REVIEW OF FORENSIC ASSESSMENT INSTRUMENTS

In the first edition of this book, nine instruments that had been designed to assess various parenting attitudes, behaviors, abilities, or styles were reviewed. Since that time, an extensive number of new par­enting measures have appeared in the literature, some of them specifically intended for use in forensic assessments of parents' capacities and cus­tody cases.

The sheer volume of parenting measures that have been developed does not allow their review here. Therefore, parenting instruments that are not reviewed in this chapter are listed in Table 2. Some of them are primarily research scales, while others are commercially marketed and intended for applied use in clinical or forensic settings. What is known about the psychometric properties of these scales varies widely, as some have been heavily researched whereas others have received scant atten­tion. Moreover, there appears to be little relationship between an instru­ment's empirical support and availability for use. For example, some of the instruments that are available commercially have been subjected to minimal research and, in some cases, are supported by no peer-reviewed publications in scientific journals detailing their characteristics.

Instruments reviewed in this chapter were selected on the basis of two general guidelines. First, some of the instruments that were included in the first edition that have continued to generate a significant amount of research have been retained in order to provide an update regarding their status. Second, we included some measures developed since the first edi­tion because they are used frequently by forensic practitioners. Exclusion of an instrument from this review should not be considered as an indication that a measure has been judged to be flawed or inappropriate for applied use, and inclusion in this review should not be interpreted as a recommendation that a particular instrument should be used.

Ackerman-Schoendorf Scales for Parent Evaluation of Custody (ASPECT)

Authors

Ackerman, M. T., & Shoendorf, K.

Author Affiliation

Wisconsin School of Professional Psychology

Table 2. Measures of Parenting Attitudes, Skills, and Parent-Child Interaction

PARENTING CAPACITY 259

Table 2. Continued

260 CHAPTER 7

PARENTING CAPACITY 261

Primary Reference

Ackerman, M.

Description

The Ackerman-Schoendorf Scales for Parent Evaluation of Custody (ASPECT) is "a clinical tool designed to aid mental health professionals in making child custody recommendations" (a, p. 1). The ASPECT is a com­posite measure that aggregates data from several sources (e.g., interviews and various psychological tests with parent and child, and behavioral observation). In terms of general psychological testing, parents are administered the MMPI-2, Rorschach, and an intelligence test. Children are administered the Rorschach, intelligence and achievement tests, and projective story-telling measures. Parents also complete a Parenting Questionnaire comprised of 57 open-ended questions regarding various issues related to parenting and child custody.

Information from these sources is combined into a "Parental Custody Index" (PCI), which is intended to provide a summary index of parenting effectiveness. The PCI comprises 56 yes/no items and can be broken down into three subscales assessing parents' self-presentation (Observational), the appropriateness of the social environment provided (Social), and the extent of parents' cognitive and emotional capacity to provide for their children (Cognitive-Emotional). Raw scores on these scales are converted to T-scores, which are based on data from the normative sample. The ASPECT also contains 12 "critical items" used to identify areas in which there may be more serious deficits in parenting than are summarized in the PCI or individual subscale scores. (For other reviews of the ASPECT, see references e, f, and h.)

Conceptual Basis

concept definition. Chapter 5 of the ASPECT manual describes the derivation of the concepts thought to be germane to effective parenting and custody decision-making, which are described in the introduction as "issues deemed essential by researchers and other mental health profes­sionals in determining appropriate custody recommendations" (a, p. 1). It appears that the authors relied on diverse sources for delineating the domains that they believed were important to assess as part of the PCI, as the literature review includes references to empirical research on parent­ing, surveys of judges and practitioners, and theoretical reviews of custody standards. Relatively little mention is made of relevant legal stan­dards and definitions, other than that the authors note that these are too vague to offer much specific guidance regarding how to operationalize the key constructs most relevant to custody decision-making.

Item-by-item rationales for each of the 56 PCI items are provided, with occasional citations to others who have argued for the importance of these content areas in relation to parenting ability. For example, support for inclusion of the Observational scale item, "Was the parent appropri­ately attired?" and the Social scale item, "Does the parent provide appro­priate discipline?" is based on reference to other commentators who have asserted that these represent important domains of parenting compe­tence. Similarly, support for the content domain associated conceptually with the Cognitive-Emotional scale item "Were there indicators of diffi­culty with emotional control on the Rorschach?" is based on reference to psychological theory regarding parenting competence and emotional reg­ulation, as well as on survey findings that the Rorschach is used widely in custody evaluations. Inclusion of some component areas, however, appears to reflect the choices of the authors independent of citations to external support. For example, inclusion of the Social scale item, "Is the child placed next to the parent on the Draw-A-Family Test?" apparently is based on the premise that children's "indirect" parental preferences are relevant to parental competence—and that this component can be assessed via projective methods (see below).

Although the ASPECT appears primarily to be a rationally-derived instrument, the manual notes that an initial collection of items presumed to measure the constructs of interest for the three subscales were examined to determine their internal consistency. Items that did not "maximize the item-to-total correlations" (a, p. 48) apparently were dropped. No other information is provided about these constructs or analyses, however.

operational definition. The attribute dimensions defined by the ASPECT developers as relevant to parenting capacity are operationalized by using the procedures noted earlier (e.g., standardized tests, unstruc­tured interviews) to reach conclusions about each facet. At the most basic level, each of the 56 items is dichotomized by the examiner as either being present or absent, and these dichotomous decisions are then combined to form a global index of parenting capacity. As can be inferred from the examples above, some of these operationalizations appear to be reason­ably consistent with legal constructs regarding custody (e.g., "Does the parent provide appropriate discipline?"), whereas others are further removed from such dimensions (e.g., "Is the parent currently taking any psychiatric medications?" and "Is the Comprehension subtest scale score on the WAIS-R at least 9?").

critique. Other than the authors' literature review, procedures often used to establish the content validity of psychological tests were not employed in the development of the ASPECT. The conceptual bases of the PCI and its subscales primarily were rationally derived, and their connec­tion to relevant functional abilities regarding parenting is unclear in many cases (e, f). Many of the domains presumed by the authors to be important to parenting have weak, indirect, or unknown connections to parenting competence, both in terms of legal standards as well as psychological the­ory. Equally important, other content areas that may be critical for ascer­taining the best placement for the child (e.g., relevant social supports) are not incorporated into the ASPECT. The statistical analyses that were used to eliminate certain domains are not described adequately in the manual.

Operationalization of the content domains that the PCI and its sub­scales claim to measure is problematic in some cases, relying on proce­dures or instruments of unknown or questionable reliability and validity (d, i). Moreover, each content area in the ASPECT is equally weighted, even though there is no conceptual reason to assume that all of these fac­tors are equivalent in terms of parenting skill or the custody decision­making process. Similarly, the "averaging" of information across multiple children (if there is more than one whose custody is in dispute) appears to be driven more by convenience than any reliance on legal standards or psychological theory.

Scoring of many items is influenced by observation of unstructured parent-child interactions of relatively limited duration ("Observations can be brief; 15 to 20 minutes should suffice": a, p. 6). The scoring of other items (e.g., "Did the parent provide oral hygiene training?") is described as "self-explanatory" (a, p. 14), although it is unclear what to do if both parents claim to have taught this skill to the child(ren). Conversely, the cognitive-emotional abilities of the parent are operationalized mostly in very concrete terms (e.g., "Was the L scale on the MMPI below 60T(65T for the MMPI-2)?"), yet there is little offered to suggest that these specific factors actually tap the dimensions of parenting relevant to making a cus­tody decision.

In short, a considerable degree of inference would be needed to con­clude that the content of the 56 item PCI has captured the requisite dimen­sions of parenting capacity germane to custody decision-making, or that it has operationalized these content domains in a manner that provides a great deal of confidence that these dimensions actually are being assessed in a reliable and valid manner. Moreover, it is difficult to accept the logic of the instrument's presumption that the multifaceted constructs relevant to determining parental competence can be condensed and reduced to a single quantitative indicator.

Psychometric Development

standardization. Procedures used to score the PCI itself appear to be relatively straightforward. However, administration and scoring pro­cedures for the ASPECT are dependent upon the measures and proce­dures that contribute to the PCI. Given recent concerns about the standardization of the Exner Comprehensive System for the Rorschach and its tendency to overpathologize (see h), there is reason for concern about its inclusion from the point of view of standardization. Even more problematic is the use of unstructured projective methods, in light of con­cerns about their lack of psychometric rigor (d). Finally, both clinical interviews and parent-child observation are designed to be unstructured, although some basic content areas to be covered in the interview are suggested.

reliability. The authors report analyses for protocols of 88 exami­nees that were coded by two raters, but the specific statistic used (e.g., pearson correlation, intra-class correlation) is not indicated. More impor­tantly, the procedure used to ascertain reliability consisted of two raters who "independently reviewed interview notes and tests from examiners using the ASPECT" (a, p. 51). Such a procedure is somewhat limited since it simply indicates that the items comprising the PCI can be coded reliably from summary information collected by individual examiners.

Not addressed by this statistic is how reliably underlying data and observations are collected and coded, which is a reasonable concern in light of the unstructured nature of the interviews and behavioral observa­tions. Thus, the authors do not provide information regarding the reliabil­ity of the components themselves (e.g., Form level on the Rorschach; evaluations of the appropriateness of the attire worn by the parent), but merely address whether or not this information can be coded reliably into the PCI. In particular, it is unknown whether the items comprising the Observational scale would be coded similarly if raters had watched videotapes of these interviews, or conducted separate interviews them­selves, rather than read someone else's interview notes. For example, two raters could agree that an evaluator indicated in his or her notes that an examinee was "cooperative," but this does not necessarily support the conclusion that they would concur if they actually had observed or partic­ipated in these interactions.

Temporal stability of the ASPECT is unknown. No test-retest analy­ses are reported.

norms. The standardization sample for the ASPECT consisted of 200 families involved in contested custody cases. The demographic com­position of the sample is primarily Caucasian (96.9%) and upper middle class. Thus the ASPECT normative sample is not representative of divorc­ing families in general, although it may not be as discrepant from the pop­ulation of divorcing families who contest custody and undergo evaluations by mental health professionals. It is also noteworthy that the disposition of the cases in the ASPECT normative sample appears to be unusual, in that fathers obtained custody in a disproportionate number of cases.

