Child .4buse ~ Neglect, Vol. 16, lap. 673-691, 1992 Printed in the U.S.A. All rights reserved.

0145-2134/92 $5.00 + .00 Copyright © 1992 Pergamon Press Ltd.

PHYSICALLY ABUSIVE PARENTS AND THE 16-PF: A PRELIMINARY PSYCHOLOGICAL TYPOLOGY CAROLYN R.

FRANCIS

Pine Bluff, AR

HONORE M. HUGHES Department of Psychology, St. Louis University, St. Louis, MO

LETITIA HITZ Department of Psychology, University of Arkansas, Fayetteville, AR

AImtract--A typology of physically abusive parents was developed based upon personality characteristics measured by the 16-PF. Cluster analysis revealed five distinct patterns, accounting for 81 of 82 profiles submitted. Significant differences among the clusters were found on 14 of the 16 factors. The following types were described: (a) Shy, withdrawn, apprehensive, sober, and restrained; tending to have the least education, the greatest number of children. (b) Parents presenting as "normal" in personality features; tending to have relatively more education, fewer children. (c) Compulsive, bold, dominant, and assertive; tending to be highly manipulative in self-presentations, have high educational levels, and be older than other types. (d) Basically passive and submissive; tending to come from families where both parents are abusive. (e) Isolated, withdrawn, suspicious, tense, and apprehensive; tending to be more psychologically disturbed. Significance tests on external variables performed to validate the solution found differences among clusters in age, education, number of children, number of parents involved in abuse, 16-PF Faking Good and Faking Bad scores, and MMPI Scales L, F, K, 4, 5, 6, 7, and 0. Distinguishing personality features and demographic characteristics of the types are discussed with a focus upon possible treatment approaches for each type. Limitations of the study and suggestions for future research are addressed.

Key Words--Physicallyabusive parents, 16-PF, Psychological typology, Assessment of physically abusive parents, Personality characteristics of physically abusive parents.

SINCE THE PUBLICATION of the 1962 article by Kempe, Silverman, Steele, Broegemueller and Silver which brought the problem of child abuse to national attention, numerous attempts have been made to understand the reasons for its occurrence. In the intervening period of time, researchers have discovered that the search for etiological variables is a complicated one, and that the causes of child abuse are related to a complex interaction among a number of factors. Within the last 10 years, some agreement has emerged that the conceptual formulation for the etiology of physical child abuse needs to be a model that is integrated and interactive, taking into account a number of dimensions: (a) characteristics of the individual parent, (b) characteristics of the individual child, (c) family interaction patterns, (d) family Received for publication January 25, 1990; final revision received September 13, ! 991; accepted December 19, 1991. Requests for reprints may be sent to Honore M. Hughes, Ph.D., St. Louis University, Department of Psychology, 221 North Grand Boulevard, St. Louis, MO 63103. 673

