Chtldilbuse & Ne&cf. Vol. 16. PP. 57-74. Pnnted in the U.S.A. All rights reserved.
1992 copynght
(X45-2134/92 $5.00 + [email protected] Q 1992 Pergamon Press pie
THE IMPACI’ OF CASE CHARACTERISTICS ON CHILD ABUSE REPORTING DECISIONS GAILL.ZELLMAN RAND, Santa Monica, CA
Abstract-Drawing distinctions on the basis of at least some case characteristics in making reporting decisions makes considerable sense. Using data from a national survey ofmandated reporters that included vignettes in which case and personal characteristics were systematically varied, this paper examines the impact of selected characteristics while controlling for others. The data reveal that respondents noticed and responded to case characteristics. Abuse-relevant judgments and reporting intentions varied, often substantially, as a function of case characteristics. Three case cham~te~stics-previous abuse, severity of abuse, and recantation-were powerful predictors of vignette outcomes. A history of previous abuse led to judgments of greater seriousness, a more salutary impact of a report, and greater likelihood of reporting. More severe abuse was more likely to be labelled as abuse, and was more likely to be reported. When the alleged victim retracted her accusation upon questioning by an authority figure, respondents were significantly less likely to intend a report. Child age, perpetrator intent and family socioeconomic status also influenced abuse-relevant judgments and reporting intentions. Respondents were more likely to intend a report when younger children, lazy or angry perpetrators and children from poorer families were portrayed. ?&jr U;or&-Case
characteristics, Reporting, Reporting intentions~ Reporting decisions.
INTRODUCTION MANDATED REPORTERS must weigh many factors when deciding whether to report suspected maltreatment that they encounter. ~h~a~te~stics of the child, the alleged perpetrator, and the circumstances surrounding a given case are potentially critical factors bearing on reporting decisions, as they may directly influence judgments about seriousness, use of the abuse label, and whether the law requires a report. These judgments in turn have been found to be closely related to reporting intentions (Zellman, 1990). Some case characteristics (e.g., visible injury) may also be important in reporting decisions because they provide professionals with “su~ci~nt evidence” that abuse or neglect has occurred. Professionals weigh lack of such evidence heavily in decisions not to report suspected abuse and neglect (Sat&bury & Campbell, 1985; Zellman, 1990a). Drawing distinctions on the basis of at least some case characteristics makes considerable sense both logically and empirically. For example, neglect may have different meanings and manifestations when a child is 4 versus 14 (Eckenrode, Powers, Doris, Munsch, & Bolger, 1988). Similarly, physical injuries inflicted by an enraged parent may provoke different reactions than would the same injuries inflicted by a parent who believes that physical punishment is an essential component of responsible child rearing. Moreover, increasingly overburThis research was funded by Grant No. 90XA- 12 I3 from the National Center on Child Abuse and Neglect, Administration for Children, Youth and Families, OHDS, DHHS. Received for publication March 12, 1990; final revision received November 2. 1990; accepted March 28, I99 1. Requests for reprints may be sent to Gail L. Zellman, RAND, I700 Main Street, P.O. Box 2 138, Santa Monica, CA 90406-2 138. 57
58
Gail L. Zellman
dened Child Protective Services (CPS) agencies appear to be doing more screening than in the past (U.S. Department of Health and Human Services, 1988) and at least some of this screening is done on the basis of case characteristics, e.g.. child age, type of abuse, relationship of child to alleged perpetrator. While such characteristics are generally used to determine how quickly a response must be made, in actual practice the lowest-priority cases may never receive attention as higher-priority cases continue to flow in (Zellman & Antler, 1990). Given this reality, some mandated reporters consider it to be both sensible and responsible to base reporting decisions on such characteristics, since it averts the potential costs of a report to the family when any benefit is highly unlikely (e.g., Zellman & Antler, 1990). A number of research studies indicate that potential reporters and other research subjects do consider case characteristics in making report-relevant judgments. For example, Giovannoni and Becerra ( 1979) found that a description of the injuries that resulted from parental behavior was a significant factor influencing professionals’ severity judgments. When a description of the consequences of the abuse was presented to professionals, most judged the act as more severe. In another study (Attias & Goodwin, 1985) recanting of an incest allegation led one-third of private practitioners to decide not to make a report to CPS. The familiarity of the potential reporter with the family was found to be a consideration in reporting decisions in another study (Morris, Johnson, & Clausen, 1985). Herzberger and Tennen ( 1988) found that college student respondents to their survey rated parental discipline as more severe and less appropriate when the child’s precipitating misbehavior was described than when it was not. More limited evidence suggests that parental intention to inflict harm or the forseeability of the injury that occurred as a result of parental action may also influence the attributions and judgments of reporters (Giovannoni & Beterra, 1979). Using data from the first National Incidence Study (U.S. Department of Health and Human Services, 198 l), Hampton and Newberger ( 1985) examined the effects of a range of case characteristics on the reporting behavior of hospital personnel. They found that hospital personnel appeared to consider family characteristics in making reporting decisions. Specifically, cases in which the child was white or relatively more affluent were less likely to be reported to CPS. These findings of apparent class and race bias, while disturbing, need to be examined in a context in which other case characteristics can be held constant, which was not possible in Hampton and Newberger’s investigation. Data from a national survey of mandated reporters that included vignettes in which case and personal characteristics were systematically varied enabled the author to examine the impact of selected case and personal characteristics while controlling for others. These data then were used to clarify the importance of personal and case characteristics on abuse-relevant judgments and reporting intentions among mandated child abuse reporters.