Critique. There are significant problems with the psychometric properties of the ASPECT. Several of the procedures used are unstandard­ized, reliability information is lacking, and the normative group is not representative of the majority of custody litigants or the typical outcome of these types of cases.

Construct Validation

The internal consistency of the ASPECT is unclear. The manual (a) notes on page 51 that the alphas for the Observational and Cognitive- Emotional subscales were reported as.00, yet in Table 10 (a, p. 52) they are reported as.50. Internal consistency for the Social subscale is reported as.72, whereas the PCI alpha is.76. Given the diversity of content areas tapped by these items, it is not surprising that internal consistency would be somewhat low. Interestingly, item-to-total correlations for sev­eral of the individual items are negative, although these items are not reverse-scored.

Evidence of construct validity for the ASPECT is based solely on the content validity issues noted above and the predictive utility research noted in the manual (see below). Despite first becoming commercially available in 1992, as of this date no peer-reviewed research has been pub­lished on the ASPECT.

critique. Clearly, further research is needed before any strong con­clusions can be offered regarding the construct validity of the ASPECT (e, f). Early reviews (c, for example) identified this as an important next step in the development of the scale, and the authors themselves noted the need for such research in the manual (p. 54), but none has been forthcoming.

Many of the components of the PCI may in fact have some indirect connection to parenting capacity (e.g., psychopathology as identified by select MMPI-2 scores, which may limit a parent's ability to parent in some cases). But the burden is on the developers to demonstrate this connec­tion, particularly given the very concrete manner in which several of these indices are employed (e.g., MMPI-2 MacAndrews Alcoholism scale > 65/'). Moreover, the bifurcation and summation of these diverse

dimensional measures (even if reliable and valid themselves) into a single numerical indicator needs to be justified by data suggesting that such methods are actually measuring some homogeneous construct of "parent­ing." The relatively low alpha level (.76 for a 56 item scale) reported in the instrument manual indicates that such an approach may not be warranted. Factor analyses of the ASPECT might provide a better concep­tualization of the instrument's underlying dimensions than the rationally- derived subscales currently used (b).

Predictive or Classification Utility

The ASPECT manual describes analyses on 57 couples from the stan­dardization sample, the results of which were compared to judges' cus­tody determinations. For those 25 cases in which there was less than one standard deviation (i.e., 10 point) difference between the parents' PCI scores, the overall hit rate was.60. In those 32 cases in which there was more than a 10-point difference, the ASPECT results were consistent with judges' decisions in 91 % of the cases.

critique. Although somewhat informative, inferences from these results are limited by several factors. First, the sample is described inade­quately, as is the performance of the participants on the PCI. No other information is presented regarding the range of scores these examinees produced other than that there was at least a "10-point difference" between parent couples in some cases. In addition, representation of minority group members in the normative sample is limited or non­existent, and a remarkably high number of fathers were awarded custody, both of which call into question the representativeness of the sample.

Perhaps most importantly, the criterion being predicted—judges' decisions—is an unknown proxy for what constitutes children's best interests. Moreover, although the PCI results apparently were not pre­sented to the judges, it is unclear exactly what (if any) mental health infor­mation they were provided about the parents and whether or not this had any impact on their custody determinations.

Aside from these issues, other reviewers have expressed concerns that construction of a score such as the PCI encourages psychologists to overstep their roles in child custody cases. Melton (f) has argued that the numeric identification of the "better" custodial parent places the exam­iner in the position of attempting to address the ultimate issue before the trier-of-fact, rather than remain in the more appropriate role of the mental health expert who simply provides relevant information to the courts. (See Chapter 1 for a discussion of views for and against addressing the ultimate legal question.)

Potential for Expressing Parent-Child Congruency

In theory, the ASPECT should provide useful information regarding the fit between the parent's abilities and the child's needs, because of the incorporation of information related to their overall functioning and the type and quality of interactions between them. In practice, however, the lack of information about the validity of the instrument leaves unan­swered whether the methods used and the ways in which the information has been combined produce PCI scores that actually reflect the true degree of congruence between the parent and the child(ren).

References

(a) Ackerman, M., & Schoendorf, K. (1992). ASPECT: Ackerman-Schoendorf Scalesfor Parent Evaluation of Custody-Manual. Los Angeles, CA: Western Psychological Services.

(b) Arditti, J. A. (1995). Ackerman-Schoendorf Scales for Parent Evaluation of Custody. In J. C. Conoley & J. C. Impara (Eds.), The twelfth mental measurements yearbook (pp. 20-22). Lincoln, NE: Buros Institute of Mental Measurements.

(c) Brodzinsky, D. (1993). On the use and misuse of psychological testing in child custody evaluations. Professional Psychology: Research and Practice, 24, 213-219.

(d) Garb, H., Wood, J., & Nezworski, M. (2000). Projective techniques and the detection of child sexual abuse. Child Abuse and Neglect, 24, 437-438.

(e) Heinze, M., & Grisso, T. (1996). Review of instruments assessing parenting competen­cies used in child custody evaluations. Behavioral Sciences and the Law, 14, 293-313.

(f) Melton, G. B. (1995). Ackerman-Schoendorf Scales for Parent Evaluation of Custody. In J. C. Conoley & J. C. Impara (Eds.), The twelfth mental measurements yearbook (pp. 22-23). Lincoln, NE: Buros Institute of Mental Measurements.

(g) Otto, R. K., Edens, J. F., & Barcus, E. (2000). The use of psychological testing in child cus­tody evaluations. Family and Conciliation Courts Review, 38, 312-340.

(h) Wellman, M. (1994). Ackerman-Schoendorf Scales for Parent Evaluation of Custody. In D. Keyser & R. Sweetwater (Eds.), Test critiques, Volume X (pp. 13-19). Austin, TX: PRO- ED.

(i) Wood, J. M., Nezworski, M. T., Garb, H. N., & Lilienfeld, S. O. (2001). The misperception of psychopathology: Problems with norms of the Comprehensive System for the Rorschach. Clinical Psychology: Science and Practice, 8, 350-373.

Bricklin Perceptual Scales (BPS)

Author

Bricklin, B.

Author Affiliation

Widener university

Primary Reference

Bricklin, B. (1990). Bricklin Perceptual Scales manual. Furlong, PA: Village Publishing.

Description

The Bricklin Perceptual Scales (BPS) (b) were specifically designed for use in child custody evaluation contexts. The developer describes the BPS as a projective measure of parents' competence, supportiveness, follow-up consistency, and possession of admirable traits (b). Scores for four scales derived from the 64 items (the same 32 for each parent) are obtained by having a child punch holes through a card with a stylus. On the cards are lines with the two extremes marked as "very well" to "not so well." The "parent of choice" is the one who receives the most positive ratings overall across the 32 responses. Thus, conclusions about parental capacity or com­petence are drawn based on the child's representation of his or her interac­tions with the parent. Bricklin claims that the information derived from the BPS is a reflection of the child's "unconscious preferences" and that this preference is more important than "objective" measures of parental behav­ior. Moreover, this preference is said to be of the utmost importance to cus­tody decision-making because "parental behavior can only be evaluated in terms of its meaning and utility to a particular child" (b, p. 14).

Conceptual Basis

concept definition. As noted above, Bricklin (b) indicates that the BPS is designed to measure children's parental preferences across four domains: competence, supportiveness, followup consistency, and posses­sion of admirable traits. Inclusion of these dimensions appears to reflect the author's belief that "conscious processes" are more vulnerable to distortion precipitated by parents motivated to have their child present a particular picture to the examiner. There are, however, no citations to other authorities or specific psychological theories on this point, other than generic refer­ences to how "unconscious processes" of the child are more accurate meas­ures of children's best interests than are "conscious" indicators.

operational definition. Because the developer believed that more objective measures of parenting capacity (e.g., children's oral reports on the BPS) were of limited value in custody cases, projective methods are used to operationalize the four domains of parenting capacity noted above. The "unconscious preferences" of children are quantified by the extent to which the parents are judged to differ based on the hole punches on the 32 items covered in the BPS. The content of most of these questions is relatively straightforward (e.g., Card 20: "If this is Mom doing very well at helping you to calm down, and this is Mom doing not so well at helping you to calm down, where on this line would Mom be?;" Card 33: "If this is Dad doing very well at solving a fight over toys, and this is Dad doing not so well at solving a fight over toys, where on this line would Dad be?"). From the comparison of the line punches across the Mother and Father cards, it can then be inferred that the optimal parent is being identified.

critique. The conceptual underpinning of the BPS is not clearly described. As noted above, although there are vague references to the importance of children's "unconscious" preferences, the notion is not linked to psycholegal constructs in ways to support the significance of the content areas purportedly assessed by the BPS (f, j). Moreover, it is unclear why the author considered overt markings on a card with clearly defined parameters (very well to not so well) to be indicative of unconscious processes and less transparent than verbal responses, and no data are offered to support such a claim (e.g., data examining correlations between children's description of their parents via card marks and verbal reports).

Contrary to Bricklin's claims, the relatively narrow range of the con­tent domains covered by the BPS severely limits the extent to which it could be considered a comprehensive measure of parenting capacity or the fit between the parent's actual abilities and the child's needs, regard­less of the adequacy of the projective nature of the data collected regard­ing the child's "true" perceptions of the parent. Despite these concerns, it should be noted that the overt content of the BPS does tap some issues important to understanding the general nature and strength of the parent-child relationship, providing it with some measure of face validity.

Finally, as noted above and consistent with concerns expressed by Melton (g) regarding the ASPECT, the structuring of BPS results such that they claim to identify the "better" parent may encourage examiners to offer opinions about matters that are ultimately legal determinations, thus over­stepping their roles in relation to custody evaluations. This is particularly troubling in relation to the BPS, in that it purports to accomplish this task based solely on input from the child.