~,74

(7. R. Francis. H. M. Hughes, and L. Hitz

climate (e.g., economic level, other stresses, situational factors) and (e) the social/community/ cultural context (e.g., Belsky, 1980; Geffner, Rosenbaum, & Hughes, 1988; Wolfe, 1985). However, given the fact that clinicians are faced with the responsibility of assessing and providing appropriate treatment for members of physically abusive families, it is not surprising that many researchers have focused upon the personality traits and other characteristics of abusive parents. While the sociological and cultural factors that contribute to the occurrence of child abuse are important, knowledge regarding the capabilities and needs of abusive parents is likely to be more salient for clinicians attempting to provide the most optimal forms of treatment. In general, the types of psychological descriptors that have been applied to abusive parents range from immature, depressed, and socially isolated to impulsive, hostile, and sadistic; they are also said to be tense and frustrated, or cold and compulsive (e.g., Wolfe, 1985). Taken as a whole, the existing body of research provides a list of psychological descriptors that is lengthy and, at times, inconsistent or even contradictory. The apparent discrepancies in research results seem to be at least partially due to methodological issues related to such matters as subject selection and choice of appropriate dependent measures (e.g., Geffner et al., 1988). However, some of the inconsistencies appear to be due to genuine differences among physically abusive parents. It has become clear that all abusive parents do not fit a particular personality type or possess a specific personality feature that gives rise to all types of physically abusive behavior. To better understand the various kinds of abusers, a number of researchers have taken a typology approach to attempting to describe the personality characteristics of abusive parents. This approach assumes that there is no single personality type that fits all abusive parents; rather there are people with various personality characteristics who exhibit abusive behavior toward children. Given the failure of researchers to identify any single personality type or feature that fits all abusive parents, and given the consistencies that appear across various published typologies, the typology approach would seem to be the most realistic in terms of accurately describing the personality functioning of abusive parents. A number of researchers have pursued the typology approach to the investigation of personality characteristics of physically abusive parents, and nearly three decades of research has produced a plethora of psychological descriptors. Overall conclusions from this literature have supported the idea that the characteristics identified likely do not apply to all abusers, but rather to several different types of abusers (Boisvert, 1972; Delsordo, 1963; Gil, 1970; Merrill, 1962; Sloan & Meier, 1983; Walters, 1975; Zalba, 1967). Of the numerous typologies for abusive parents that have been developed, none are based upon actual personality test data. Some of the typologies are focused upon environmental factors as being the primary precipitators of child abuse, rather than parental factors (e.g., Gil, 1970; Walters, 1975). Moreover, in those typologies that do include specifications of personality characteristics, those descriptions are generally based upon clinical impression (Boisvert, 1972; Delsordo, 1963; Kent, Weisberg, Lamar & Marx, 1983; Merrill, 1962; Zalba, 1967). While Sloan and Meier (1983) did administer a number of personality tests to their subjects, the results of these tests were simply used to further describe subjects who had already been classified. Waiters (1975) pointed out several difficulties that may arise with typologies constructed from clinical contact. These potential problems include the unequal assignment of weight to various factors due to unconscious biases on the part of the investigator, an often limited number of subjects, and a lack of randomness. Additionally, when a typology is developed from clinical contact, or from the subjective impressions of an investigator, rarely are there any clear guidelines for assignment of an individual to a specific category. In the present study, the Sixteen Personality Factor Questionnaire (16-PF) (Cattell, Eber, & Tatsuoka. 1982) has been used as a basis for the development of a typology for physically

Typologyof physicallyabusiveparents

675

abusive parents. The 16-PF was selected as a dependent measure because of its direct applicability to the types of personality characteristics being investigated and the likelihood that clinicians would use it in an evaluation. That is, the 16-PF is specifically designed to measure many of the personality characteristics that have been attributed to physically abusive parents in the literature (see Table 1). By using standardized measures of these personality characteristics, greater accuracy (i.e., less bias) has been obtained in the determination of the personality functioning of abusive parents. The typology generated by this investigation is a "psychological typology," in that categories have been delineated solely on the basis of those personality characteristics measured by the 16-PF. Thus, to further facilitate research into the personality characteristics of abusers and to maximize the match between abusers and appropriate treatment, the purposes of the present study were to: (a) identify, through standardized measurements, the personality characteristics commonly found among abusive parents; and (b) identify and describe any patterns that might exist among these personality characteristics, such that a psychological typology might be derived which characterized various types of physically abusive parents.

METHOD Subjects for this study were selected from among families that had been referred for a psychological assessment to the Center for Children in Crisis, Le Bonheur Children's Medical Center, Memphis, Tennessee, by caseworkers at Shelby County Department of Human Services on the basis of the following criteria: 1. Physical abuse was defined for the purposes of this study as an overt act of physical aggression directed toward a child. 2. Individuals were determined to be the actual perpetrators of the physical abuse. 3. Only substantiated cases of physical abuse, with no allegation or substantiation of sexual abuse, were considered for inclusion in the study. 4. The abuse occurred in a parent-child relationship. Natural parents, stepparents, "common law" parents, and legal guardians were included; extended family members, babysitters, or temporary caretakers were excluded. 5. Only individuals who completed a 16-PF as part of their evaluations were included. Families were generally referred to the Center for Children in Crisis (CCC) immediately following detection of abuse, with evaluations completed within 3-4 weeks of referral. Referrals are typically based upon a need for more complete information regarding a family's functioning, with recommendations for the most appropriate forms of intervention provided. The subject sample totalled 82 (45 men and 37 women) ranging in age from 19-49 years. Fifty of the subjects were white, and 32 were nonwhite (primarily African-American). Forty-five of the subjects were married, including seven couples where both parents were identified as child abusers; 26 were single, divorced or separated, and 11 were cohabitating. Materials 16-PF. 16-PF data on each subject were obtained via Form A of the 16-PF. The 16-PF (Cattell et al., 1982) is a factor analytically derived personality questionnaire designed to measure one's standing on 16 personality traits or factors. Briefly described, the personality scales included on the 16-PF are: reserved versus outgoing, concrete versus abstract, unstable versus stable, submissive versus dominant, sober versus enthusiastic, expedient versus conscientious, shy versus bold, realistic versus intuitive, trusting versus suspicious, practical versus

