And the multidimensional neglectful behavior scale

Chilent

Development and Preliminary Psychometric Properties of the Multidimensional Neglectful Behavior

DOI: 10.1177/1077559504269530

Thed at:

Additional services and information for Child Maltreatment can be found at:

Email Alerts:

>> - Nov 10, 2004

Downloaded from at UQ Library on August 1, 2012

Crystal A. MacAllum
Westat

Wendy Brown
New Hampshire Division for Children, Youth and Families

Authors’ Note: This research was supported by grants from Na-tional Institute of Child Health and Human Development (R01MD39144-01), Office of Juvenile Justice and Delinquency Pre-vention (2002-JW-BX-0002), and National Institute of Mental Health (T32MH15161).

409

Some of the most widely used instruments measur-ing neglect such as the Childhood Level of Living Scale (CLL; Polansky, Chalmers, Buttenweiser, & Wil-liams, 1978), and the Home Observation for Measure-ment of the Environment (HOME; Caldwell & Bradley, 1984) are observational measures that may be completed by child welfare workers or other pro-fessionals who know the family well, or by profession-als from a variety of backgrounds (DeVoe & Kaufman Kantor, 2002). Trocme’s Child Neglect Index (CNI; 1996) was developed to reflect the legal definition of neglect in Ontario, Canada, and was designed as a substantiation tool to be used mainly by child welfare workers or researchers using the Ontario legal stan-dards. Other instruments designed for adult infor-mants require respondents to report on their own child-rearing attitudes, for example, the Adult-Adolescent Parenting Inventory (AAPI; Bavolek, 1984), or behaviors. For some instruments, such as the Childhood Trauma Questionnaire (Bernstein et al., 1994) and the Multidimensional Neglectful Behavior Scale, Form A: Adolescent and Adult Recall Version (Harrington, Zuravin, DePanfilis, Ting, & Dubowitz, 2002; Straus, Kinard, & Williams, 1995), adults retrospectively evaluate their childhood expe-riences of abuse and neglect. In fact, few instruments are available that rely on child self-report measures of neglect, although some exist (e.g., McGee, Wolfe, Yuen, Wilson, & Carnochan, 1995; the Parental Acceptance-Rejection Questionnaire [PARQ], Rohner, 1986). McGee and colleagues (1995) asked adolescents about a range of maltreatment experi-ences but confined questions about neglect to behav-ioral examples reflecting improper care or lack of attention, whereas Rohner’s (1986) focus was on emotional neglect and abuse.

Instruments vary widely in regard to the types of neglect addressed. Some measures assess only one component of neglect, such as physical or emotional neglect (e.g., the Childhood Trauma Interview, Fink, Bernstein, Handelsman, Foote, & Lovejoy, 1995), whereas others evaluate a wide array of neglectful behaviors (e.g., Child Well-Being Scales, Magura & Moses, 1986). Instruments also vary as to their objec-tives. Although the majority of these instruments have been developed specifically to measure child neglect, others have instead focused on risk assessment or pro-gram evaluation (e.g., Magura & Moses, 1986).

CHILD MALTREATMENT / NOVEMBER 2004

Downloaded from at UQ Library on August 1, 2012

These principles provide a basis for distinguishing abuse from neglect. For example, acts of psychologi-cal aggression against a child (e.g. “you dummy”) are sometimes considered neglect because they disregard the child’s need for a supportive and loving relation-ship and for nurturance. Previous research has dem-onstrated that verbal attacks can be more damaging developmentally than physical attacks (Vissing, Straus, Gelles, & Harrop, 1991); however, we regard verbal attacks as acts of psychological abuse, not neglect. To illustrate the importance of keeping them separate, let us consider the proposition that neglect

conjunction with separate measures of etiological or consequence variables that are relevant for a particu-lar research or clinical assessment (Straus & Kaufman Kantor, in press). The four core areas that are encom-passed by the multidimensional measure of neglect are the domains of physical needs, emotional needs, supervision needs, and cognitive needs.