METHODS This paper presents data from a national survey of mandated reporters about their child abuse reporting behavior. For coordination with a fieldwork component of the project (Zellman & Bell, 1990), the professionals were clustered in 15 states. Using data from the 1983 county and city data books, the 50 states were divided into four strata, defined by two dimensions: large urban versus smaller rural states, and states with relatively rich, well-educated populations versus ones with poorer, less well-educated populations. Within each stratum, states were selected using a form of stratified random sampling (Chromy, 1979) that took into account 1984 per-capita reporting rates (American Humane Association, 1986) and whether reports must be made to the police (this variable was selected because of substantial variation
Child abuse reporting decisions
59
across states and because we believed it might influence the inclination to report). Probability of selection increased with state population (proportional to population raised to the 0.4 power, except for two large states that were sampled with certainty). The resulting sample included equal numbers of states with populations under 3.0 million, 3.0-9.5 million, and over 9.5 million. General and family practitioners, pediatricians, child psychiatrists, clinical psychologists, and social workers were sampled from directories of their various professional organizations. Those clinical psychologists and social workers who indicated that they did not see children were excluded from the sample, which eliminated about 20% of these professionals. Surveys were also mailed to principals of public schools and heads of child care centers sampled from commercial lists of each. Pediat~cians, psychologists, social workers, and school principals were ove~mpled because we believed that they were likely to see the largest numbers of children and have the most reporting experience. The survey was completed by 257 pediatricians, 2 10 psychologists, 230 social workers, and 267 principals. The forms were returned by 104 general and family practitioners, 103 child psychiatrists, and 126 child care providers. Different versions of the surveys were developed for physical and mental health care providers, school principals, and child care providers. Most of the differences involved slight wording changes that reflected respondents’ different work environments. To improve response rate, a follow-up postcard and a second copy of the survey were sent. Phone calls to nonrespondents followed. A total of 1,196 professionals responded to the survey, an overall response rate of 59%. Response rates were only 38% for general and family practitioners; the response rate for all other groups was at least 55%. Of the public school principals contacted, 69% responded to the survey. While these response rates are not high, they are comparable or superior to response rates obtained in similar surveys of professionals (e.g., 32%, James, Womack, & Stauss, 1978; 63%, Kalichman, Craig, & Follingstad, 1990; 60%, Saulsbury & Campbell, 1985; 61%, Saunders, 1988). Although we were unable to collect data about nonrespondents, it is reasonable to assume that professionals who chose to complete the survey were more interested in and perhaps more concerned about child abuse and child abuse reporting than were nonresponders. Differences in response rates by profession, combined with the data reported below, support this assumption. At the very beginning of the survey form, before respondents described their own reporting behavior and something about the nature of their professional work, each respondent read and responded to five vignettes, each of which brietiy described a case of possible abuse or neglect. The vignettes provided common stimuli across respondents and thus controlled to some extent for differences in the types of children the respondents interact with, and the incidents that they see. The vignettes also permitted the exploration of the independent contributions of case characteristics to intended reporting and to other decisions that bear on reporting intentions. The vignettes were preceded by a statement acknowledging that the level of information was obviously limited and that in real life respondents would no doubt attempt to collect additional information before making a reporting decision, but that we would appreciate their making the best decision possible on the basis of the information provided. After each vignette, respondents answered six questions about the incident described (see Appendix A for the six questions). The first five questions asked respondents to make abuserelevant judgments about: ( 1) seriousness of the incident; (2) whether the incident should be labelled “abuse” or “neglect,” (as approp~ate); (3) whether the law would require a report in this instance; and the likeliho~ that (4) the child and, (5) the rest of the family would benefit from a report. The sixth item asked respondents to indicate how likely they would be to report the incident if they encountered it. On the basis of open-ended pretesting with mandated
Gail L. Zellman Table 1. Vignette Generation Design Respondent Group
Vignette
Gender
Ptimary Care Physicians’
Factors
Age
Mental Health Provider?