Psychometric Development

standardization. Some general guidelines for administration and scoring are described in the manual, but the manual encourages examin­ers to deviate from these procedures (see, e.g., p. 22 and p. 58 in reference b). Thus, the BPS is only loosely standardized in its administration format. Scoring is relatively straightforward, in that each card has a grid on the back that provides a numeric indicator identifying where the hole has been placed by the child.

reliability. Adequate reliability data for the BPS are lacking. Although no information regarding inter-rater reliability is provided, inter-rater reliability may not prove problematic given the objective scor­ing of the instrument (assuming it is administered in a reliable manner and deviations from recommended procedures do not occur). Test-retest data (n = 12) are described in a format that is difficult to comprehend, but they offer some indication of stability over relatively short time frames (i.e., "parents of choice" apparently remained the same over a time frame that was seven months or less). The description of the studies providing this information is insufficient to judge their methodological rigor. Unpublished dissertations that have examined the test-retest reliability of the BPS (d, i, k) have not been so encouraging.

norms. There is no clearly delineated normative sample for the BPS. Although the author makes various references to a "data-based system," mostly this appears to consist of anecdotal accounts of other clinicians who have used the instrument. These accounts apparently have not led to accumulation of cases to provide a normative sample.

critique. Previous reviewers (e, f, j, l) consistently have noted the problems described above (subjective administration, unknown reliabil­ity, absence of norms), which have yet to be addressed in any meaningful fashion. The logic for the instrument as described by its developer sug­gests that these issues may never be addressed by the instrument's devel­oper; the manual indicates that issues such as test-retest reliability are of limited concern and that there are "no reasons to expect the measures reported here to exhibit any particular degree of stability" (b, p. 42).

Construct Validation

No information has been provided regarding the internal consistency of the BPS.

Construct validation data reported in the BPS manual focuses on the instrument's relationship with another measure designed by the same author, the Perception of Relationships Test (PORT; described later in this chapter), and from the instrument's association with other examiners' perceptions of which of the two parents would be the better caretaker.

Regarding the PORT, Bricklin (b) cites two studies in which the BPS and PORT resulted in the same "parent of choice" in a total of 46 of the 55 cases (84%). As further evidence of the validity of the BPS, Bricklin (a) administered the BPS and a set of questionnaires to 23 children and their parents. The questionnaires queried the children and parents regarding the parents' child rearing behaviors and responsibilities. Parents who were identified by the children as taking on more child rearing responsi­bilities were identified as the "parent of choice." In 21 of 23 cases (91%), the same "parent of choice" was identified by the children's responses to the questionnaires and the BPS.

Interestingly, Bricklin claims that this level of agreement provides even greater support for the validity of the BPS than a higher rate of agreement might have demonstrated. He argues that children's responses to the BPS are non-verbal and represent less conscious feelings about their parents, whereas their responses to questionnaires represent more con­scious perceptions. Thus, perfect agreement between the two measures would not be expected and indeed would prove problematic, Bricklin argues. With the same sample, Bricklin also examined the rates of agree­ment between the "parent of choice" as identified by the children's BPS responses and the parents' perceptions of their involvement with their children. In 13 of 17 (76%) cases there was agreement between parents' perceptions and the "parent of choice" identified by the BPS.

A number of other studies examining the validity of the BPS are described in a photocopied test manual supplement (c), but the descriptions are so limited (one to two lines per study) that it is not possible to ade­quately critique them. Of particular interest is the claim that in a sample of 1,765 cases, the "parents of choice" identified by the children's BPS responses was consistent with the judgments of psychologists who had access to "clinical and life history data." These findings appear essentially to be psychologists reporting to Bricklin how frequently the BPS results were consistent with their clinical impressions in custody cases they evaluated.

One study (d) examined the association between the BPS and a widely used and extensively researched measure of parent-child function­ing, the family Environment Scale (FES; h). Fifty-eight children between the ages of 6 and 12, and their parents, were administered the BPS, Children's Version of the Family Environment Scales (CVFES), and FES at two sessions, one week apart. There was very little association between these measures, with poor agreement between the child's "parent of choice" as identified by the BPS and the parents' consensus agreement of who was the "parent of choice."

critique. Although the BPS' status as a projective measure for assessing unconscious process creates some methodological obstacles, it would not be difficult to conduct factor analyses on existing BPS data. One might also consider a multi-trait, multi-method research project to investigate its construct validity, but this is made difficult by the lack of a clear description concerning what construct the BPS claims to measure.

There have been no BPS studies published in peer reviewed outlets, and the research data provided in the manual and its supplements are so abbreviated that it is not possible to judge the adequacy of the samples, methodology or the results reported. Studies with large sample sizes are minimally described, and may simply reflect the extent to which examin­ers who have bought and used the BPS form clinical opinions consistent with its conclusions (and which may have been shaped by the BPS results themselves). Such information does little to support the argument that results of the BPS actually can identify the more appropriate custodial parent.

Predictive or Classification Utility

Bricklin (a) described 29 cases of contested custody in which he con­ducted evaluations using the BPS. In 27 of these cases (93%) the parent identified as the "parent of choice" by the BPS was awarded custody by the judge. It is not reported whether Bricklin offered an ultimate opinion on custody in these cases that was consistent with BPS scores, which obvi­ously would raise concerns about criterion contamination.

Critique. The same criticisms noted above regarding the relative absence of construct validation data also apply here. The predictive utility of the BPS is essentially unknown.

Potential for Expressing Parent-Child Congruency

The identification of the "parent of choice" assumes that the BPS pro­vides an index of the degree of fit between the parents' capacities and the child's needs. As such, in principle the BPS should provide a good indica­tion of the degree of congruence between each parent and the child. Unfortunately, the conceptual, psychometric, and validation limitations of the BPS preclude such a conclusion from being drawn. Even assuming that the BPS measures children's perceptions in a reliable and valid manner, the presumption that these perceptions of the parent are de facto indicators of the true level of fit between them is an unsubstantiated claim in need of empirical support before it should be used to influence custody decisions.

References

(a) Bricklin, B. (1990). Parent Awareness Skills Survey manual. Furlong, PA: Village Publishing.

(b) Bricklin, B. (1990). Bricklin Perceptual Scales manual. Furlong, PA: Village Publishing.

(c) Bricklin, B., & Elliott, G. (1997). Critical child custody evaluation issues: Questions and answers. Test manuals supplement for BPS, PORT, PASS, PPCP. Furlong, PA: Village Publishing.

(d) Gilch-Pesantez, J. (2001). Test-retest reliabilityand construct validity: The BricklinPerceptual Scales. Unpublished doctoral dissertation. Indiana University, Pennsylvania.

(e) Hagin, R. A, (1992). Bricklin Perceptual Scales. In J. L. Conoley & J. C. Impara (Eds.), The twelfth mental measurementsyearbook (pp. 117-118). Lincoln, NE:Buros Institute ofMental Measurements.

(f) Heinze, M. C., & Grisso, T. (1996). Review of instruments assessing parenting compe­tencies used in child custody evaluations. Behavioral Sciences and the Law, 14, 293-313.

(g) Melton, G. B. (1995). Ackerman-Schoendorf Scales for Parent Evaluation of Custody. In J. C. Conoley & J. C. Impara (Eds.). The twelfth mental measurements yearbook (pp. 22-23). Lincoln, NE: Buros Institute ofMental Measurements.

(h) Moos, R., & Moos, B. (1994). family Environment Scale: Manual (3rd ed.). Palo Alto, CA: Consulting Psychologists Press.

(i) Muir, K. (1997). Children's perceptions of their parents and susceptibility to suggestion. Unpublished doctoral dissertation. California School of Professional Psychology, San Diego.

(j) Otto, R. K., Edens, J. F., & Barcus, E. (2000). The use of psychological testing in child cus­todyevaluations. Familyand Conciliation Courts Review, 38, 312-340.

(k) Scallon, C. (1997). Changes in children's perceptions of their 'parent of choice' following partic­ipation in a Roman Catholic post-divorce support group. Unpublished doctoral dissertation. Chicago School of Professional Psychology, Chicago.

(l) Shaffer, M. B. (1992). Bricklin Perceptual Scales. InJ. L. Conoley & J. C. Impara (Eds.), The twelfth mental measurements yearbook (pp. 118-119). Lincoln, NE: Buros Institute of Mental Measurements.

Perception of Relationships Test (PORT)

Author

Bricklin,B.

Author Affiliation

Widener University

Primary Reference

Bricklin, B. (1989). Perception of Relationships Test manual. Furlong, PA: Village Publishing.

Description

The Perception of Relationships Test (PORT) (a) was designed specifically for use in child custody evaluation contexts and is described as a projective test that is based on the placement of the child's human figure drawings.

It was designed to assess the "psychological closeness" of the child to each parent, as well as the types of interactions that occur with each parent. The PORT is comprised of seven drawing tasks and is scored based on the con­figuration of the child, parents, and family within the child's drawings. For example, distances are measured between the child's drawing of the child and the parents in some items. Scores result in the identification of a "pri­mary caretaking parent." Thus, the PORT is not a measure of parenting capacity or ability, but rather is offered as an objective, structured measure of the connection between the parent and child, based on the child's repre­sentations. (For a complete review of the PORT, see d).

Conceptual Basis

concept definition. Bricklin (b, p. 1) writes that the PORT measures the "whole organism or gut-level responses a child has toward a parent [which] are much more reflective of what the child's actual interactions or experiences with that parent have been." As with the BPS, apparently no attempts were made to connect the content of the PORT to legal stan­dards, and the theory upon which it is based, as described in the manual, is difficult to discern.

operational definition. The relevant content domains operational­ized by the PORT appear to be similar to the BPS, in that unconscious preferences for one parent over the other are proposed to be reflected in the obtained drawings. The content of interest to determine this prefer­ence is not overtly manifest, and must be assessed via indirect methods. Thus, the PORT presumably assesses the "best interests of the child" by delineating latent constructs that reflect the quality of the relationship between the parent and the child.

critique. The PORT lacks a firm connection between the content of the test, parenting capacity, and relevant legal standards. However, the PORT does claim to assess a general factor (i.e., the child's perception of his or her relationship with each parent) that is nonetheless relevant to issues of custody. The absence of a coherent, empirically supported theory regard­ing the relationship between human figure drawings and parenting is a particular limitation, however, as are data casting doubt on the reliability and validity of projective drawing techniques more generally (see, e.g., e).