412 Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT

who are the child’s friends. The MNBS-CR computer-administered interviewing system used branching of questions based on the age of the child to present age-

from age 10 years to age 14 or 15 years are generally categorized as early adolescents by developmental psychologists (e.g., Compian, Gowen, & Hayward, 2004; Silverberg & Steinberg, 1990), and this is consis-tent with the practices of the Journal of Early Adolescence that has been dedicated to work concerning 10- to 14-

The development of child report measures must consider children’s terminology for objects and events, children’s linguistic development at different ages (vocabulary and degree of grammatical com-plexity understood), and the social and cultural aspects of language to which children are accus-tomed. Questions about the credibility and reliability of children’s reporting have been regarded as the foremost of challenges in child assessment. For exam-ple, it has been suggested that the child’s desire to conform or please may lead them to agree with infor-mation that is untrue. However, even very young chil-dren (age 3 to 6 years) are less suggestible for central events and actions that they understand (Bruck & Ceci, 1999; Powell, Roberts, Ceci, & Hembrooke, 1999). In this regard, the research on children’s epi-sodic or event memory is relevant. Preschool-aged children’s memories of events have been found to be vulnerable to suggestions offered by adults; older chil-dren are less susceptible (Ceci, Leichtman, Putnick, & Nightingale, 1993). In the absence of suggestion, however, even very young children will relay accurate information about an experienced event (White, Leichtman, & Ceci, 1997). The major difference between preschool and older children is in the amount of detail they provide. The bulk of the evi-dence indicates that school-aged children are able to provide relatively detailed and veridical accounts of events they have experienced (Ceci, Fitneva, &

regarding the time period of care assessed. Younger children were excluded from the study if they had lived in foster care for more than 6 months. Older children were excluded from the study if they had lived in foster care for more than 1 year.

Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT 413

the same for pictures of male and female caregivers and male and female children.

The Pilot Study: Phase 1

Sample. Cognitive testing was conducted in 2001 us-ing the Audio-Computer Assisted Self Report Inter-view (ACASI) and pictorial versions of the scale on a convenience sample of 47 children drawn from the community (most from after-school programs) and 19 children from the Maine Foster Care System and the Spurwink Child Abuse Forensic Program. The sample was almost equally divided by gender and age. However, in the clinical sample, three fourths of the children tested were in the younger age group, by design.

The MNBS-CR Core Domains

Emotional neglect. Previous conceptualizations of emotional neglect have been largely consistent on viewing psychological abuse or aggression (harmful insults) as separate from emotional neglect (the fail-ure to provide nurturance or to express affection to the child). Some have opted to categorize emotional maltreatment more broadly (e.g., Rose & Meezan, 1993). Others have considered parental substance abuse, antisocial behavior, and domestic violence to be equivalent to emotional neglect (e.g., Slack, Holl, Altenbernd, McDaniel, & Stevens, 2003) or focused

Supervisory neglect. Lack of supervision is the most common situation that brings caregivers and children to the attention of child welfare (Zuravin, 2001). Many of the items in the MNBS-CR reflect those same concerns, and the nine items in the MNBS-CR Super-visory subscale encompass areas such as knowledge of the child’s whereabouts, and the provision of adult su-pervision. We also defined this category of neglect to include broad but essential aspects of parental social-ization responsibilities such as attending to children’s misbehavior and limit setting. Most would agree that it is the duty of parents to socialize children into prosocial, noncriminal activities, by communicating that certain actions are against the law and unaccept-able to the parent. In fact, some of the correlation be-tween neglect and delinquency found in other research (e.g., Maxfield, Weller, & Widom, 2000) may be due to such parental failures. Unless these items are specifically measured and studied, the etiologic relationships between neglect and particular conse-quences may not be empirically clear. For example, one item states, “This child’s parent cares if she or he does things like steal.” Another states, “This child’s parent does not make sure that she or he goes to school.”