676

('. R. Francis. H. M. Hughes, and L. Hitz Table 1. Characteristics Attributed to Abusive Parents Which Relate to the 16PF

Factor A Reserved versus Outgoing "'Lacking in warmth" (Merrill, 1962) "'Interpersonal ditficulties" (Paulson, Affifi, Thomason, & Chalel; 1974) "~Socially isolated" (Cohn, 1982; Gaudin & Pollane, ~983; Starr. 1982) "Loneliness" (Milner & Wimberly, 1980) "Feelings of isolation and loneliness" (Spinetta, 1978) Factor C Unstable versus Stable "Failed to establish age-appropriate ego identities" (Steele & Pollack, 1968) "Tendency to get upset or angry" (Spinetta, 1978) "'Respond primarily to frustration" (Kertzman, 1978) "Less able to integrate emotion with reality" (Amberg, 1977) Factor E Submissive versus Dominant "Hostile and aggressive" (Egeland, Jacobvitz, & Sroufe, 1988; Merrill, 1962; Sloan & Meier, 1983; Susman, Trickett, Ianotti, Hollenbeck, & Zahn-Waxler, 1985) "Passive and dependent" (Egeland et al., 1988; Kent, Weisberg, Lamar, & Marx, 1983: Merrill, 1962; Sloan & Meier, 1983) "Conflicts with aggression" (Paulson et al., 1974) "'Anger and acting-out" (Friedich, Tyler, & Clark, 1985) Factor G Expedient versus Conscientious "'Rigid" (Kent et al., 1983; Merrill, 1962; Milner & Wimberly, 1980; SIoan & Meier, 1983) "Lacking in flexibility" (Merrill, 1962) "'Concerned with 'right vs. wrong' and "respect for authority' " (Kent et. al., 1983) Factor H Shy versus Bold "Passive and dependent" (Kent et al., 1983; Merrill, 1962; Sloan & Meier, 1983) "Fear of external threat and control" (Spinetta, 1978) "Shy, restrained, timid, threat-sensitive" (Kokkevi & Agathonos, 1987) Factor L Trusting versus Suspicious "Untrusting" (Kent et al., 1983) "Paranoid ideation" (Paulson et al., 1974) "Demanding, jealous and suspicious" (Oates, Forrest, & Peacock, 1985) Factor N Genuine versus Shrewd " "Sick but slick' syndrome" (Wright, 1976) Factor O Secure versus Apprehensive "Poor self-concept" (Anderson & Lauderdale, 1982; Kent et al., 1983; Milner & Wimberly, 1980; Rosen, 1978) "Low self-esteem" (Evans, 1980; Melnick &Hurley, 1969) "Feelings of depression and worthlessness" (Steele & Pollack, 1968) "'Distressed" (Milner & Wimberly, 1980) "'Experiencing identity/role crisis" (Sloan & Meier, 1983) Factor Q2 Joiner versus Loner "Socially isolated" (Cohn, 1982; Gaudin & Pollane, 1983; Starr, 1982) "'Feelings of isolation and loneliness" (Spinetta, 1978) "'Fear of external threat and control" (Spinetta, 1978)

(continued)