Child Care Providers
Elementary Principals
Secondary Principals
X
Neglect No
F
5
F/M
4 or 8
3. Ingested pills
F/M
18months
4. Left alone/ dws
M
13
I.
medication
2. Latchkey
child
Perpetrator intent: lazy/poor/ incompetent Previous incidents: yes/ “o SES: clinic/office Prior relationship: yes/no Age: 4 or 8 Gender Previous incidents: yes/
X’
X
X
X
X
X
X
X
IlO
SES: clinic/office Perpetrator intent: retarded/upset with child gender Gender Previous incidents: yes/ “* Severity: mother selling drugs to kids/not selling Gender Race: black/white
X
X
Physical Abuse 5. Son hit with belt
M
6or 14
M
6 months
7. Teen beaten
F
14
8. l-lx”
F
14
F
24
6.
Infant slapped
Age: 6 or 14 Prior relationship: yes/no Severity: moderate/ severe Perpetrator intent: anger/learned SES: welfare/ middle class Ethnicity: Hispanic/white Prior relationship: yeslno SES: attorney/ carpenter Recants: yes/no
X
X
X
X
X
X
X
X
X
X
X
X
x
Sexual Abuse molested
9. Adultabu~d as a childd
SES: attorney/ carpenter Recants: yes/no Severity: rubbed breasts/exposed self/intercourse Risk: not rema~ed/two young stepchitdren Access: known address-lives nearby/left state -no known address
X
X
Child abuse reporting
decisions
61
Table 1. (Continued) Respondent
Vignette
Gender
Age
Factors
Primary Care Physicians”
Mental Health Provide&’
Child Care Providers
Group
Elementary Principals
Secondary Principals
Sexual Abuse (conlinued) 10. Boy molested
M
3 or 8
11. Gonorrhea
F
6
12. Father admits fondling
F
6or
14
Age 3 or 8 Perpetrator relationship: babysitter/ estranged father Race: black/white Race: black/white Prior relationship: yes/no Age:6or14 Child gender SES: accountant/ machinist/ accountant seen socially Perpetrator intent: father drunk/ child provocative
X
X
X
X
X
X
X
X
X
X
Note. SES = socioeconomic status. a The category includes family/general
practitioners and pediatricians. b The category includes child psychiatrists, clinical psychologists, and social workers. ’ An X indicates that the professional group identified in the column heading received the vignette described in that row. d This vignette has its own metric and is not included in many analyses.
reporters in a variety of professions, we determined that the first five items captured issues that professionals often consider in deciding whether or not to report suspected maltreatment, a hypothesis borne out in our data (Zellman, 1990). The sixth item-likelihood of reporting the case described-was the ultimate outcome variable, and measures the respondent’s behavioral intention to report. While behavioral intentions are obviously not the same as actual behavior, they have been found to be significant predictors of actual behavior in a number of studies across a broad range of behaviors (e.g., Ajzen & Fishbein, 1980; Sheppard, Hartwick & Warshaw, 1988). Each respondent received five vignettes selected from a set of 12 core vignettes that we devised. These core vignettes included four cases of possible neglect, three of physical abuse, and five of sexual abuse (see Appendix B for one version of each of the 12 core vignettes). We included only five vignettes per survey to limit respondent burden. Each core vignette itself varied several factors ( 12 in all) found to be important in the child abuse literature, e.g., severity, history of previous abuse, and family socioeconomic status. The number of factors that were varied within each core vignette ranged from two to five. (See Table 1 for presentation of these factors and their levels by core vignette.) Given the fairly large number of core vignettes ( 12) and the staggering number of possible combinations of vignette variations (in the millions), we had to be sure that each respondent got a unique subset and that presentation of variations was balanced across respondents. A series of computer programs used arithmetic sequences to get us substantially closer to achieving perfect balance in assignments than would have been possible with random assignment (Zellman & Bell, 1987). These programs provided each respondent with a unique subset of vignettes, appropriate to his or her profession, which varied the twelve factors in a fairly unobtrusive manner. Since two vignettes were very similar (#7 and #8), the program ensured that no respondent received both of them. (See Zellman & Bell, 1990 for further discussion of these programs.)
