Male and Female Perpetrated Partner Abuse: Testing a Diathesis-Stress Model 
by Reena Sommer


Table of Contents

Chapter 3

Chapter 5


CHAPTER 4 


WINNIPEG HEALTH AND DRINKING 
SURVEY - WAVE 2

                        Methodology

     This chapter describes the methodology employed throughout this
     longitudinal project.  Having already addressed this project's
     sampling technique and procedure of data collection in the previous
     chapter, a brief description of the Wave 2 response rates will be
     presented in the following section.
 

     Response Rate

     Data collection for Wave 2 began during the summer of 1991 and
     was completed by the fall of 1992.  Of the 1257 persons interviewed
     in Wave 1, 83 moved away from the city, eight were deemed ineligible
     (i.e., died or were institutionalized), 57 could not be contacted and
     121 refused to be reinterviewed. The number of completed interviews
     for Wave 2 was 988 and provided an overall response rate of 78.7
     percent.
 

           Independent Variables and Measures

     The following is a list and explanation of the instruments and
     measures utilized throughout both waves of this project.  A battery
     of demographic, alcohol abuse, personality and partner abuse items
     were administered in Wave 1 and then were readministered in Wave 2.
     Items designed to assess changes that might have occurred during the
     two year period between Wave 1 and Wave 2 as well as questions that
     might also elicit further information regarding the circumstances
     surrounding episodes of partner abuse were added to the follow-up
     interviews in Wave 2.  These items will be noted below where
     applicable.

     Demographic variables. Questions were included to measure the
     following demographic variables: (1) age, (2) gender, (3) marital
     status, (4) race, (5) religion, (6) employment status and (7) income.
     Items regarding changes with respect to marital status, family size,
     employment status, as well as income were added to the Wave 2
     questionnaire (Appendix B. Part 1).

     Alcohol abuse.  Two constructs were selected to assess the extent of
     alcohol abuse by the respondents (i.e., alcohol consumption and
     alcohol dependence).

     1) Alcohol consumption:

     Alcohol consumption was measured by the Volume Variability Index
     (Cahalan & Cisin, 1986; Room, 1982).  This instrument contained nine
     questions and measured the quantity-frequency (with an added
     indicator for binge drinking) of wine, beer and liquor consumption.
     (Appendix B, Part 2).  This measure is particularly relevant because
     of the association between alcohol consumption and the perpetration
     of violence consistently reported throughout the literature.  The
     index selected for use in this research was the number of ounces of
     alcohol consumed per day.

     2) Alcohol dependence:

     An index was constructed to measure alcohol dependence (Sommer,
     Barnes & Murray, 1990).  The Alcohol Dependence Index was constructed
     by assigning "0" and "1" values to scores which respectively fell
     below and above the scale cutpoints (based on those reported in
     clinical research on alcohol dependence). The values were then summed
     to produce a possible scale range of 0 to 3 whereby a score of 0
     indicated no indicators of alcoholism while a score of 3 reflected
     being alcoholic on all indicators.  It employed the following
     measures:

          A) Michigan Alcoholism Screening Test (SMAST) (Porkorny &
          Miller, 1972) is a 13 item test intended to screen individuals
          in the general population and one which has been widely used in
          other studies (Appendix B, Part 3). The SMAST was designed to
          produce a more effective,shorter, self- administered and more
          easily scored version of the original MAST. In a study comparing
          the two versions of the MAST, reliability coefficient alphas
          computed for the two comparison group scores, yielded
          coefficients only slightly lower for the SMAST (Seltzer, Vinokur
          & Rooijen, 1975). However, validity coefficients on the other
          hand were found to be slightly higher for the same shortened
          scale version. In light of these findings, the authors concluded
          that the SMAST is as effective as the MAST in screening for
          alcoholism.

          B) The Alcohol Dependence Data Schedule (SADD) (Raistrick,
          Dunbar & Davidson, 1983) is a 15 item instrument which has been
          administered to both clinical and non-clinical samples and has
          been found to strongly distinguish alcoholics from nonalcoholics
          (Appendix B, Part 4). The split- half reliability obtained by
          Raistrick et al. (1983) using the short form (i.e., the 15 item
          scale used in this study) was .87. Other research by Jorge and
          Mazur (1985) employing this shortened version obtained a split
          half reliability of .88 when used in an interview format and .82
          when self- administered. Results of the Wave 1 provided an
          estimate of internal consistency of .82 for females and .68 for
          males.

          3) A subscale of the NIMH Diagnostic Interview Schedule Version
          III, Revised (DIS) which provides a number of indices of
          alcoholism that follow the diagnostic criteria of the Diagnostic
          and Statistical Manual (DSM-III) (American Psychiatric
          Association, 1980) was developed by Robins, Helzer, Croughan,
          Williams and Spitzer (1979). The subscale selected for use in
          this research is the "lifetime diagnosis of alcohol
          dependence/abuse" (Appendix B, Part 5).
 