Physical neglect. There is a fair amount of consensus regarding parental responsibility to provide for the physical needs of the child for adequate food, shelter, clothing, and medical care, and this category of ne-glect is also typically addressed in state statutes (Tower, 1996; Zuravin, 2001). The Physical Neglect subscale is the largest subscale included in the MNBS-CR (12 items) because it must capture the multi-dimensionality of this construct. This domain assesses

Alcohol use. This subscale includes two items: allow-ing a child to use alcohol, and an indicator displaying parental intoxication, “This child’s parent gets drunk and cannot take care of him or her.” The Substance Abuse subscale was included because there is a rather extensive literature establishing the linkages between parental substance abuse and neglectful parenting (e.g., Famularo, Fenton, & Kinscherff, 1992; Harring-ton, Dubowitz, Black, & Binder, 1995). However, un-less clear linkages to maltreatment can be established, parental substance abuse is one of those recognizably harmful factors, along with corporal punishment, that may be overlooked in defining the risks to chil-dren or not charging parents with maltreatment. We included this subscale because it can permit a further

CHILD MALTREATMENT / NOVEMBER 2004

Child Depression. The mood-affect subscale (six items) of the Depression Profile for Children (Harter & Nowakowski, 1987) is included in the ACASI pro-gram to establish construct validity. Pictures for each item were developed as part of our computerized as-sessment. We selected this subscale based on commu-nications with our colleague, Milling Kinard, who, at our request, analyzed data from her studies of mal-treated children (e.g., Kinard, 1998) to determine which of the depression subscales most strongly corre-lated with child neglect compared to other maltreat-ment types. Across all maltreatment types, Kinard (1998, personal communication) found that the al-pha reliability coefficient for child ratings of depression using the mood-affect subscale was .77.

Other measures. For the clinical sample, limited data on the maltreatment history of the child and child and family characteristics were collected from clinical records of the forensic program. Data on IQ were con-sidered important relative to appreciating the com-prehensibility of the program to a range of children but were not consistently available. For the commu-nity sample, data on learning deficits were obtained from teachers in the after-school programs attended by the children.

Computer implementation. A basic essential in assess-ing the status and experience of children is to create an atmosphere in which the child feels able to trust the interviewer and feels safe (Garbarino & Stott, 1989). This is as important in developing and admin-

displayed in conjunction with the response categories (e.g., Richters & Martinez, 1993) to capture the degree or intensity of the child’s identification with the selected behavior. The program also allows chil-dren to change their responses if they wish, and it prompts the child if she or he skips a question. Another innovation is the inclusion of an interactive computer game, midway into the program to mini-mize children’s possible fatigue or boredom in doing the measurement program. On average, the program took between 30 and 40 minutes for completion.

It is also important to establish a context for ques-tions—and to convey that questions are not a demand to comply but a request for information. These were areas that were considered in gaining assent from children and in implementation of the MNBS-CR. Some children had already been through a forensic interview with staff psychologists or caseworkers. Although some have questioned if those children might be contaminated by prior interview (e.g., foren-sic interviews are conducted for purpose of investiga-tion and fact finding as opposed to the research inter-view), research indicates that contamination should not be a concern (Goodman, Aman, & Hirschman, 1987). If children were distressed or fatigued by other assessments, the testing was rescheduled.

• “This child’s (boy’s/girl’s) parent does not help him or her with homework.”
• “This child does not have clothes to wear to school and to play.”
• “Thischilddoesnothavehisorherownbedtosleepin.”

Other changes were made to simplify wording and vocabulary. For example, in an item asking older chil-dren about whether the parent cared if the child did things such as shoplift, the word shoplift was changed to steal. The majority of changes made as a result of the pilot were made to the graphic representation of items to ensure that the picture accurately reflected the item, and to minimize potential confounds in the picture so that the child would focus on neglectful and non-neglectful parental behaviors.

Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT 417

neglect; and 6 items measuring depression for a total of two 52-item scales (one for younger children, and

The majority of children were White, consistent with ethnic distributions in the states of Maine and New Hampshire. At the time of testing, 46% of the children lived with a biological parent. Analysis of record data on parents revealed that a number of fam-ily risk factors were present. Developmental disabili-ties were present in 16% of parents, and mental illness (mainly depressive disorders) had been diagnosed in almost one half of the parents. Another 41% of par-ents had a history of substance abuse problems, and 62% were experiencing current domestic violence.

