this exploration examines the co-occurrence of partner victimization.

The current findings also indicate that asking women about experiences of CCBs in their relationships represents a potentially promising technique to screen for questions about physical partner violence. However, due to the modest cut-point identified and the use of probabilities to determine such cut-point, the proposed 3+ CCB threshold for predicting the likelihood of physical IPV should be used only as part of a comprehensive assessment in line with the views of professionals working in the violence sector (Boxall et al., 2015). If validated in additional populations, the 3+ CCB cut-point may prove a helpful adjunct to other reliable and valid proxy IPV measures which are used to initially identify IPV. This study supports the position that the use of more sophisticated violence typologies, which include counts of CCB types as well as their frequency, may indeed be applicable to the day-to-day work of violence practitioners (Boxall et al., 2015).

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: The current study was funded by the National Drug and Law Enforcement Research Fund.

ORCID iD

Brittany Patafio

Author Biographies

Brittany Patafio, Bachelor of Psychology (Honors), is a PhD candidate in the School of Psychology at Deakin University. Her research interests focus on mental health literacy and help-seeking in youth, community safety, and violence typologies.

Peter Miller, PhD, is a professor of violence prevention and addiction studies at the School of Psychology, Deakin University. His research interests include alcohol and drug-related violence, predictors of violence (including family and domestic violence), and Corporate Political Activity of alcohol and dangerous consumptions industries.

Arlene Walker, PhD, is a registered psychologist and a senior lecturer in Organizational Psychology at Deakin University. Her research focuses on employee health and well-being and the impact of family and domestic violence on the workplace.

Kerri Coomber, PhD, is a research fellow in the School of Psychology at Deakin University. Her research focuses on the epidemiology of alcohol and drug-related harms and violence. Her work builds theoretically informed models that drive policy and legislative change.

Ashlee Curtis, PhD, holds a jointly funded research fellow position with Deakin University and Odyssey House Victoria. Her research focuses on the complex relationship between substance use and offending behavior, particularly in regard to responsivity to interventions.

Gery Karantzas, PhD, is an associate professor in the School of Psychology at Deakin University. He received his PhD from La Trobe University in 2008 and his research focuses on couple and family relationships, including partner maltreatment and aggression. He has over 70 publications and his research activities have been funded by grants from the Australian Research Council, the Academy of Social Sciences in Australia, the National Medical Health and Research Council, and beyondblue.

Richelle Mayshak, PhD, is a lecturer at Deakin University and an active researcher within the Violence Prevention Group and CEDAAR in the School of Psychology. She has a keen interest in the management of large datasets, and her research interests include alcohol-related violence; links between social media use and violence; and sexual violence, alcohol, and drug use.

Nicholas Taylor, Graduate Diploma (Psychology), is a PhD candidate in the School of Psychology at Deakin University. His research focuses on how policy interventions and environmental factors influence alcohol and drug-related harms and violence.

Shannon Hyder, PhD, is a lecturer in the School of Psychology at Deakin University. His research focuses on violence typologies and their intersection with mental health. He teaches across several research methods units in undergraduate and postgraduate courses and is a member of the Psychology department’s data science team.

References

Aizpurua

Copp

Ricarte

J. J.

Vázquez

(2017). Controlling behaviors and intimate partner violence among women in Spain: An examination of individual, partner, and relationship risk factors for physical and psychological abuse. Journal of Interpersonal Violence, 36(1–2), 231–254.

Australian Bureau of Statistics. (2013a). Census of population and housing: Socio-Economic Indexes for Areas (SEIFA), Australia, 2011. http://www.abs.gov.au/ausstats/abs@.nsf/Lookup/by%20Subject/2033.0.55.001~2011~Main%20Features~Main%20Page~1

Australian Bureau of Statistics. (2013b). Personal safety, Australia, 2012: Experience of emotional abuse by a partner. http://www.abs.gov.au/ausstats/abs@.nsf/mf/4906.0

Beck

C. J.

Raghavan

(2010). Intimate partner abuse screening in custody mediation: The importance of assessing coercive control. Family Court Review, 48(3), 555–565. https://doi.org/10.1111/j.1744-1617.2010.01329.x

Black

M. C.

Basile

K. C.

Breiding

M. J.

Smith

S. G.

Walters

M. L.

Merrick

M. T.

Chen

Stevens

M. R.

(2011). The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. https://www.cdc.gov/violenceprevention/pdf/nisvs_report2010-a.pdf

Bonomi

A. E.

Anderson

M. L.

Reid

R. J.

Carrell

Fishman

P. A.

Rivara

F. P.

Thompson

R. S.

(2007). Intimate partner violence in older women. The Gerontologist, 47(1), 34–41. https://doi.org/10.1093/geront/47.1.34

Boxall

Rosevear

Payne

(2015). Domestic violence typologies: What value to practice? Trends and Issues in Crime and Criminal Justice, 494, 1–9. https://www.aic.gov.au/publications/tandi/tandi494

Breiding

M. J.

Chen

Black

M. C.

(2014). Intimate partner violence in the United States—2010. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention.

Charles

Perreira

(2007). Intimate partner violence during pregnancy and 1-year post-partum. Journal of Family Violence, 22(7), 609–619. https://doi.org/10.1007/s10896-007-9112-0

10.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum.

11.