 

     Personality.

     Several instruments measuring the major dimensions of personality
     were selected for use in this research. They are as follows:

     1) Eysenck Personality Questionnaire (EPQ-R) (1985):
     The EPQ-R is a 100 item scale containing three personality subscales:
     Neuroticism (EPQN), Introversion-Extraversion (EPQE), and
     Psychoticism (EPQP). The EPQ-R also contains a validity scale, the
     Lie Scale (EPQL) (Appendix B, Part 6). The Lie Scale may also be
     considered a measure of social conformity.  While Neuroticism and
     Introversion-Extraversion are well established tests of personality,
     the Psychoticism dimension is newer and has been a source of
     controversy.  Earlier versions of this scale have been criticized for
     low reliabilities (i.e., Cronbach's Alpha = .74 for males and .68 for
     females) (Torrubia & Muntaner, 1987) and skewed distributions
     (Eysenck & Eysenck, 1975).  The most recent version (EPQ- R)
     (Eysenck, Eysenck & Barrett, 1985) seems to have overcome these
     handicaps. Wave 1 data provided the following reliability coefficient
     alphas for the EPQ-R subscales: EPQN (males .85; females .85), EPQE
     (males .82; females .80), EPQP (males .60; females .61)  and EPQL
     (males .83; females .82) (Sommer, 1990; Sommer et al., 1992).

     2) MacAndrew Scale (MAC) (MacAndrew, 1965):
     The MAC is a 51 item subscale of the Minnesota Multiphasic
     Personality Inventory MMPI) (Appendix B, Part 7). It has been
     successfully cross validated against samples similar to those for
     which it was originally developed (alcoholics versus non-substance
     abusing psychiatric patients) (MacAndrew, 1980). MacAndrew (1980)
     found that 85% of male alcoholics assessed by the MAC scale were
     classified as "secondary psychopaths" (i.e., neurotic extroverts)
     according to Eysenck's model of personality.  Whereas results of Wave
     1 data indicated that the MAC was a predictor of partner abuse for
     female respondents (Sommer et al., 1992), tests of reliability
     performed on these data provided low estimates of internal
     consistency for both males (.43) (Sommer, 1990) and females (.54)
     (Sommer et al., 1992).  These were likely due to the multidimensional
     nature of the scale.  Earlywine, Fine and Martin (1990) considered
     the MAC a fallible indicator of underlying constructs. In spite of
     the MAC's low estimate of reliability, these authors determined that
     a confirmatory factor analysis demonstrated associations between it
     and personality measures.  Other research by MacAndrew (1980) showed
     that the MAC is still considered an appropriate measure because of
     its widespread use and its demonstrated ability to distinguish
     between alcoholic and nonalcoholic populations.

     3) Neuroticism Index:
     In order to avoid the problems associated with multicollinearity,
     several highly correlated scales were combined to form a Neuroticism
     Index (Sommer, Barnes & Murray, 1990).  A conservative approach
     toward selection of the measures based on the strength of
     intercorrelations supported this procedure (Sommer et al., 1992).
     Support for combining these measures was also found in the literature
     on personality (Krisha, 1980; Rosenberg, 1979; Roy, 1977).  The index
     was constructed by averaging the means of the scales' z scores.
     Transformations were performed to ensure that all composite measures
     were scored in the same direction.  The following instruments were
     included in this measure:

          A) The Rosenberg Self-Esteem Scale (Rosenberg, 1965) is a 10
          item scale which has been found to relate to alcoholism
          (Beckman, 1978) as well as with perpetrators of partner abuse
          (Goldstein & Rosenbaum, 1985; Rouse, 1988; Barnes et al., 1990)
          (Appendix B, Part 8). This scale provided satisfactory estimates
          of internal consistency on Wave 1 data for males (Alpha =.83)
          (Sommer, 1990) and females (Alpha =.86) (Sommer et al., 1992).

          B) The Trait Anxiety Scale (Spielberger, 1970), a widely used
          measure of anxiety consists of 20 items (Appendix B, Part 9).
          Since our earlier research on abuse among male college students
          found neuroticism to be a significant predictor of abuse
          (Barnes, Greenwood & Sommer, 1991), this particular measure was
          therefore thought to be relevant in this present examination of
          partner abuse. This measure provided a satisfactory estimate of
          internal consistency in Wave 1 for males (Alpha =.84) (Sommer,
          1990) and females (Alpha =.88) (Sommer et al., 1991).