Psychometric Development

standardization. Administration and scoring guidelines in the PORT manual are inconsistent and difficult to comprehend. The manual (a), which is augmented by various supplements (c), provides little coordinated structure for how to administer the test.

reliability. Limited reliability data are provided in the manual. Inter-rater reliability is not described, and test-retest estimates are based on very small studies that are not described in sufficient detail to address their methodological adequacy. One unpublished dissertation examining the reliability of the PORT offered little support for the PORT's stability (f).

norms. No normative group and no normative data are offered for the PORT.

critique. Limitations of the PORT regarding standardization and reliability are similar to those for the BPS (subjective administration, unknown reliability, absence of norms). Despite publication of the PORT over 13 years ago, no published research to remedy these limitations has appeared.

Construct Validation

Information about internal consistency analyses is not available. Research described in the manual (a) compares the identification of the "primary caretaking parent" by the PORT with other instruments and procedures of questionable validity. Bricklin (a) reported that the "primary caretaking parent" identified by the PORT was identical to the "parent of choice" identified by the BPS in 19 of 23 (83%) cases. In another study, clinicians used extensive clinical data to make judgments about 30 children and their parents. These determinations were consistent with the PORT findings in 28 of the 30 cases. The PORT manual also describes a study that compared PORT scores to judgments of clinicians who viewed 30 parent-child interactions. According to Bricklin (a), the PORT- identified "primary caretaking parent" agreed with the clinical opinions regarding parent of choice with better than 90% accuracy.

A number of other studies are referenced in a later published supple­ment (c), but the descriptions of the studies are too brief (i.e., one to two lines per study) to allow for an adequate critique of their methodology. Of these studies, perhaps most interesting is the description of a sample of 1038 cases, in which an agreement rate of 89% was obtained when PORT "primary caretaking parents" were compared to the judgments of "inde­pendent psychologists based on all clinical and life-history data avail­able." These findings, however, appear to be psychologists reporting to Bricklin how frequently the PORT results were consistent with their clini­cal impressions in custody cases they evaluated.

critique. Given the theoretical basis of the PORT and the method it uses to operationalize key variables, it would be difficult to conduct well-controlled studies to examine whether it is measuring the construct of interest, because its author considers overt measures to be invalid indi­cators of the child's "true" parent. The studies noted above, which appear to lack experimental rigor and/or suffer from various methodological limitations (e.g., criterion contamination), at best provide limited support for the construct validity of the PORT, which is insufficient to justify its use in applied settings.

Predictive or Classification Utility

The PORT manual (a) describes a series of studies that compared PORT scores to judges' decisions regarding custody. In 80 of 87 cases (92%), the judges granted custody to the parent identified as the PCP by the PORT.

Bricklin also claims (a, pp. 43-47) that the PORT may be useful for identifying children who have been physically or sexually abused, because their responses to the projective method can reveal the conse­quences of such abuse. No data supporting these claims are provided and general concerns regarding the utility of projective drawings more gener­ally raise questions about the validity of such claims (see, e.g., f).

Critique. One could conduct well-controlled studies using tradi­tional research methodologies that assess whether the "primary caretaking parent" identified by the PORT is "objectively" preferred by the child, and or is actually competent to care for the child. But such studies have not been done. Concerns about the predictive utility of the PORT parallel the con­cerns noted above for the BPS. There are only limited data to support the conclusion that the PORT identifies the better custodial parent, as defined by a judge's determination, and this information itself may have influenced the outcome being predicted. More generally, concerns have been expressed regarding whether this criterion measure is particularly informa­tive regarding what actually constitutes a child's "best interests' (g).

Potential for Expressing Parent-Child Congruency

Similar to the BPS, in theory the PORT was developed specifically to provide an index of the congruence between a child and his or her par­ents. In the absence of better data, however, the case has not been made that this task is accomplished using the results of this measure.

References

(a) Bricklin, B. (1989). Perception of Relationships Test manual. Furlong, PA: Village Publishing.

(b) Bricklin, B. (1993). Test Manuals Supplement #9. Furlong, PA: Village Publishing.

(c) Bricklin, B., & Elliott, G. (1997). Critical child custody evaluation issues: Questions and answers. Test manuals supplement for BPS, PORT, PASS, PPCP. Furlong, PA: Village Publishing.

(d) Conger, J., (1995). Perception-Of-Relationships Test. InJ. Conoley & J. Impara (Eds.), The twelfth mental measurements yearbook (pp. 747-748). Lincoln, NE: Buros Institute of Mental Measurements.

(e) Garb, H., Wood, J., & Nezworski, M. (2000). Projective techniques and the detection of child sexual abuse. Child Abuse and Neglect, 24, 437-438.

(f) Muir, K. (1997). Children's perceptions of their parents and susceptibility to suggestion. Unpublished doctoral dissertation. California School of Professional Psychology, San Diego.

(g) Otto, R. K., Edens, J. F., & Barcus, E. (2000). The use of psychological testing in child custody evaluations. Family and Conciliation Courts Review, 38, 312-340.

Parent Awareness Skills Survey (PASS) and Parent Perception of Child Profile (PPCP)

Author

Bricklin, B.

Author Affiliation

Widener University

Primary References

Bricklin, B. (1990). Parent Awareness Skills Survey manual. Furlong, PA: Village Publishing; Bricklin, B., & Elliott, G. (1991). Parent Perception off Child Profile manual. Furlong, PA: Village Publishing.

Description

The Parent Awareness Skills Survey (PASS) is described as a "clinical tool designed to illuminate the strengths and weaknesses in awareness skills a parent accesses in reaction to typical child care situations" (a, p. 4). The PASS is comprised of 18 childcare scenarios and was designed to sample relevant parenting behaviors that could be employed with chil­dren of various ages. After being presented with a particular situation the parent provides his or her reaction, with follow-up questioning by the examiner as needed. Responses are evaluated and scored based on guide­lines in the manual.

The Parent Perception of Child Profile (b) was designed to assess parental understanding and awareness of a child's development and needs across eight areas: interpersonal relations; daily routine; health history; developmental history; school history; fears; personal hygiene; and communication style. According to the manual, the PPCP can be self-administered or administered by an examiner in an interview format. The manual indicates that data need not be gathered in all eight cate­gories, and the examiner can decide which issues are most critical for a particular child and parent.

Conceptual Basis

concept definition. The content domains assessed by the PASS and the PPCP are face valid and have some intuitive appeal. The dimensions thought to be relevant to custody decision-making appear to have been delineated by Bricklin and his colleagues using rational means and reviews of the child development and parent-child interaction literature.

operational definition. For the PASS, operationalization of the functional parenting abilities appears to be rooted in the common sense belief that strengths and weaknesses in child rearing abilities can be assessed, in part, by querying parents directly about how they would respond to various child care scenarios. The method presumes that verbal responses are useful indicators of the actual abilities thought to be ger­mane to effective parenting.

Similarly, the functional abilities needed for effective parenting are presumed to be represented by the overt responses parents make to inquiries on the PPCP that reflect their level of knowledge about a child's needs and development. Quantification of these responses is presumed to lead to the identification of the parent who is better able to meet these needs.

critique. The content of the PASS and the PPCP generally appears to assess relevant domains of parenting. However, the adequacy and thor­oughness of content coverage has not been evaluated systemically, and the choice of dimensions was not linked with any particular psychological model of parenting or legal criteria regarding custody issues. Further con­tent validation procedures would be useful for determining the appropri­ateness and comprehensiveness of these measures.

Psychometric Development

standardization. Neither measure can be described as standard­ized. The PASS manual provides some scoring guidelines, although the author notes, "The evaluator... can apply his or her own standards in assigning the suggested scores. The PASS allows for wide latitude in scor­ing since its main purpose is to discover the relative (rather than absolute) strengths and weaknesses any individual or compared set of respondents manifest" (a, p. 11). The PPCP manual provides no scoring guidelines and, as noted above, examiners are free to administer only certain sections of the instrument.

reliability. Reliability estimates for these measures have not been reported.

norms. No norms are available for either measure.

critique. Despite the fact that both measures can be scored, there is no evidence that these scores are reliable, nor is there a way to compare them to the performance of anything approaching a reference group. Importantly, although the authors note that the PASS is supposed to pro­vide a relative comparison of the parents (which does not obviate the need for reliability data and a standardization sample), it also is claimed that the PASS can be administered to one parent with scores having inter­pretive significance (a, p. 16). Given the absence of more supportive infor­mation, this would seem to be an overstatement of what the PASS can provide. At best, both measures should be considered semi-structured interviews rather than methods for psychometrically sound quantifica­tion of parenting knowledge or ability.

Construct Validation

No data are provided to support the construct validity of either the PASS or the PPCP. No peer-reviewed, published reports have examined their validity.

critique. It is unclear exactly what the scores on these measures mean, although it is likely that they reflect to some extent the clinical judg­ment and intuition of the examiner, given the non-standardized nature of the tasks and the lack of normative data. Unlike the child-report measures developed by Bricklin that claim to offer projective measures of "uncon­scious" preferences, the PASS and the PORT easily could be validated using commonly employed methodologies and statistics (e.g., factor analysis, multi-trait and multi-method approaches).

Predictive or Classification utility

No data are available to examine the predictive utility of either measure.

critique. Clearly, the absence of such data place serious restrictions on what examiners may conclude about the meaning of scores on the PASS or the PPCP in relation to custody decision-making. As such, the reporting of specific scores does not appear warranted at present.

Potential for Expressing Parent-Child Congruency

Similar to Bricklin's child-report measures, in theory the PASS and PPCP could provide useful information regarding the parent's functional abilities and their fit with the needs of the specific child, if there were data available to judge the adequacy of the instruments. Used as semi­structured interviews, these two adult measures might be able to identify potential strengths and weaknesses of parents at a more global (i.e., less quantitative) level, which could then be combined with information from other relevant data sources about the child's needs.

References

(a) Bricklin, B. (1990). Parent Awareness Skills Survey manual. Furlong, PA: Village Publishing.

(b) Bricklin, B., & Elliott, G. (1991). Parent Perception of Child Profile manual. Furlong, PA: Village Publishing.

Child Abuse Potential Inventory (CAPI)

Author

Milner,J. S.

Author AiJiliation

Department of Psychology, Western Carolina University

Primary Reference

Milner, J. S. (1986). The Child Abuse Potential Inventory manual (2nd ed.). Webster, NC: Psytec Corporation.