Typology of physically abusive parents

677

Table 1. (Continued) Factor Q3 Careless versus Controlled "Rigid" (Kent et al., 1983; Merrill, 1962; Sioan & Meier, 1983) "Compulsive" (Merrill, 1962; Paulson et al., 1974; Sloan & Meier, 1983) "Action-oriented rather than delay of gratification" (Steele & Pollack, 1968) "Undisciplined, self-conflicted, following own urges" (Kokkevi & Agathonos, 1987) "Liable to act without sufficient consideration and thought" (Oates et al., 1985) "Impulsivity, anger, acting-out" (Friedrich et al., 1985) Factor Q4 Relaxed versus Tense "Frustrated" (Kent et al., 1983; Merrill, 1962; Waiters, 1975) "Distressed" (Milner & Wilberly, 1980) "Experiencing stress" (Cohn, 1982; Egeland et al., 1980; Gaines, Sandgrund, Greene, & Power, 1978; Gaudin & Pollane, 1983; Justice & Justice, 1982; Perry, Wells, & Doran, 1983) "Tension" (Egeland et al., 1988)

imaginative, genuine versus shrewd, secure versus apprehensive, traditional versus liberal, joiner versus loner, careless versus controlled, and relaxed versus tense. Each of the 16 personality scales is comprised of 10 to 13 items. Test items are presented in the form of statements or questions that call for one of three possible responses from the individual. For example, "I talk about my feelings: (a) only if necessary, (b) in between, (c) readily, whenever I have a chance," or "In thinking of difficulties in my work, I: (a) try to plan ahead, before I meet them, (b) in between, (c) assume I can handle them when they come." Each of the 16 personality scales is scored separately, yielding a standardized score (sten score) of one to 10. This score indicates one's standing on a specific personality factor, relative to a normative sample.

Shipley-Institute of Living Scale.An estimate of each subject's level of intellectual functioning was obtained via the Shipley-Institute of Living Scale (SILS). The SILS is a two-part inventory, designed to measure intelligence (Bartz & Long, 1970). Part I is a vocabulary test, comprised of 40 items. Part II is an abstract-thinking test, comprised of 20 items. Each item on the SILS is scored as either correct or incorrect. Scores for each part are combined to give an estimate of one's level of intellectual functioning. Minnesota Multiphasic PersonalityInventory.The Minnesota Multiphasic Personality Inventory (MMPI) was administered to 54 of the 82 subjects included in this study. (Administration of the MMPI depended primarily upon the amount of time available to the subject and the examiner during the evaluation period.) The MMPI is a 566-item true/false questionnaire, designed to assess major personality characteristics that affect social and personal adjustment (Hathaway & McKinley, 1967). The test provides measurement on 10 clinical scales, which assess various aspects of an individual's personality functioning, and on three validity scales, which basically assess an individual's test-taking attitude (Dahlstrom, Welsh, & Dahlstrom, 1972, 1975; Graham, 1978).

RESULTS To identify and describe any patterns of personality characteristics, cluster analytic procedures were used. All analyses of the data were conducted using Statistical Package for the Social Sciences (SPSS-X) computer programs. First, a hierarchical agglomerative cluster analysis was performed on the 16-PF data (the 16 sten profile scores) obtained from the 82

678

('. R. Francis. H. M. Hughes. and L. Hitz Table 2. Fusion Coefficients for Cluster Solutions Number of Clusters

Fusion Coefficient

10 9 8 7 6 5 4 3 2 1

102.021561 106.137085 107.666580 114.466553 117.260376 120.744781 123.576462 128.958054 149.028778 227.789734