          C) The Baron Ego-Strength Scale is another subscale of the MMPI
          containing 67 items (Appendix B, Part 10). It measures a general
          factor of the capacity for personalityintegration (Greene,
          1980).  According to Greene (1980), persons who score high on
          this measure are thought to have secure self concepts and are
          able to cope with situationalstress.  Low scorers on the other
          hand, are more likely to experience chronic, personality
          problems.  It was suggested that those with less integrated
          personalities might experience lower levels of impulse control
          and resort to more primitive methods of coping such as violence
          (Sommer, 1990). Analyses of Wave 1 data provided somewhat low,
          yet satisfactory estimates of internal consistency for males
          (.67) (Sommer, 1990) and females (.70) (Sommer et al., 1992).

          D) A fourth measure included in the Neuroticism Index is the
             EPQN described above.
 
 

     Family background.

     In order to address the issue of violence within the family of
     origin, two questions were included in the Wave 2 interview schedule
     (Appendix B, Part 11).
 

     Stress.

     Twelve items have been included in the Wave 2 interview schedule to
     assess the extent of life stress events experienced by abusers and
     their spouse/partners.  The items selected are similar to some of
     those included in the Holmes and Rahe (1967) Life Stress Events Scale
     (Appendix B, Part 12).
 

                       Dependent Variable

     Conflict Tactics Scale.

     In this project, the dependent variable, partner abuse was measured
     by an abridged version of Straus's (1979) Conflict Tactics Scale
     (CTS) (Form A).  The scale measures the frequency with which abusive
     actions occur as well as the degree of its severity.  An individual
     scoring low on the CTS is indicative of someone whose experience with
     partner abuse is both infrequent and less severe than one scoring
     high on the same measure.  In this research, both the prevalence
     (i.e., abuse ever occurring during the course of a relationship) and
     incidence (i.e., reports of partner abuse incidents during the past
     year) of perpetrated partner abuse were assessed by this measure.  In
     its original form, the CTS incorporates the following three modes of
     conflict resolution: (1) the use of rational discussion, argument and
     reasoning, (2) the use of verbal and nonverbal acts, and (3) the use
     of physical force against another person.

          For the purposes of this study, six items of the CTS measuring
     "physical force" tactics were selected for analyses.  Items
     associated with emotional abuse (i.e., rational discussion, argument
     and reasoning) were not considered relevant to this investigation of
     physical abuse and were therefore excluded.  As well, the most severe
     "physical force" conflict tactics (i.e., beat up, threatened with a
     knife or gun, used a knife or gun) were also omitted because of low
     endorsement rates demonstrated in past research (i.e., 0% to 4%)
     (Brinkerhoff & Lupri, 1988; Malone et al., 1989; Marshall & Rose,
     1990; Smith, 1987; Stets & Pirog-Good, 1989).

          Furthermore, whereas many studies have employed the full scale
     in their analyses, portions of the scale have also been used in other
     research (Brinkerhoff & Lupri, 1988; Roscoe & Benaske, 1985; Smith,
     1987). In fact, the reliability of an abbreviated version used in
     Wave 1 of this project (Cronbach's Alpha =.79 for males and
     Cronbach's Alpha =.94 for females) was found to show greater internal
     consistency than that of the entire scale used in our earlier
     research on male perpetrated courtship violence (Cronbach's Alpha
     =.57) (Barnes, et al., 1991).

          Various versions of the scale have been used in face-to-face
     interviews (Kennedy & Dutton, 1989; Schulman, 1981; Smith, 1987;
     Sommer, Barnes & Murray, 1990, 1991; Straus et al., 1980), telephone
     interviews (Straus & Gelles, 1986) and mail surveys (Straus, 1979).
     However as noted previously, the CTS is limited in its ability to
     elicit information on the following: (1) the circumstances
     surrounding the violent episode occurred (i.e., who initiated the
     episode or did it occur in self defence), (2) whether the consumption
     of alcohol was involved prior to the violent episode, (3) the
     consequences of violent episodes (i.e., physical injuries) and (4)
     whether episodes of abuse had been reported to the police.  The
     revised version of the CTS employed in this study included items
     designed to overcome most of the noted limitations of previous
     research (Appendix C).
 
 

                        Data Analysis

     The data analyses conducted in this second phase of the research
     followed a similar format to that conducted during its first phase
     using SAS statistical packages (SAS Institute Inc., 1986).  Data from
     both Waves 1 and 2 of the Winnipeg Health and Drinking Survey (Barnes
     & Murray, 1989) were used to test the theoretical issues in this
     research.

          In light of the diminished response rate reported earlier for
     Wave 2, data analysis began with an investigation into the effects of
     sample attrition. Inspecting the data for possible distortions
     resulting from missing data is important because nonrandom attrition
     may pose a threat to internal and external validity (Stacy, Newcomb &
     Bentler, 1991).  Determinations regarding the proportion of partner
     abusers from Wave 1 found among those who refused to participate or
     could not be contacted for Wave 2 data were made.  In addition,
     possible differences between study dropouts and completers with
     respect to Wave 1 sample characteristics (i.e., socio-demographic and
     personality factors, alcohol consumption and CTS scores) were
     evaluated.