Description

The Child Abuse Potential Inventory (CAPI) (i) was developed by Milner and his colleagues to assess parental risk for physical child abuse. It was designed to be a screening tool for use by protective service work­ers when investigating suspected cases of child abuse. The CAPI Inventory Form IV was released for use in applied settings in 1986. It uses a 160-item forced-choice questionnaire on which examinees indicate that they agree or disagree with the question when applied to themselves or their own circumstances. The items typically refer to some psychological condition or opinion (e.g., "I am often lonely inside," "Children should never be bad"). Items were written at a third-grade reading level and typ­ically can be administered in 10 to 20 minutes. A Spanish version of the instrument also is available (c, i).

Within the 160 items of the CAPI are 77 items constituting a physical abuse scale. Scoring templates, providing weighted scores for each item, are used to obtain a raw score sum across the 77 items. The abuse scale also provides six factor scores (distress, rigidity, unhappiness, problems with the child and oneself, problems with the family, and problems with others). However, the author recommends only the total physical abuse scale score for use in assessing the potential for physical child abuse.

Although the items composing the physical abuse scale have remained unchanged since the inventory's introduction in 1977, three validity scales more recently have been included to assist in detecting response distortion. These validity scales (a lie scale, a random response scale, and an inconsistency scale) are used in various combinations to form three validity indexes: Faking Good, Faking Bad, and Random Response. Most recently, two special scales (ego-strength scale, j; loneli­ness scale, 1) were designed to provide the test user with additional clini­cal information about the respondent. An applied manual (1) provides information to assist test users in administration and interpretation, and a revised technical manual (i) provides information about the development and validity of the inventory.

Milner cautions the user against applying the CAPI as a diagnostic tool for labeling purposes, recognizing that a score on this instrument alone does not warrant any diagnosis or conclusion about an examinee. Yet many clinical situations could make use of an instrument of this type for purposes of identifying the need for further assessment, cross-checking data from other sources in a more comprehensive evaluation, exploring potential treatment needs, or evaluating treatment programs designed to modify parental dysfunction. Based on an accumulation of many high risk and control samples, the Manual identifies cutoff scores for the upper 5% of normative samples. Milner cautions that these cutoffs are arbitrarily chosen, and notes that cutoff criteria must be determined by individual users based on their own purposes and local samples. The manual con­tains a detailed discussion of the effects of base rates on predictive accu­racy, as well as their importance when interpreting CAPI scores.

Conceptual Basis

concept definition. Milner (i) began with a survey of over 700 arti­cles and books, cataloguing personality traits and background variables that research and theory suggested were characteristic of parents who abuse or neglect their children. The variables were grouped logically into homogeneous categories (e.g., feelings of isolation and loneliness, nega­tive childhood experiences often including abuse). Milner notes that, although "psychiatric and interpersonal perspectives" on child abuse were considered important in defining the relevant content domain, they did not rely on any one particular theory of abuse when developing the item pool.

As can be ascertained by the review of special scales noted above, the CAPI was designed with awareness that respondents might not approach the testing situation in a completely forthright manner. As such, con­structs related to response distortion were also identified and attempts were made to operationalize and measure these factors that might affect the validity of the assessment process.

operational definition. In terms of delineating the criterion of inter­est, the CAPI manual indicates that child abusers were operationally defined during the development of the scale as those parents who were defined by social service workers as having committed abuse, based pri­marily on the North Carolina child abuse reporting law in place at that time. Notably, this did not necessarily mean that the case had been adju­dicated, only that it was the case worker's opinion that sufficient evidence was present to warrant such a conclusion.

Regarding the instrument itself, item generation and evaluation was an iterative process. First, the test developers wrote 15 to 20 items for each of the concepts derived from their literature review, creating a 334-item pilot form of the CAPI. This stage involved input from child protective service workers in the field and research staff members. In the first pilot test of this measure, Milner and Wimberley (t) identified the items that best discriminated between 19 abusive and 19 non-abusive parents (matched on nine demographic and background variables). Through a series of analyses, a subsample of items was identified that appeared most promising for differentiating abusers from non-abusers. A revised, 160- item measure was then constructed, which incorporated all of the promis­ing items (some with slight modifications) and several new ones based on input from social service workers. A second study, using a much larger sample (n = 130), was conducted to further examine the utility of the 160 items. A total of 77 items significantly discriminated between the abusive and non-abusive parents. These items ultimately were used to create the Abuse scale (the predictive accuracy of which is described below).

The three validity scales included in the CAPI were developed inde­pendent of the above process. In order to create the Lie scale, Milner first developed 59 items that focused on denial of common but undesirable attributes (e.g., "I never act silly"). These items were then presented to samples of college students and P.T.A. groups to identify those that were endorsed relatively infrequently (i.e., response rates approximating 15%). A total of 18 items with socially desirable item content and low endorse­ment rates were identified that ultimately were incorporated into a later edition of the CAPI.

For the Random Response (RR) scale, the manual (i) describes two studies in which 65 filler items from an earlier version of the CAPI were used to develop an 18-item index of infrequent responding. Although also comprised of items with low endorsement rates, unlike the content of the Lie scale these items were not intended to reflect socially desirable behav­iors. Also, items initially were selected for inclusion only if they failed to correlate with the Abuse scale score.

Finally, the Inconsistency scale was developed by identifying item pairs that were either similar (10 item pairs) or opposite (10 item pairs) in content. Response patterns that indicate inconsistent responding to similar items or consistent responding to dissimilar items are scored as evidence of inconsistent or random responding.

Because none of these scales was considered by Milner to be a pure measure of a specific response style (e.g., high scores on the Lie scale ostensibly can indicate either socially desirable or random responding), response distortion indices based on the configuration of these scales were developed. For example, the Faking-good index is derived based on consideration of both Lie scale scores and RR scores. Faking good is con­sidered to be present when there is an elevated Lie scale score combined with a normal range RR score.

critique. The CAPI was not developed as an index of dimensions of parenting attitudes or abilities generally, but specifically to identify par­ents who are at high risk for abuse. Thus, the CAPI appropriately began with a survey of research on that parent type, not a review of important parental attitudes in general. Its content, therefore, is relevant for address­ing questions of high risk of abuse, but it may not be relevant for address­ing other questions of parental ability that may be important for custody decision-making. One would expect, for example, that there are more and less adequate parents, by some other criterion, among non-abusive par­ents, and the CAPI was not designed to address such questions.

The concept and operational definitions of the CAPI appear to provide adequate content relevance and content coverage for its intended purpose. One question that might be raised is whether the process of item reduction (from 334 to the shorter, final version) employed test develop­ment samples of subjects with an adequate range of sociodemographic features to avoid selection of items relevant for only certain subgroups of abusers. This question will be addressed later in the review of validity studies.

Psychometric Development

standardization. The CAPI can be administered to individuals or groups and requires only a pencil and four-page questionnaire for each respondent. Items are rated dichotomously (agree-disagree). The inventory can be hand-scored using templates or computer-scored. To decrease scor­ing errors, computer scoring is recommended by the test author.

reliability. For the 77 items that contribute to the Abuse scale, inter­nal consistency estimates are reported in the technical manual and can be found in many subsequent studies. For the general population, at-risk, neglectful, and physically abusive groups, split half reliabilities range from.96 to.98, and KR-20 reliabilities range from.92 to.95. Split half reli­abilities and KR-20 coefficients by age, gender, ethnic, and education within the above groups range from.93 to.98 and.85 to.96 respectively. Temporal stability estimates are not available for abusers. For non- abusive examinees, however, test-retest reliability for the Abuse scale for 1-day, 1 week, 1 month, and 3 month intervals has been reported as.91,.90,.83, and.75, respectively (i). Internal consistency reliabilities for the six abuse factor scales are lower than the full abuse scale and more variable.

For the Lie scale, KR-20 reliability coefficients have been reported in the.70s for most samples. Relatively similar values have been reported for the Random Response and Inconsistency scales. Test-retest reliability (non-abusive subjects) has been reported as.86 (one day) and.83 (one week) for the Lie scale, although lower values have been obtained for the other two validity scales.

Notably, in addition to general reliability information, the CAPI man­ual also provides an overview of the Standard Error of Measurement (SEM) for the Abuse scale, subscales, and validity scales, as well as a description of how these values should be used in interpreting individual scores.

norms. Descriptive statistics for 836 respondents are presented in the CAPI manual (i, Table 2.7). In addition, similar information is pre­sented for various subgroups, with a cumulative N of over 5,000 respon­dents drawn from various studies. The manual notes that "while these data are provided for general comparison purposes, they must be viewed with caution because the studies are not always directly comparable" (i, pp. 13-14). In situations in which samples of parents are being assessed who are demographically similar, it is suggested that users consider developing their own local norms.

critique. The structure of the CAPI allows for ease of administra­tion and scoring, although Milner reports unpublished data in the manual to suggest that field workers often make errors in the computation of scale scores. Also, substantial evidence supports the internal consistency and temporal stability of measurement with the CAPI scales and subscales. The clear presentation of information related to the SEM in the manual also is an advantage. One somewhat unusual aspect of the CAPI is that it does not use standardized scores, which makes interpretation of the infor­mation provided cumbersome at times.

Although the CAPI manual is exhaustive in many respects, the nor­mative group could be described in greater detail and, like the subgroups reported in other tables, this group appears to be primarily a sample of convenience. No discussion is provided to indicate that the sample was census-matched or geographically diverse. Although the various sub­samples referenced in the manual appear to be diverse (e.g., location, age, socioeconomic status), the usefulness of the CAPI would be enhanced by a more extensive description of the available normative data.

Construct Validation

Initial factor analyses conducted on a sample of 65 abusing and 65 non-abusing parents revealed seven factors: Distress, Rigidity, Child with Problems, Problems from Family and Others, Unhappiness, Loneliness, and Negative Concept of Child. Factor scores for Distress, Rigidity, and Unhappiness most significantly distinguished abusers from the non­abusers. Subsequent factor analyses of unpublished data (n = 220) are reported in the manual and used to support a six-factor structure that generally is similar to the seven factors noted above (see i, Table 3.2). Despite some support for these subscales, the manual indicates that only the total Abuse score should be used to screen for child abuse.