Crossman

K. A.

Hardesty

J. L.

(2018). Placing coercive control at the center: What are the processes of coercive control and what makes control coercive? Psychology of Violence, 8(2), 196–206. https://doi.org/10.1037/vio0000094

12.

Cunradi

C. B.

Caetano

Schafer

(2002). Socioeconomic predictors of intimate partner violence among white, black, and Hispanic couples in the United States. Journal of Family Violence, 17(4), 377–389. https://doi.org/10.1023/A:1020374617328

13.

Graham-Kevan

Archer

(2003). Intimate terrorism and common couple violence. A test of Johnson’s predictions in four British samples. Journal of Interpersonal Violence, 18, 1247–1270. http://doi.org/10.1177/0886260503256656

14.

Graham-Kevan

Archer

(2008). Does controlling behavior predict physical aggression and violence to partners? Journal of Family Violence, 23(7), 539. https://doi.org/10.1007/s10896-008-9162-y

15.

Hamberger

L. K.

Larsen

S. E.

Lehrner

(2017). Coercive control in intimate partner violence. Aggression and Violent Behavior, 37, 1–11. http://doi.org/10.1016/j.avb.2017.08.003

16.

Hardesty

J. L.

Crossman

K. A.

Haselschwerdt

M. L.

Raffaelli

Ogolsky

B. G.

Johnson

M. P.

(2015). Toward a standard approach to operationalizing coercive control and classifying violence types. Journal of Marriage and Family, 77(4), 833–843. https://doi.org/10.1111/jomf.12201

17.

Hosmer

D. W.

Lemeshow

(2000). Applied logistic regression. John Wiley.

18.

Johnson

M. P.

(2006). Conflict and control: Gender symmetry and asymmetry in domestic violence. Violence Against Women, 12(11), 1003–1018. https://doi.org/10.1177/1077801206293328

19.

Johnson

M. P.

(2008). A typology of domestic violence: Intimate terrorism, violent resistance, and situational couple violence. http://ebookcentral.proquest.com/lib/deakin/detail.action?docID=1084935

20.

Johnson

M. P.

Leone

J. M.

(2014). Intimate terrorism and situational couple violence in general surveys: Ex-spouses required. Violence Against Women, 20(2), 186–207. https://doi.org/10.1177/1077801214521324

21.

Kelly

J. B.

Johnson

M. P.

(2008). Differentiation among types of intimate partner violence: Research update and implications for interventions. Family Court Review, 46(3), 476–499. https://doi.org/10.1111/j.1744-1617.2008.00215.x

22.

Langhinrichsen-Rohling

McCullars

Misra

T. A.

(2012). Motivations for men and women’s intimate partner violence perpetration: A comprehensive review. Partner Abuse, 3(4), 429–468. http://doi.org/10.1891/1946-6560.3.4.429

23.

Leone

J. M.

Johnson

M. P.

Cohan

C. L.

Lloyd

S. E.

(2004). Consequences of male partner violence for low-income minority women. Journal of Marriage and Family, 66(2), 472–490. https://doi.org/10.1111/j.1741-3737.2004.00032.x

24.

Miller

Cox

Costa

Mayshak

Walker

Hyder

Tonner

Day

(2016). Alcohol/Drug-Involved Family Violence in Australia (ADIVA): Final report. https://www.aic.gov.au/publications/ndlerfmonograph/ndlerfmonograph68

25.

Myhill

(2015). Measuring coercive control: What can we learn from national population surveys? Violence Against Women, 21(3), 355–375. https://doi.org/10.1177/1077801214568032

26.

Policastro

Finn

M. A.

(2015). Coercive control and physical violence in older adults: Analysis using data from the national elder mistreatment study. Journal of Interpersonal Violence, 32(3), 311–330. https://doi.org/10.1177/0886260515585545

27.

Próspero

(2008). Effects of masculinity, sex, and control on different types of intimate partner violence perpetration. Journal of Family Violence, 23(7), 639–645. https://doi.org/10.1007/s10896-008-9186-3

28.

Sorenson

S. B.

Upchurch

D. M.

Shen

(1996). Violence and injury in marital arguments: Risk patterns and gender differences. American Journal of Public Health, 86(1), 35–40. https://doi.org/10.2105/AJPH.86.1.35

29.

Stark

(2007). Coercive control: How men entrap women in personal life. http://ebookcentral.proquest.com.ezproxy-f.deakin.edu.au/lib/deakin/detail.action?docID=415172

30.

Stark

(2009). Coercive control: The entrapment of women in personal life. Oxford University Press.

31.

Stark

(2013). The dangers of dangerousness assessment. Family & Intimate Partner Violence Quarterly, 6(2), 13–22. http://www.fcadv.org/sites/default/files/Stark%2020132.pdf

32.

Wagers

S. M.

(2015). Deconstructing the “power and control motive”: Moving beyond a unidimensional view of power in domestic violence theory. Partner Abuse, 6, 230–242. https://doi.org/10.1891/1946-6560.6.2.230

33.

Wagers

S. M.

Wareham

Sellers

C. S.

(2019). Testing the validity of an internal power theory of interpersonal violence. Journal of Interpersonal Violence. Advance online publication. https://doi.org/10.1177/088626051983409

34.

Winstok

(2013). From a static to a dynamic approach to the study of partner violence. Sex Roles, 69(3–4), 193–204. https://doi.org/10.1007/s11199-013-0278-z