Gail L. Zellman
62
RESULTS In order to simplify the interpretation of the vignette analyses, responses to the vignettes were converted to a scale of 0 to 100, using a linear transformation to retain original scale properties. Higher numbers are associated with increased propensity to see an event as serious and reportable. On our 5-point scales we reassigned values so that 1 (e.g., extremely serious) = 100, 2 (e.g., very serious) = 75, 3 (e.g., somewhat serious) = 50, 4 (e.g., not very serious) = 25, and 5 (e.g., not at all serious) = 0. On the 4-point scales, values were reassigned so that 1 = 100, 2 = 66.66, 3 = 33.33, and 4 = 0. To capture the definitional subjectivity implicit in child abuse reporting (e.g., Herzberger & Tennen, 1988; Meriwether, 1986) we did not provide survey respondents with definitions of abuse or neglect. Instead, we used language similar to that found in the reporting laws. Respondents appeared to notice and respond to differences in vignettes. As shown in Table 2, abuse-relevant judgments varied, often substantially, as a function of selected case characteristics such as severity. Differences in vignette responses as a function of respondent profession did emerge, particularly on the sexual abuse vignettes, but they are not presented here because they would have detracted from this paper’s main points. The reader is referred to Zellman and Bell ( 1990) Chapter 7 for presentation of these data.
Predicting Vignette Outcomes We used analysis of variance to test for the effects of the factors varied on each vignette. Although employing analysis of variance with ordinal data is not its classic use, it represents a valid test of the hypothesis that a factor is unrelated to a given outcome. Separate analyses were run for each of the six outcome measures. Preliminary tests for two-way interactions revealed that very few were significant. Of all 2-way interactions, only 5% were statistically significant at a < .05, which is the percentage that would be expected by chance. Thus, we focus our analyses on main effects. While most ANOVAS were significant due to large sample size, the amount of variance accounted for (R2 values) is generally small, with more than 80% of R’s less than .20, a pattern similar to that found by Kalichman, Craig, & Follingstad ( 1990). Nevertheless, some of the relationships shown in Table 3 warrant discussion. As shown in Table 3, case and family characteristics were generally more important in predicting ratings of seriousness, use of the abuse label, and likelihood of reporting than they were in predicting ratings of the anticipated benefits of a report for the child or the rest of the family.
Eflticts
ofCase Characteristics
Three case characteristics-previous 3. Each consistently predicts variations
Previous abuse. A history of previous
abuse, severity, and recantation-stand in vignette outcomes across vignettes.
out in Table
abuse, which was varied on three of the four neglect vignettes, had a significant impact on all five abuse-relevant judgments and on behavioral intention in virtually every instance, as shown in Table 2. Vignettes in which an earlier, similar incident was noted were rated significantly more serious and were more likely to be described as neglect. The law was perceived to more consistently require a report in these cases, and a report was believed to have a more salutary impact on both the child and the rest of the family. In each vignette, previous abuse was a significant predictor of intention to report, with likelihood of reporting higher when abuse had occurred in the past. The most
Childabut reporting decisions
63
striking eikts of previous incidents appear on Vignette #3 (Table 2), in which a child ingests a prescription drug. In this case, the knowledge that this had occurred before apparently allowed respondents to rule out the reasonable hypothesis that the ingestion was an accident unlikely to occur again.