          Assessing changes and stability in prevalence rates of partner
     abuse (i.e., has abuse "ever" occurred) as well as in other
     independent measures reported in Wave 1 and Wave 2 were achieved by
     way of frequency and correlational analyses.  As was the case in the
     analysis of Wave 1 data, Pearson correlations were also performed
     within the Wave 2 data to assess the bivariate relationships between
     partner abuse.

          The relationship between alcohol consumption (Wave 2) and
     partner abuse (Wave 2) were examined for linearity by collapsing the
     former variable into three levels (i.e., low, medium and high alcohol
     consumption) and then examining differences in the partner abuse
     score means.  Linearity is established when the means of the partner
     abuse scores are found to increase across levels of alcohol
     consumption.

          The final phase of the data analysis tested the major hypotheses
     of this research using logistic regression analyses.  Given that
     reported incidence rates for perpetrated partner abuse (10-14%) fall
     well below the 25 percent cutoff considered to be appropriate for
     inclusion as a dependent measure in a multiple regression analyses
     (Cleary & Angel, 1984), the selection of this approach was considered
     appropriate.  The application of this procedure in previous spousal
     violence research (Bland & Orn, 1986; Kalmuss & Seltzer, 1986;
     Seltzer & Kalmuss, 1988) provided additional support for use in this
     research.

          The logistic regression approach has two distinct features; the
     dependent variable is dichotomous, and the effects of the independent
     variables are presented as odds ratios.  In order to conform with the
     former, this study's dependent measure, "current perpetrated partner
     abuse" (i.e., the number of times an individual perpetrates partner
     abuse during the past year) was collapsed into the following two
     levels: abuse as reported at any level of severity and/or frequency
     (coded as 1) and no abuse (coded as 2).

          According to Halli and Rao (1992), the odds ratio forms the
     backbone of logistic regression.  In this research, logistic
     regression analyses made it possible to establish the ratio between
     perpetrating and not perpetrating current partner abuse based on a
     number of underlying and situational factors.  The magnitude of each
     predictor was also assessed through an examination of the parameter
     estimate in relation to the standard error.  Finally, the fit of the
     model being tested (i.e., the degree to which the predictors tested
     were needed to model the dependent measure) was determined by way of
     chi-square analysis.

          In spite of the appropriateness of logistic regression in this
     research, there are a number of limitations inherent in the use of
     this approach that are in need of consideration.  For example, a
     problem associated with the severe skewness of the dependent variable
     is the potential instability  of coefficient estimates using this
     statistical application (Seltzer & Kalmuss, 1988).  In order to
     compensate for this problem, the following strategies put forth by
     Seltzer and Kalmuss (1988) were adopted in the logistic regression
     analyses conducted in this research:

          1) Rather than including all the main effects and interaction
          effects in one model, the interactions were estimated in two
          separate models (i.e., one examining interactions with stress
          and the other examining interactions with current alcohol
          consumption).

          2) Conservative standards for the interpretation of results were
          also employed.  Thus, "statistical significance was attributed
          only when a coefficient was twice the value of the standard
          error" (p. 481, Seltzer & Kalmuss, 1988).

          The second issue relates to a loss of sensitivity in the
     dependent measure due to the transformations performed.  The
     quantitative differences once present in the continuous variable,
     "current perpetrated partner abuse" were eliminated when it was
     collapsed into two levels.  Thus, instead of assessing the
     perpetration of current partner abuse as ranging from low to high
     (with a possible range of 0 to 12), the transformed variable can only
     be interpreted as indicating either the presence or absence of
     current perpetrated partner abuse. To the extent that a dichotomous
     dependent measure is a necessary condition of logistic regression,
     its resulting lack of sensitivity must be viewed as an unresolved
     limitation in this approach, to be dealt with theoretically.

Next: Chapter 5



 

Table of Contents

Chapter 3

Chapter 5

References



HTML by Walter H. Schneider

 

___________
Updates:
2001 02 10 (format changes)


Walter H. Schneider
Box 62, Bruderheim, Alberta, Canada  T0B 0S0
    Tel: (780) 796-2306
e-mail: sheep_@telusplanet.net
Website: http://www.fathersforlife.org or http://forever.freeshell.org

Children need Both Parents -- Once a Parent, a Parent for Life
Subscribe to The Liberator, Men's International Unity Magazine,
Published by the Men's Defense Association

You are visitor #    since 2001 06 19
at the Web site of Fathers for Life

© Copyright Information