Community sample. We also tested a comparison group of 87 children in the general community, drawn from aftercare programs in New Hampshire and Maine, after gaining parental consent and child assent. The community sample included children who completed the MNBS-CR at four New Hampshire sites (see Table 1). These children came from seven after-school programs in the more urban communi-ties of New Hampshire. More than one half of the pro-grams are Boys and Girls Clubs. The majority of this sample were girls and more likely to be younger than children in the neglect clinical sample. With respect to ethnicity, almost three fourths were White. Minori-ties were more prevalent in this sample because the after-school programs were drawn from more urban areas of New Hampshire. Of the children interviewed in the community sample, 10% were known to have learning disabilities.

Age

Peabody Picture Vocabulary Test. Some measure of the child’s cognitive ability is important because pre-vious research has suggested that neglect may be asso-ciated with cognitive deficits and poor academic performance (Kendall-Tackett & Eckenrode, 1997). In addition to the computerized version of the MNBS-CR, the Peabody Picture Vocabulary Test III (PPVT III) was given to children so that the child’s receptive vocabulary could be considered in interpreting the results. This widely used measure is quickly and easily administered and demonstrates excellent reliability (internal consistency alpha, 0.92 to 0.98; split half 0.86 to 0.97). Recent revisions of this instrument in-clude a better balance of ethnicity and gender in the illustrations. The PPVT was not used as a criterion in excluding children from the study or from analysis. We examined PPVT scores as correlates of the child’s reporting of neglect.

Child Behavior Check List. Data from the parent re-port version of the Child Behavior Check List (CBCL;

couple of additional questions. The number of items observed depended on how comfortable the inter-viewer felt about the child’s comprehension. When the interviewer felt that the child comprehended the questions and responses, the child could have privacy to answer the questions. Children could use ear-phones for the audio portion of the computer pro-gram and were reminded to raise their hands if there were any problems or questions. The interviewer sat in another part of the room that was nonintrusive to the child (exception with 6-year-olds, or any children where comprehension was a concern). Other technical aspects of the program and administration were as described above for the pilot study.

Ineligible Cases

because of comprehension difficulties, two because of identified mental retardation of the child after administering the ACASI, one because the child indi-cated he had not always answered honestly, one because parental time constraints led to an early ter-mination of the interview, and one because the child was unable to pay attention during the interview. This resulted in a final sample of 140 older children (116 clinical, 24 community) and 177 younger children (114 clinical, 63 community).

Analytic Approach

relatively uncommon, types of neglect such as paren-tal abandonment. In addition, the expanded measure

TABLE 2: Reliability Summary for MNBS-CR Scores for Clinical

Analyses conducted. Most of the analyses in this arti-cle focus only on the portion of the clinical sample with neglect allegations and, comparatively, the sam-

Among the older children with neglect concerns, the full core version of the MNBS-CR had high inter-nal consistency reliability (α = .94, expanded = .95). For the four primary neglect domains, the alpha coef-ficient of internal consistency reliability was highest for Supervisory Neglect (α = .85), with similar high alphas for emotional and physical neglect. The alpha for cognitive neglect was slightly lower, due most likely to the fewer items in this scale. Additional scales also yielded adequate internal consistency coeffi-

CHILD MALTREATMENT / NOVEMBER 2004

cients for depression, general appraisal, abandon-ment, and for exposure to conflict. The Alcohol Use subscale yielded a very low internal consistency; however, this includes just two items.