Severity ofabuse. Severity, defined in the physical and sexual abuse vignettes in terms of the injury sustained (e.g., Son Hit With Belt, Vignette #5, Table 2) or perpetrator behavior (Teen Molested, Vignette #8, Table 2) was a significant predictor of seriousness ratings, use of the abuse label, the law’s requirements, and likelihood of reporting. In each instance, the more severe condition was more likely to be rated serious, and to be labelled abuse. The law was perceived as more iikely to require a report in the serious condition and respondents were more likely to intend one. Severity was not a significant predictor of impact on the rest of the family in either vignette, but was a significant predictor of impact on the child in the physical abuse vignette, where a report was perceived as more likely to benefit a child in the serious condition. On the single neglect vignette in which severity was varied (#4, in which a teen is left alone by a drug-using mother who is described as also selling drugs in the “severe” condition), severity proved to be unimportant in predicting any vignette outcome. We suspect that the severity variation meant little in the context of a case that many would argue was at the margins of reportable neglect.
~~~~~~~~~~~.Whether or not an overheard allegation of physical or sexual abuse was later recanted by a teenager had a significant impact on ratings of seriousness, use of the abuse label, beliefs about what the law required, and likelihood of reporting (Vignettes #7 and #8, Table 2). In every case, an indication that the alleged victim had retracted her accusation when questioned by an authority figure led to ratings that substantially reduced the likelihood of reporting. Recanting had a much smaller effect on the likely benefits of a report, perhaps because respondents believed that even if the abuse did not occur, such an accusation indicated that the family was troubled and in need of help. Child age. While less powerful or consistent than the above factors, child age was a significant contributor to vignette outcomes, as shown in Table 2. In three of the four vignettes in which age was varied (Vignettes #S, #IO, # 12) incidents alleged to have happened to younger children were rated as more deserving of a report on all six outcome variables. In the neglect vignette in which child age was varied (#2, Table 2), leaving a child of four alone at home for long periods during the day was seen as more problematic than when the child was eight. Hitting a six-year-old with a belt (Vignette #5> was consistently viewed as more problematic than when a I4-year-old was hit. Such judgments appear to be institutionalized in some CPS agencies. Interviews in CPS agencies in six states during the field-based portion of this study revealed that in several agencies, incidents of alleged physical abuse in which a parent is the perpetrator and the victim is an adolescent are defined as out of scope and are not accepted n priori as reports (Zellman & Antler, 1990). On one of the two sexual abuse vignettes, the six-year-old variation was rated more prohlematic on each vignette outcome than when the same vignette portrayed a 14-year-old (Vignette #f 1). In the fourth vignette in which age was varied, however, the direction of the effect was reversed and weaker. In this vignette (#9), in which a boy is described as fondling other children as a result of his own sexual abuse, respondents were somewhat more likely to rate the abuse as serious, and to label the incident as abuse when the child was eight as opposed to three. There was no age effect on ratings of what the law required, on the predicted efficacy of a report, or on the Iikelihood of reporting.
64
Gail L. Zellman Table 2. Mean Vignette Outcomes
Seriousness
by Vignette Variations
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Family
Likelihood of Report