physically abusive parents. The purpose of this initial procedure was to determine the most viable number of clusters that seemed to exist within the data. The cluster analysis was carried out using a between groups average linkage method, with squared Euclidean distance as a similarity measure. An examination of the fusion coefficients for each cluster solution revealed a noticeable increase in coefficient values between the seven- and eight-cluster solutions (see Table 2). This suggested the eight-cluster solution to be the most viable for further examination (Aldenderfer & Blashfield, 1984). The eight-cluster solution generated by the hierarchical agglomerative cluster analysis placed the 82 subjects into eight clusters, based upon similarities among their 16-PF profiles. The number of subjects placed within each cluster were as follows: Cluster l, N = 47; Cluster 2, N = 4; Cluster 3, N = l; Cluster 4, N = 20; Cluster 5, N = l; Cluster 6, N = 5; Cluster 7, N = 3; Cluster 8, N = I. Examination of the eight clusters described above revealed the presence of three clusters containing only one subject. This seemed to suggest that there were actually five clusters within the data, which accounted for all but three of the 82 subjects. Therefore, further analyses were focused upon a five-cluster solution rather than an eight-duster solution. To refine the five-cluster solution, and to obtain more specific information on the nature of each cluster, the data were reanalyzed using the Quick Cluster procedure included in the SPSS-X computer program. The Quick Cluster analysis was performed using squared Euclidean distance as a similarity measure. Nearest centroid sorting was used as the cluster-forming procedure, with a five-cluster solution specified. Under this procedure, five distinct profile patterns were computer-selected to serve as initial cluster centers (centroids). The remaining profiles were then placed into one of the five clusters, based upon their levels of similarity to the cluster centers. The Quick Cluster analysis placed the 82 subjects into five clusters, with the following numbers of subjects in each cluster: Cluster 1, N = 1; Cluster 2, N = 19; Cluster 3, N = 23; Cluster 4, N = 22; and Cluster 5, N = 17. Thus, under the Quick Cluster procedure, two of the three subjects previously unaccounted for were placed into clusters. The one remaining subject, who was placed alone in Cluster 1, was determined to be an "outlier" within the data set. In cluster analysis, outliers are defined as subjects possessing such atypical features that they do not cluster with other subjects on the variables measured (Anderberg, 1973). Since the presence ofoutliers tends to distort the algorithms used in duster analysis, the Quick Cluster analysis was performed again with this subject omitted from the data set. The resultant cluster solution placed the 81 remaining subjects into five clusters containing the following number of subjects: Cluster 1, N = 19; Cluster 2, N = 18; Cluster 3, N = 10; Cluster 4, N = 28; Cluster 5, N = 6. To further ensure that this five-duster solution was indeed the most viable, additional Quick Cluster procedures were performed for three-, four-, six-, and seven-cluster solutions. Examination of the clusters generated by these procedures

679

Typologyof physicallyabusive parents Table 3. Cluster Comparisons on 16-PF Scores Mean Sten Scores

Factor

Cluster 1 (N= t9)

Cluster 2 (N= 18)

Cluster 3 (N= 10)

Cluster 4 (N= 28)

Cluster 5 (N= 6)

Fratio

Factor A Factor B Factor C Factor E Factor F Factor G Factor H Factor I Factor L Factor M Factor N Factor O Factor Q 1 Factor Q2 Factor Q3 Factor Q4

5.11 4.42 4.95 4.74 4.00* 5.00 3.84* 6.42 6.32 4.53 6.63 7.11' 4.69 6.95* 5.11 6.21

6.28 5.89 6.78* 6.00 6.11 5.67 6.22 5.72 6.00 5.00 5.72 5.17 6.22 5.44 6.06 5.17

6.30 5.80 6.10 7.10. 6.30 6.20 7.60* 6.40 3.90* 6.60* 5.30 3.50* 4.20* 4.90 7.00* 3.90*

4.75 5.00 6.25 4.00* 4.00* 6.04 5.46 5.57 5.11 4.14' 7.32* 5.04 4.57 5.79 7.39* 4.29*

2.83* 5.83 3.33 7.33* 5.33 5.50 4.50 6.50 7.50* 3.67* 4.50 9.00* 6.83* 7.83* 5.67 8.67'

7.41t 2.76* 7.78* 10.39t 7.78* 1.40 n.s. 11.00. 1.59 n.s. 6.48* 4.79** 5.15' 15.13' 4.18** 5.03* 7.61' 14.10.