Substantial data in addition to theses factor analytic results support the construct validity of the CAPI abuse scale. The studies are far too extensive to review here in adequate detail, so only a brief summary will be provided. (For more extensive reviews, see i, n, o). The majority of this research has found expected correlational relationships between CAPI abuse scores and a variety ofindividual, family, and environmental risk factors. Positive rela­tionships generally have been reported between CAPI scores and exami­nees' childhood history of abuse, social isolation and lack of social support, and negative family interactions and family problems. Studies also have found that individuals with higher CAPI scores report higher levels of negative affective states such as anger, depression, and anxiety. Relationships in the expected direction (i.e., an inverse relationship) have been reported for CAPI scores and personal characteristics such as self­esteem and ego-strength. Individuals with elevated CAPI scores also tend to have an external locus of control. Higher levels of perceived life stress and distress have been related to elevated scores on the CAPI Abuse scale, as well as deficits in knowledge regarding child development issues.

Expected relationships also have been observed between negative evaluations of children's behavior, inappropriate expectations for chil­dren's behavior, and CAPI abuse scores. Parenting characteristics such as use of harsh discipline and parenting practices have been studied and findings are in the expected direction, i.e., elevated CAPI scores are asso­ciated with use of more harsh discipline techniques, less positive parent­ing practices, and problems in parent-child interactions. Mixed findings have been found for studies investigating the relationship between CAPI abuse scores and authoritarianism, however, as well as their association with infant attachment problems.

Another line of research supporting the validity of the CAPI has addressed whether it is sensitive to the effects of treatment. Several studies have found that CAPI abuse scores decrease after treatment interventions such as an ecological-based intervention program, a parent training pro­gram, in-home treatments, and an intensive multi-modal intervention pro­gram. For example, the CAPI was administered pre- and post-treatment to a sample of 42 high-risk parents who participated in a 16-week program of films, speakers, and small group discussions designed to prevent child abuse and improve parenting knowledge and abilities. Prior to receiving this training, these high-risk parents scored significantly higher when compared to the CAPI normative sample, and their post-training scores were significantly lower than their pre-training scores. CAPI scores obtained for 33 of these parents seven weeks after post-training evaluation were also significantly lower than pre-training scores and not significantly different from post-training scores (x). Similar results have been obtained in other treatment outcome studies as well (see reference o for a review).

Support for the validity scales of the CAPI also has been demon­strated. The Lie scale correlates positively with the Marlowe-Crown Social Desirability Scale (.32) and with the MMPI Lie scale (.49) (d). Discriminant validity is evidenced by insignificant correlations between the CAPI Lie scale and scores on Rotter's Locus of Control scale (d), the MMPI Ego Strength scale and parents' grade point average (v), and the CAPI Abuse scale (h). More recent results indicated that parents instructed to fake good, fake bad, or respond randomly were identified with a high rate of accuracy using the CAPI validity measures (p). The validity indexes correctly detected as invalid 94.7% of the protocols generated by general­population parents and 91.1% of those generated by at-risk parents.

critique. The available data provide considerable support for the construct validity of the CAPI scales. The factor structure is reasonably stable and generally consistent with the concepts originally derived from the literature on child abuse. CAPI scores relate to several other measures of personality and response distortion in theoretically consistent ways. Further, the CAPI Abuse scale appears to be sensitive to change as a func­tion of remediation directed toward reducing child abuse potential. In the context of forensic evaluations, the accumulated evidence provides important convergent validity for interpreting a parent's CAPI Abuse score as an indicator of potential risk with regard to a child's welfare. Further, the CAPI validity scales may be helpful in detecting cases in which the CAPI Abuse scale results cannot be interpreted in the usual manner, due to the examinee's attempts to create an overly favorable impression or failure to attend to the item content.

In the first edition of this text, two cautions were noted regarding the construct validity of the CAPI. The first was the caveat that the results reviewed in support of the CAPI Abuse scale did not allow one to assume that a high score on the CAPI is indicative of past or future child abuse. The studies reported here do not eliminate that concern, although there is now greater evidence that non-abusing parents who are simply distressed are relatively unlikely to get extremely high scores on the CAPI. The sec­ond concern related to the potential impact of sociodemographic charac­teristics. The early studies employed samples consisting predominantly of white, lower socioeconomic females, with a tendency toward rural resi­dence, and it was unclear whether the same relationships between CAPI scores and variables examined in these studies would be found with ethnic minority samples, with fathers, or with middle- to upper- socioeconomic class samples. Research addressing many of these questions has now been conducted and, although differences have been noted in terms of group mean scores (see Table 2.7; Milner, 1986), the studies generally seem to sug­gest that the CAPI measures a similar construct in these populations.

Predictive or classificatory utility

Both concurrent and predictive validity studies have been conducted in order to investigate the ability of the CAPI Abuse scale to discriminate between known abusers and non-abusers. In an initial study (u) using dis­criminant analysis, the CAPI Abuse scale correctly classified 125 out of 130 subjects (96% hit rate, all misclassifications false negatives), using a sample of 65 abusive parents and 65 non-abusive parents matched on nine demographic and familial variables, including ethnic background.

In more diverse populations, subsequent studies using discriminant analysis have resulted in classification rates in the mid-80% to low-90% range (e.g., references r, s). Use of the conservative standard-weighted scoring procedures (as opposed to statistically optimized cut scores) in one study resulted in an overall classification rate of 90.2% for a sample of physical child abusers and comparison parents (k). This study used only valid protocols and the 215-point cut score recommended in the manual. A slightly lower overall rate of correct classification (86.4%) was obtained when invalid protocols were included in the analyses. Use of an optimally-derived 166-point cut-off resulted in a slightly higher overall classification rate compared to the 215 point cut-off. Similarly, another study (a) reported correct classification rates of 87.7%, 73.3% and 100% of physical child abusers, non-abusive comparison parents with childhood history of abuse, and non-abusive comparison parents without childhood history of abuse, respectively, using standard scoring procedures and the 215 point cut-off, and including invalid protocols. Use of the 166-point cut-off resulted in correct classification rates of 96.7%, 60.0%, and 83.3% for the three groups of subjects previously mentioned.

In a truly predictive (rather than concurrent or retrospective) validity study (q), a group of 200 participants were followed in a treatment pro­gram for parents at risk for parenting problems. Forty-two parents were reported for confirmed abuse (11), neglect (15) or failure-to-thrive (16) during the follow-up. Pretreatment CAPI scores were above the cutoff score for all of the abusive parents, and CAPI scores correlated with abuse (.34), neglect (.19), but not failure-to-thrive findings (.12). Despite these generally positive findings, the abusive parents accounted for only 10.7% of the 103 at risk parents who had pretreatment CAPI scores above the cutoff; that is, the majority of parents with pre treatment scores above the cutoff criterion apparently did not subsequently abuse their children.

Use of the CAPI Abuse scale to classify other types of abusive groups (as compared to the identified target group of suspected physical child abusers) has resulted in somewhat lower classification rates (e.g., lower 70%s). Such abusive groups include "mildly" abusive parents (g) and physically abusive and neglectful parents (b). One study has reported markedly lower classification rates for the CAPI than have been obtained in most other research (e), finding that only 28% of a combined group of physically and sexually abusive parents referred from a treatment group were correctly classified using the CAPI. Milner (o) has criticized this methodology, however, due to the inclusion of sexual offenders and use of persons already in treatment.

Finally, the classification accuracy of the CAPI has also been studied using non-abusive groups. Specificity rates for low-risk mothers (f), nurturing mothers (i, k), and nurturing foster parents (b) have been 100%. However, there is some evidence that parents of children with a history of medical illnesses and injuries (m) and developmental disabilities (w) may be at greater risk to be misclassified on the Abuse scale. Milner has tenta­tively noted that the higher scores of parents with children who have suf­fered injuries may suggest that some of these injuries actually resulted from abusive behavior, although this has not been tested directly.

critique. Published studies clearly indicate the ability of the CAPI to discriminate between parents who have or have not abused their chil­dren in the past, including samples with diverse sociodemographic char­acteristics. However, one should remember that these results are not sufficient to conclude that abuse has occurred, or will in the future, based on the CAPI alone.

There are various reasons for this assertion, including ethical issues regarding appropriate test use, impact of varying base rates across sam­ples, and costs of both false positive and false negative errors. Central to these reasons is the fact, recognized by Milner (i), that the cost of identify­ing 100% of the future abusive parents in his samples has been a very large false positive rate. Almost 90% of the parents with above-cutoff scores did not subsequently abuse their children (although all abusers did meet this criterion). It should be noted, however, that all of the parents were receiving treatment (which also highlights the importance of situa- tional/contextual factors that are not accounted for by instruments such as the CAPI). At least some of the high pretreatment CAPI parents who did not subsequently abuse their children might have done so if treatment had not been provided. Nevertheless, the study results serve as a graphic reminder that an instrument can be highly successful at identifying abusers, yet incapable of being used alone as a practical and accurate diagnostic indicator of future abuse because of its potential for overprediction.

Potential for Expressing Parent-Child Congruency

The CAPI was not intended to assess various parenting abilities or skills. Thus it is not meaningful to consider how parenting dimensions on the measure might be used to compare a parent's abilities to the parallel needs of a specific child.

References

(a) Calisco, J., & Milner, J. S. (1992). Childhood history of abuse and child abuse screening. Child Abuse and Neglect, 16, 647-659.

(b) Couron, B. (1982). Assessing parental potentials for child abuse in contrast to nurturing (Doctoral dissertation), United States International University, 1981. Dissertation Abstracts International, 42, 3412B.

(c) De Paul, J., Arruabarrena, I., & Milner, J. S. (1991). Validation de una version Espanola del Child Abuse Potential Inventory para su uso en Espana. ChildAbuse and Neglect, 15, 495-504.

(d) Ellis, R., & Milner, J. (1981). Child abuse and locus of control. Psychological Reports, 48, 507-510.

(e) Holden, E., Willis, D., & Foltz, L. (1989). Child abuse potential and parenting stress: Relationships in maltreating parents. Psychological Assessment, 1, 64-67.

(f) Lamphear, V., Stets, J., Whitaker, P., & Ross, A. (1985, August). Maladjustment in at-risk for physical child abuse and behavior problem children: Differences in family environment and mar­ital discord. Paper presented at the meeting of the American Psychological Association, Los Angeles.

(g) Matthews, R. (1985). Screening and identification of child abusing parents through self­report inventories (Doctoral dissertation, Florida Institute of Technology, 1984). Dissertation Abstracts International, 46, 650B.