Community sample. Within the younger child com-munity sample, the alpha coefficient for the core and expanded versions of the MNBS-CR was .61, similar to that found for young children in the neglect sample. As shown in Table 2, alphas for individual scales were lower, with Physical Neglect showing the highest in-ternal consistency of the four core scales, and other scales were markedly low with the exception of De-pression. It should be noted that if individual items were removed from the Cognitive, Supervision, and Exposure to Conflict scales, the estimates of internal consistency reliability would improve significantly (.26, .17, and .43, respectively).

sample was similar to, but lower than, their younger child counterparts. It should be noted that if items were removed from the Supervision scale for older children in the community group, the estimate of internal consistency reliability would improve signifi-cantly (.43). It was not possible to calculate alpha coef-ficients for abandonment or alcohol use because of zero variance for one of the two items constituting

the tables). Neglected children in the younger group scored significantly lower on the PPVT than children without neglect concerns in the family (Neglect: M = 94.24, No neglect: M = 100.54; t = 2.33, p < .05). More-over, PPVT standard scores were related to physical neglect (r = –.28, p < .01) among 6- to 9-year-olds. Al-though 10- to 15-year-olds with neglect histories scored slightly lower on the PPVT than those individu-

CHILD MALTREATMENT / NOVEMBER 2004

Downloaded from at UQ Library on August 1, 2012

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

6- to 9-year-olds, domestic violence was significantly associated with more supervision neglect (r = .31, p < .05). Contrary to expectations, younger children with parents who were mentally ill reported less emotional neglect (r = –.35, p < .01).

Downloaded from at UQ Library on August 1, 2012

422 Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT

Looking first at the emotional domain for younger children, most items appeared to discriminate well, in the expected direction, except for “This child’s par-ent does not do fun things with him.” In addition, no children in either sample reported that the parent does not tell him or her that she or he is loved. Among older children, similar patterns emerged, all items worked in the expected direction, and there was some reporting of parental failure to express verbal

Youths Age 10 to 15 Years

NOTE: Neglect Total (Core) includes Emotional, Cognitive, Super-vision, and Physical subscales. Neglect Total (Expanded) includes Emotional, Cognitive, Supervision, Physical, Abandonment, Alco-hol Exposure, and one Exposure to Conflict item (“Parent lets other people in the house hurt him or her”).

In regard to the supervisory domain, several items emerged with low rates of occurrence for both of the youngest samples, including “Leaves him places where he does not feel safe”; “Does not make sure he goes to school”; “Leaves him or her alone.” For older children, some of the same issues emerged. In addi-tion, the items, “Does not call from work after school to check on the child,” did not appear to differentiate between the older samples well.

As noted above, physical neglect is a heteroge-neous concept, which is more often present among clinical than community samples, even though indi-vidual occurrences are low. As anticipated, gaps in the provision of care, which were consistent with severe neglect, are more likely to be reported by younger and older clinical sample children relative to commu-nity children. Exceptions were noted in items measur-ing the cleanliness of the home, permissiveness about junk food, and the warmth of the home environment. However, most of the latter items (except for the warmth of the home) likely did not present serious concerns to the welfare of the child.

Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT 423

ducted. It would seem that the parentification item may be viewed as socially desirable by all children. Furthermore, as the data showed, feeling hungry is not equivalent to children’s reporting of inadequate food in the house. The most poignant and telling item of this scale is the item indicating that the child did not feel that someone loved him or her. As antici-pated, neglected children reported this more often than community children.

Older children differed considerably from youn-ger children in regard to the reliability scores for exposure to conflict and violence. Older children reported these events more often, and older children had higher alpha levels for this subscale than did younger children. Possibly younger children were more fearful about reporting violence in the family or may have regarded telling such family secrets as tanta-mount to disloyalty to the parent. This suggests a need for further examination.

Support for the construct validity of the MNBS-CR was provided by the finding that two of the four major subscales were correlated with depressive symptoms in the predicted direction for younger children, and all four were correlated with depressive symptoms among older children. However, these findings may also be biased by the shared method variance of same child reporters on these MNBS scales.

Further support for the construct validity of the MNBS-CR was provided by findings supporting a sig-nificant association between a history of domestic vio-lence documented by clinicians, and supervisory neglect of younger children. (e.g., a history of domes-tic violence increased the child’s self-reporting of neglect). Among older children, a history of parental

Downloaded from at UQ Library on August 1, 2012

replace the item, “This child’s parent leaves him or her places where she or he does not feel safe”; (d) Physical scale: revision of one item, “This child’s par-ent did not take him or her to the doctor’s in the past year,” will replace “This child does not go to the doc-tor for check-ups”; (e) Exposure to Conflict: addition of one item, “This child sees grown-ups in the house throwing things”; (f) Alcohol Use: addition of one item, “This child’s parent uses drugs and cannot take care of him or her.”