Normative range = 4.5-6.5. *p < .05, **p < .01,*p < .001. *Outside normative range. Note.

suggested them to be less desirable than those generated by the five-cluster solution. Therefore, the five-cluster solution was retained for further analysis. Analyses of variance were performed comparing the five clusters on each o f the 16 personality factors measured by the 16-PF. The results o f these analyses, shown in Table 3, revealed between-group differences, which were significant beyond the .01 probability level on Factors A, C, E, F, H, L, M, N, O, Q 1, Q2, Q3, and Q4. Between-group differences on Factor B were significant at the .05 probability level. No significant between-group differences were found on factors G or I. Significant between-group differences on 14 o f the 16 factors suggest the five profile patterns to be rather distinct in terms o f the personality profiles that they reflect. Validation o f the five-cluster solution was performed via significance tests on 37 external variables, which were not utilized in generating the cluster solution. This validation procedure has been described as being one o f the better techniques available for validating cluster analysis (Alenderfer & Blashfield, 1984). One-way analyses of variance were performed for each o f the following continuous variables: age, n u m b e r o f years of education, monthly household income, n u m b e r of children in home, n u m b e r o f children abused, age o f child abused, severity of abuse, age at birth o f first child, level o f intelligence, 16-PF Faking Bad score, and the 13 clinical and validity scales on the MMPI. These analyses o f variance revealed significant differences among the five clusters on 13 o f these variables: age, n u m b e r o f years o f education, n u m b e r of children in the home, 16-PF Faking G o o d score, 16-PF Faking Bad score, and Scales L, F, K, 4, 5, 6, 7, and 0 on the MMPI. These significant results, which lend support to the validity o f the five-cluster solution, are presented in Table 4. T o identify the sources o f the significant differences found on the above-listed variables, planned comparisons using the least significant differences (LSD) procedure were performed within each o f the 5 clusters on the 13 variables identified. The results o f the LSD procedures are presented in Table 5. Chi-Square tests were performed for each o f the following categorical variables: gender, race, marital status, relationship to abused child, n u m b e r o f parents involved in abuse, disciplinary versus impulsive abuse, children removed or not removed from home, abuse charges filed or not filed, broken or intact h o m e as a child, past involvement with child abuse agency, past legal difficulties, spouse abuse, marital problems, and e m p l o y m e n t status. These Chi-

680

C. R. Francis, H. M. Hughes, and L. Hitz Table 4. Cluster Comparisons on External Variables ( A N O V A s ) Mean Scores or Values

Variable

Cluster I (N = 19)

Cluster 2 (N 18)

Cluster 3 (N = 10)

Cluster 4 (N = 28)

Cluster 5 (N - 6)

/;'ratio

Age YRSED NOCHHOME FG score a FB score a Scale L b'c Scale F Scale K Scale 4 Scale 5 Scale 6 Scale 7 Scale 0

31.5 11.2 2.9 6.4 4.0 48.9 56.1 50.3 67.3 49.1 58.3 53.3 55.4

29.6 13.3 1.7 7.4 1.7 54.7 53.3 58.7 67.2 48.7 59.5 53.0 49.6

37.3 14.0 1.8 9.3 t.2 59.7 48.4 64.7 61.2 58.6 53.9 52.7 44.7

28.9 12.0 2.6 8.5 1.6 59.1 52.8 57.3 57.2 53.3 54.6 50.9 54.0

27.5 12.0 2.0 2.7 5.0 40.0 65.0 40.0 70.7 61.7 75.0 67.3 57.7

3.53** 5.09* 2.50* 8.74* 11.81' 5.77* 3.49** 6.34* 3.59** 3.09* 5.85* 3.43** 2.95*

Note. All nonsignificant analyses of variance were omitted. Y R S E D = n u m b e r o f years o f education; N O C H H O M E - n u m b e r of children in the home; FG score = 16-PF Faking Good score; = [FB] Faking Bad score. a 16-PF Faking Good and Faking Bad scores are given as raw scores, bFor Minnesota Multiphasic Personality Inventory (MMPI) scales, df= (4, 53). For other variables, df= (4, 76). CMMPI scale scores are given as T scores. * p = 5 < 1> 5 < 5> 5 < 4 < 2 < 5> 5> 3
1,4 4>2 12,3,4 13 1

Physically abusive parents and the 16-PF: a preliminary psychological typology.

A typology of physically abusive parents was developed based upon personality characteristics measured by the 16-PF. Cluster analysis revealed five di...
1MB Sizes 0 Downloads 0 Views