(h) Milner, J. (1982). Development of a lie scale for the Child Abuse Potential Inventory. Psychological Reports, 50, 871-874.

(i) Milner, J. (1986). The Child Abuse Potential Inventory: Manual (2nd ed.). Webster, N.C: Psytec.

(j) Milner, J. (1988). An ego-strength scale for the Child Abuse Potential Inventory. Journal of Family Violence, 3, 151-162.

(k) Milner, J. (1989). Additional cross-validation of the Child Abuse Potential Inventory. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 1, 219-223.

(l) Milner, J. (1990). An interpretative manual for the Child Abuse Potential Inventory. Webster, N.C: Psytec.

(m) Milner, J. (1991). Medical conditions and the Child Abuse Potential Inventory specificity. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 3, 208-212.

(n) Milner, J. (1994). Assessing physical child abuse risk: The Child Abuse Potential Inventory. Clinical Psychology Review, 14, 547-583.

(o) Milner, J., Murphy, W., Valle, L., & Tolliver, R. (1998). Assessment issues in child abuse evaluations. In J. Lutzker (Ed.), Handbook of child abuse research and treatment (pp. 75-115). New York: Plenum.

(p) Milner, J., & Crouch, J. (1997). Impact and detection of response distortions of parenting measures used to assess risk for child physical abuse. Journal of Personality Assessment, 69, 633-650.

(q) Milner, J., Gold, R., Ayoub, C., & Jacewitz, M. (1984). Predictive validity of the Child Abuse Potential Inventory. Journal ofConsulting and Clinical Psychology, 52, 879-884.

(r) Milner, J., Gold, R., & Wimberley, R. (1986). Prediction and explanation of child abuse: Cross-validation of the Child Abuse Potential Inventory. Journal of Consulting and Clinical Psychology, 54, 865-866.

(s) Milner, J., & Robertson, K. (1989). Inconsistent response patterns and the prediction of child maltreatment. Child Abuse and Neglect, 13, 59-64.

(t) Milner, J., & Wimberley, R. (1979). An inventory for the identification of child abusers. Journal of Clinical Psychology, 35, 95-100.

(u) Milner, J., & Wimberley, R. (1980). Prediction and explanation of child abuse. Journal of Clinical Psychology, 36, 875-884.

(v) Robertson, K., & Milner, J. (1983). Construct validity of the Child Abuse Potential Inventory. Journal of Clinical Psychology, 39, 426-429.

(w) Rodriguez, C., & Murphy, L. (1997). Parenting stress and abuse potential in mothers of children with developmental disabilities. Child Maltreatment: Journal of the American Professional Society on the Abuse of Children, 2, 245-251.

(x) Thomasson, E., Berkovitz, T., Minor, S., Cassle, G., McCord, D., & Miner, J. (1981). Evaluation of a family life education program for rural "high risk" families: A research note. Journal of Community Psychology, 9, 246-249.

Parenting Stress Index (PSI)

Author

Abidin, R. R.

Author Affiliation

University of Virginia

Primary Reference

Abidin, R. (1995). Parenting Stress Index manual (3rd ed.). Odessa, FL: Psychological Assessment Resources.

Description

The Parenting Stress Index (PSI) (a) was developed to assess stress within parent-child systems. Based on Abidin's model of dysfunctional parenting, the PSI in its current form is a 120-item self-report measure that assesses stressors related to both child and parent issues, as well as life event stressors. The PSI, which is intended for parents who have children between the ages of one month and 12 years, is comprised of several scales and subscales. In a portion of the pSi called the "parent domain" (54 items), three subscales are intended to assess the "personality and pathol­ogy" (a, p. 30) of the parent: Depression, Competence, and Attachment. Another 4 subscales were developed to assess situational stressors thought to be influential regarding the level of distress experienced by the parent: Spouse, Health, Role Restriction, and Isolation. Subscales within the 47- item "child domain" scale included four thought to be characteristics of temperament: Adaptability, Demandingness, Mood, and Distractibility- Hyperactivity. Finally, two subscales—Acceptability and Reinforces Parent—were considered to reflect interactive aspects of the parent-child relationship. The 19-item Life Stress scale, which assesses the presence of various stressors outside the parent-child relationship, contains no sub­scales and is considered optional. The PSI also includes a measure of defensiveness called the Defensive Responding scale.

These subscales can be aggregated to form separate Child Domain and Parent Domain scores, as well as a combined Total Stress score. The Life Stress scale is not included in this total. Raw scores on these scales are converted to percentile scores that are based on the normative group (described below).

Conceptual Basis

concept definition. Abidin (a) offers a detailed model of parenting stress that is reflected in the subscales developed for the PSI. The PSI is based on the premise that dysfunctional parenting results from stressors related to the three areas noted above (parent and child characteristics, life events) and that the effects of these stressors on parenting ability are addi­tive. Abidin identified these content domains using primarily a rational- theoretical approach, based on reviews of the infant development, parent-child interaction, attachment, child abuse and neglect, child psy­chopathology, childrearing practices, and stress literatures, as well as his clinical experience.

The PSI is presented as an instrument that may provide information that is relevant to legal decision-making. For example, it is described in the publisher's test catalogue as being useful in assessing for child abuse potential and custody decision-making. However, it does not appear to have been designed with this particular objective in mind. The manual does not discuss legal constructs and issues in the context of the instru­ment's conceptualization or development in any significant way, although the psychological content domains that the PSI is intended to assess clearly have implications for legal decision-making in these areas.

operational definition. An iterative process was used to opera­tionalize the content domains of the PSI. This began with the develop­ment of a large item pool by the author and his colleagues, followed by pilot testing with mothers of children younger than three years of age, then a review by a panel of six experts in the area of parent-child interac­tion. The experts appear to have addressed the content and construction of each individual item, although it is not clear that they were involved in refining the broader content domains the items were intended to measure. (An unpublished doctoral dissertation provides a comprehensive over­view of the content validational procedures employed for the PSI: see ref­erence d). For the original 150-item PSI, the manual notes that 95% of the items directly focused on content that had been demonstrated to be rele­vant to parenting stress in at least one research study.

Unlike previous editions, each item on the PSI now contributes only to one subscale. Most of the items are rated by the parent on a 5-point Likert scale (Strongly agree to Strongly disagree) although there are a few exceptions to this format (e.g., the Life Stress scale uses a dichotomous format). Some of the items are descriptive in nature (i.e., reflecting specific behaviors the child does or does not engage in), whereas others are framed to reflect how the child's behavior affects the parent or whether the child's behavior deviates from the parent's expectations. Raw scores for the scales and subscales are converted to percentile ranks, which are reported in 5-point intervals.

critique. Although somewhat more detail could be provided in the manual, the content validation procedures used in the development of the PSI are a step ahead of many instruments that currently are in use. There is a clearly identified psychological theory on which the PSi is based, and the content of the items themselves has been derived from empirical research. Moreover, unlike other instruments included in this review, the PSi does not purport to measure legal constructs specifically; thus it avoids some of the pitfalls associated with such claims (e.g., that perform­ance on this scale can identify the "better" custodial parent).

Psychometric Development

standardization. The paper-and-pencil format of the PSI is such that administration and scoring are relatively straightforward. Both hand­scoring and computer scoring options are available.

reliability. Test-retest reliability estimates are reported over rela­tively short durations (e.g., 3 weeks, 3 months) as well as a one-year follow-up. In a relatively small sample (n = 37), the one-year stability coefficients for the Child, Parent, and Total Stress scores were.55,.70, and.65, respectively (a).

norms. The normative sample reported in the PSI manual (a) con­sisted of 2,633 mothers who were recruited primarily from "well-child­care" pediatric clinics in central Virginia. Others were recruited from day care centers, health maintenance organizations, and general pediatric clinics. A small number of subjects (3.6%) had been referred for diagnostic services related to behavioral or health problems. Almost all were from the East Coast of the United States. No attempt was made to match census data or stratify the sample, and the author acknowledges that the norma­tive group is a sample based on opportunity.

A description of a much smaller normative group of men (n = 200) also is included in the manual. Unlike the maternal sample (11% African American, 10% Hispanic), the male group was almost exclusively (95%) Caucasian. They also appear to be more educated than would be expected, in that almost half were college graduates (compared to 27% of the mothers). PSI scores for men were somewhat lower than for women, although it is not clear if this is in any way a function of demographic differences.

The manual also includes statistics for a sample of 223 Hispanic parents from New York City who completed a Spanish-language version of the PSI. Raw scores for this sample appear to be somewhat elevated in comparison to the English-language version, although it is unclear whether the Hispanic sample included data from fathers rather than mothers only. Also, SES level appears to be lower for the Hispanic sample overall.

critique. The PSI appears to be reasonably sound psychometrically. Administration and scoring are straightforward. Given that stress is not construed necessarily as a stable construct, the test-retest reliability esti­mates seem appropriate. Further detail would be helpful regarding some issues, especially inclusion of the Standard Error of Measurement for the scales and subscales. The biggest limitation of the PSI is the normative sample, which is essentially a very large sample of convenience. More detail could be provided regarding various issues, such as some break­down of PSI scores in relation to the race/ethnicity of the mother.

Construct Validation

Internal consistencies for the normative group in the manual ranged from.70 to.83 for the child subscales and from.70 to.84 for the parent subscales. The Child Domain, Parent Domain, and Total Stress scores all were above.90. Somewhat lower values have been reported for a cross- cultural sample in the manual.

The published literature relating to the construct validity of the PSI is voluminous and cannot be reviewed here in detail. (See references a and b for more detailed reviews.) Results of this research generally are support­ive, although there are significant concerns regarding the factor structure of the scale. Analyses reported in the manual indicate that many (and in some cases, the majority) of the items on the subscales do not load sub­stantially on the underlying dimensions of interest, and some items cross­load onto other scales (see Tables 7-9, a).

Concerns regarding its factor structure notwithstanding, the conver­gent validity of the PSI has been well established, in that the scales and subscales tend to correlate in expected directions with other theoretically relevant constructs (e.g., marital dissatisfaction, childhood illnesses, child­hood behavior problems, social isolation). Importantly, criterion measures in many of the studies cited in the manual and in professional journals are structured behavioral observations of parent-child interactions, rather than simply other self-report measures. For example, one study (h) found that, in a sample of 49 pre-schoolers and their mothers, scores from the Child Domain correlated highly with Maternal Involvement, Negative Affectivity, and Sociability scores coded from another instru­ment (Parent-Child Early Relational Assessment System), as well as with attachment status.