The preliminary results suggest that the MNBS-CR has a great deal of potential as a tool for screening of neglect and clinical assessment of maltreated chil-dren by child protection agencies, researchers, and clinicians. Additional data collection, particularly with a larger comparison sample, is indicated, before clinical implementation. Some of the absences of anticipated findings noted above may be because of the small sample sizes, and consequently, low statisti-cal power. Further research using a multisample approach is indicated, including additional testing of children from the general community, CPS, and a more ethnically diverse sample. We also plan to develop and test a social desirability measure with the MNBS-CR. Future analyses will examine the concur-rent validity of the MNBS-CR by analyzing, for exam-ple, the concordance of the MNBS-CR with specific neglect clinical classifications and with analyses of video-taped parent-child interactions, and the con-cordance of neglect reporting between siblings. We plan to conduct specificity and sensitivity analyses so that specific cut points can be established for clinical and research usage, and we plan to explore other approaches to scoring the revised measure including the use of a weighted, severity approach, and the development of minor and severe scores of neglect.

Downloaded from at UQ Library on August 1, 2012

Kaufman Kantor et al. / MULTIDIMENSIONAL NEGLECTFUL BEHAVIOR SCALE–CHILD REPORT 427

Bornstein, M. H. (1992). Perception across the lifespan. In M. H.

Bornstein & M. E. Lamb (Eds.), Developmental psychology: An advanced textbook (3rd ed., pp. 155-209). Hillsdale, NJ: Lawrence Erlbaum.

Ceci, S. J., Leichtman, M., Putnick, M., & Nightingale, N. (1993).

Age differences in suggestibility. In D. Cicchetti & S. Toth (Eds.), Child abuse, child development, and social policy (pp. 117-137). Norwood, NJ: Ablex.

DeVoe, E. R., & Kaufman Kantor, G. (2002). Measurement issues in child maltreatment and family violence prevention programs. Trauma, Violence, & Abuse, 3(1), 15-39.

English, D. J. (1998). The extent and consequences of child mal-treatment. Future of Children: Protecting Children From Abuse and Neglect, 8, 39-53.

Gaudin, J. M., & Dubowitz, H. (1997). Family functioning in neglectful families: Recent research. In J. D. Berrick, R. P.

Barth, & N. Gilbert (Eds.), Child welfare research review (Vol. 2, pp. 28-62). New York: Columbia University Press.

CHILD MALTREATMENT / NOVEMBER 2004

Harrington, D., Dubowitz, H., Black, M. M., & Binder, A. (1995).

Harter, S., & Pike, R. (1984). The Pictorial Scale of Perceived Com-petence and Social Acceptance for Young Children. Child Devel-opment, 55, 1969-1982.

Kaufman Kantor, G., & Bluhm, B. (2002, August). Parental substance abuse and child maltreatment: Evaluation results from the NH IV-E waiver project. Paper presented at the Victimization of Children and Youth: An International Research Conference, Portsmouth, NH.

Kuhn, D. (1992). Cognitive development. In M. H. Bornstein & M.

E. Lamb (Eds.), Developmental psychology: An advanced textbook (3rd ed., pp. 211-272). Hillsdale, NJ: Lawrence Erlbaum.

Montelone (Ed.), Recognition of child abuse for the mandated reporter (2nd ed., pp. 105-118). St. Louis: G.W. Medical Publishing.

National Middle School Association. (1996). Research summary #1: Grade configuration. Westerville, OH: Author.

Powell, M. B., Roberts, K. P., Ceci, S. J., & Hembrooke, H. (1999).

The effects of repeated experience on children’s suggestibility. Developmental Psychology, 35, 1462-1477.