It also should be noted that the PSI has been translated into several other languages, and these versions appear to have psychometric proper­ties that are reasonably similar to the English version. Such cross-cultural stability would appear to support the significance of the parenting con­structs assessed by the PSI.

critique. The body of research on the PSI is so extensive that simple, evaluative summary statements are difficult to provide. The factor struc­ture reported in the manual is problematic, which raises some concerns about the dimensions that actually underlie the subscales (although these subscales generally do have adequate internal consistencies). Another (potentially related) criticism is that the intercorrelations between some of the subscales are rather high, suggesting that the constructs being assessed are not as distinctive as theorized. Clearly, however, the PSI appears to be embedded in a conceptual and nomological net that supports its use as a global measure of stress, even if some issues have yet to be resolved.

Predictive or Classification Utility

Scores on the PSI have been shown to correlate with several outcome measures of interest to forensic psychologists. For example, parents' PSI scores are sensitive to the effects of intervention. One study (c) reported decreases in PSI scores following completion of an intervention program for parents of children with aggression management problems. PSI scores also are associated with abuse potential. For example, one study (f) found significant differences on all PSI scales and subscales when comparing samples of physically abusive and non-abusive mothers, and also obtained theoretically consistent correlations between PSI scores and behavioral observation of abusive mothers' interactions with their chil­dren in a structured activity. Similarly, another study (e) reported that negligent mothers obtained much higher PSI scores than did controls.

Only one published study could be located that examined the predic­tive validity of the PSI Defensiveness scale (g). That study reported that, in a sample of parents instructed to "fake good," the classification accuracy of this scale was below 50%.

critique. There is clear evidence that parenting stress is associated with adverse outcomes, and that PSI scores seem to be capable of identifying those who, in a relative sense, are at greater risk for engaging in dysfunc­tional behaviors such as abuse or neglect. Of some concern with the PSI, however, is the suggestion in the manual to use a normatively-based cut score (85th percentile) for determining when to make referrals. The basis for this recommendation is not particularly clear and should be described in greater detail. More generally, given concerns about the normative sam­ple, making absolute (as opposed to relative) predictions about abuse potential based on the percentile scores seems questionable at this time. For example, even if large, positive correlations are obtained between PSI scores and abusive behavior, or group differences are obtained between abusing and non-abusing parents, such information does not address the accuracy of specific cut scores, nor does it take into consideration the importance of base rates when examining predictive accuracy.

Potential for Expressing Parent-Child Congruency

The PSI was not designed specifically to address parent-child congru­ence, although the scores obtained do purport to provide information regarding the extent to which various characteristics of the child in ques­tion appear to be a significant source of distress for the parent. As such, it seems that information from the PSI might be useful in conjunction with other data about the child's functioning, which might allow for a compar­ison of the parent's attributions to more "objective" measures. For exam­ple, a child may be quite accurately perceived by one parent as a significant source of stress because the child very clearly has an external­izing behavior problem (and realistically would strain the resources of anyone), whereas another child is seen as a source of stress by a parent when in fact the child would be considered "low maintenance" by most standards. Also, there would be reason to believe that comparisons might be possible across two parents in a custody dispute regarding the extent to which whatever stress they are experiencing is considered to be attrib­utable to the child.

References

(a) Abidin, R. (1995). Parenting Stress Index: Professional manual (3rd ed.). Odessa, FL: Psychological Assessment Resources.

(b) Abidin, R. (1997). Parenting Stress Index: A measure of the parent-child system. In C. P. Zalaquett & R. J. Wood (Eds), Evaluating stress. Lanham, MD: Scarecrow Press.

(c) Acton, R., & During, S. (1992). Preliminary results of aggression management training for aggressive parents. Journal of Interpersonal Violence, 7, 410-417.

(d) Burke, W. (1978). The development of a technique for assessing the stresses experienced by parents of young children. Unpublished doctoral dissertation. University of Virginia, Charlottesville.

(e) Ethier, L., Lacharite, C., & Couture, G. (1993). Childhood adversity, parental stress and depression of negligent mothers. Unpublished manuscript, University of Quebec at Trois- Rivieres, Canada.

(f) Mash, E., Johnston, C., & Kovitz, K. (1983). A comparison of the mother-child interac­tions of physically abused and non-abused children during play and task situations. Journal of Clinical Child Psychology, 12, 337-346.

(g) Milner, J., & Crouch, J. (1997). Impact and detection of response distortions of parenting measures used to assess risk for child physical abuse. Journal of Personality Assessment, 69, 633-650.

(h) Teti, D., & Gelfand, D. (1991). Behavioral competence among mothers of infants in the first year: The mediational role of maternal self-efficacy. Child Development, 62, 918-929.

Parent-Child Relationship Inventory (PCRI)

Author

Gerard, A.B.

Author Affiliation

Western Psychological Services

Primary Reference

Gerard, A. B. (1994). Parent-Child Relationship Inventory manual. Los Angeles: Western Psychological Services.

Description

The Parent-Child Relationship Inventory (PCRI) (b) is a self-report ques­tionnaire that was designed to assess adults' attitudes regarding parent­ing and their children. The 78 Likert-type items (rated from strongly agree to strongly disagree) combine into seven content areas: Parental Support; Satisfaction with Parenting; Involvement; Communication; Limit Setting; Autonomy; and Role Orientation. Two validity scales also are included. One is intended to identify socially desirable responding, whereas the other is comprised of highly inter-correlated item pairs and functions as an indicator of response inconsistency.

Conceptual Basis

concept definition. The content domains that the PCRI was designed to measure (noted above) were identified through a fairly rigorous process,

involving both qualitative (e.g., expert judges, literature reviews) and quantitative (e.g., factor analytic) methods (b).

operational definition. Initial item generation for a 345-item ver­sion was driven by a review of existing research on parenting, followed by factor analyses of the prototype measure that identified 14 relevant dimensions.

critique. The content validation procedures used in the development of the PCRI seem to have resulted in the development of scales that repre­sent at least some of the most relevant dimensions of relationship quality. Concerns have been expressed, however, that it is not comprehensive enough in terms of important content domains and that the construct of par­enting itself may be too contextual to assess using a self-report measure (d).

Psychometric Development

standardization. Given the self-report nature of the instrument and the simplicity of the scoring procedures (hand- and computer-scored options are available), it seems unlikely that any significant problems would occur with the PCRI in relation to administration or scoring.

reliability. Reliability estimates for the PCRI generally have been acceptable. In the standardization sample, the median alpha reported for the scales was.80 (range =.70 to.88) and the mean test-retest correlation for a subsample of 82 participants over a 5-month follow-up ranged from.44 to.71 (b).

norms. As reported in the manual (b), the normative sample for the PCRI consisted of 1,139 parents who were identified through schools and day-care centers, but with an overall response rate of only 4.4% to the ini­tial solicitation letter. The resulting sample was better educated and less culturally diverse than the U.S. population. Race and education effects were examined by pulling a stratified sample of 240 parents who approx­imated census data. Although results indicated some significant differ­ences in relation to both demographic factors, separate norms for racial and educational groups were not developed because the obtained differ­ences were presumed to identify substantive differences in parenting across these groups. Separate norms were developed for mothers and fathers, however, given that significant differences were obtained between these groups on five of the seven PCRI scales.

critique. As with most of the parenting measures reviewed, the normative sample for the PCRI has some limitations. The initial 4.4% response rate is reason for concern and, although it may be that the differ­ences between racial groups may in fact represent legitimate differences, at present this assertion remains unsubstantiated.

Construct Validation

Relatively few studies have been conducted on the PCRI, although results have been encouraging. The factor structure of the instrument seems to be reasonably well supported. Aside from these results, the man­ual notes three studies that have examined the validity of the PCRI. First, in a sample of 35 couples undergoing court-ordered mediation, PCRI scores were generally low and correlated in expected directions with the Personality Inventory for Children (c). Second, a shortened version of the PCRI was used to predict disciplinary styles among a sample of 174 par­ents. Those who had higher scores on the Communication scale were more likely to reason with their child and less likely to use spanking. The third study reported analyses on 26 adolescent mothers who were consid­ered to be 'at-risk.' As expected, these mothers obtained mean scores that were below the standardization sample on all scales. (See references a and d for reviews of these studies.)

critique. More research is needed regarding the construct validity of the PCRI. Although the factor analytic and other item-level analyses are encouraging, more extensive research using better samples and methodologies (e.g., employing adequate control groups) is needed before strong claims can be made regarding exactly what it is that the PCRI measures.

Predictive or Classification Utility

At present, no data are available that address the predictive or classi­fication utility of the PCRI.

critique. In the absence of such information, examiners should be very cautious regarding any claims made about the predictive validity of this instrument in forensic contexts. With essentially no data to suggest that PCRI scores identify those whose parenting abilities actually result in children who experience "better" developmental outcomes (e.g., higher academic achievement, lower rates of delinquency or substance abuse), it would be difficult to defend empirically claims that higher scores predict "better" parenting or that low scores predict inadequate parenting (a, d).

Potential for Expressing Parent-Child Congruency

If further empirical support can be provided for the PCRI, it seems that its use in combination with measures of child functioning might pro­vide relevant information regarding congruence. However, at present such results would have to be considered speculative.

References

(a) Boothroyd, R. (1998). Review of the Parent-Child Relationship Inventory. In J. C. Impara & B. S. Plake (Eds.), The thirteenth mental measurements yearbook (pp. 717-720). Lincoln, NE: Buros Institute of Mental Measurements of the University of Nebraska-Lincoln.

(b) Gerard, A. (1994). Parent-Child Relationship Inventory (PCRI): Manual. Los Angeles, CA: Western Psychological Services.

(c) Lachar, D. (1984). Personality Inventory for Children manual. Los Angeles: Western Psychological Services.

(d) Marchant, G., & Paulson, S. (1998). Review of the Parent-Child Relationship Inventory. In J. C. Impara & B. S. Plake (Eds.), The thirteenth mental measurements yearbook (pp. 720-721). Lincoln, NE: Buros Institute of Mental Measurements of the University of Nebraska- Lincoln.