Neglect Vignettes I. No Medication Intent L=Y No money Child effectively resists Previous incidents Yes No 2. Latchkey Child SES Clinic Your practice Prior relationship Yes No Child age 4 8 Child Gender Male Female Previous Incidents Yes No 3. Ingested Pills SES Clinic Your office Intent Upset re child gender Retarded Chtld gender Male Female Previous incidents Yes No 4. Left Alone/Drugs Severity Mother using drugs Mother using and selling to teens Gender Male Female Race Black White Neglect Means
16
74 63
78 64 51
61 55 39
62 59 45
47 47 37
61 57 39
75 68
70 59
47 57
60 51
41 40
59 46
71 76
81 17
72 71
64 61
49 45
74 69
76 16
71 80
70 73
63 63
49 46
71 72
86 66
88 69
83 59
70 55
52 42
83 60
71 15
79 78
71 71
64 61
49 46
74 70
79 14
71 81
75 68
66 60
48 47
75 68
80 85
64 65
54 56
54 60
46 48
56 59
81 84
63 66
56 54
54 60
43 50
55 60
83 82
64 64
54 56
55 59
47 46
57 58
89 76
84 45
79 31
71 43
58 35
84 31
76 80
83 85
75 78
64 60
50 42
80 82
79 77
85 83
77 77
63 61
50 42
81 81
76 80 78
84 84 73
74 80 63
57 61 60
42 50 47
19 83 65
Physical Abuse Vignettes 5. Son Hit With Belt Child age 6 I4 Prior relationship Yes No Severitv Bruised thigh in past Observed welts on neck Intent Anger Learned SES Welfare Middle class
65
71 68
74 71
71 68
56 54
46 42
61
69 69
13 72
71 68
55 55
45 44
64 62
64 15
66 79
61 78
52 58
43 4s
54 13
70 69
74 71
71 69
55 55
4s 43
64 62
69 70
72 74
70 69
54 56
43 46
63 63
65
Child abuse reporting decisions Table 2. (Continued)
Seriousness
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Family
Likelihood of Report
Physical Abuse Vignettes (continued) 6. Infant Slapped Ethnicity Alvakdo Greenberg Prior relationship Yes No 7. Teen Beaten SES Prominent attorney Carpenter Recants Yes No Physical Abuse Means
73 75
74 71
60 64
60 64
50 46
61 57
72 16
71 75
63 61
61 63
46 50
51 62
79 82
11 78
16 79
56 65
38 48
72 79
73 89 75
65 85 73
66 89 70
54 28
40 46 47
65 87 65
Sexual Abuse Vignettes 8. Teen Molested SES Prominent attorney Carpenter Recants Yes No Severity Rubbed breasts Exposed himself Intercourse 9. Adult Molested as Child Access Father out of state Father in state Risk No other children Father remarried, stepchildren 10. Boy Molested Child age 3 8 Perpetrator relationship Male sitter Estranged father Race Black White 1 I. Gonorrhea Race Black White Prior relationship Yes No 12. Father Admits Fondling Child age 6 14 Child Gender Male Female SES Machinist Accountant Accountant seen socially Intent Drunk, lost control Child provocative Sexual Abuse Means
14 78
71 1-l
70 76
53 55
32 41
66 72
66 86
64 84
60 85
41 61
35 39
55 83
72 72 85
71 12 81
68 68 82
53 51 58
37 39 34
64 66 11
71 73
89 88
31 34
49 43
38 39
33 35
71 73
96 81
27 31
38 54
32 45
25 44
18 81
71 74
75 78
65 62
56 53
75 71
81 78
15 69
18 15
66 61
61 48
78 74
80 79
73 72
18 75
66 62
58 51
78 14
95 95
92 91
98 95
18 78
53 57
99 96
95 96
91 92
96 98
17 79
54 55
97 98
91 88
93 90
92 86
68 63
50 49
88 83
89 90
92 91
89 89
63 67
47 53
85 86
90 89 89
92 93 90
91 89 87
65 66 64
51 52 46
88 85 83
89 90 83
92 92 83
89 89 83
66 64 47
51 49 50
88 85 83
Note. All entries were converted from questionnaire propensity to see an event as serious and reportable.
responses to a scale of O-100. Higher numbers SES = socioeconomic status.
are associated
with increased
66
Gail L. Zellman Table 3. Vignette Outcomes as a Function of Vignette Variation (F Values)
Seriousness
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Familv
Likelihood of Reoort
Neglect Vignettes I. No Medication Intenta Prewous incidents R’
2. Latchkey Child SES Prior relationship Child age Child gender Previous incident R2 3. Ingested Pills SES Intent Child gender Previous incidents Rl 4. Left Alone/Drugs Severity Gender Race RI
62.22”* 27.X0*** .2l
28.21*** 14.99*** .I I
17.01*** 15.99*** .08
7.89*** 9.95*** .04
28.40*** 26.82*** .03
0.33 0.0 I I27.06*** I .69 7.58** .24
4.53+ 2.13 107.81*** 0.07 5. IX* .22
0.1 I I .62 88.18*** 0.02 7.07** .I9
I .64 0.04 33.5 I *** 1.38 4.65* .I0
2.48 I.12 15.35*** 0.98 0.32 .0X
4. I?’ 0.23 82.18*** 2.35 7.84*’ .I8
5.158 1.26 0.29 25.94*** .I4
0.10 1.07 0.00 224.85’** .49
0.65 0.70 0.16 204.97*** .47
2.34 2.24 1.30 64.68*** .25
0.17 3.13 0.07 34.52*** .I5
0.46 I .72 0.06 23 I .20*** .50
0.14 0.31 0.02 .Ol
0.33 0.0 I 1.35 .02
20.98*** 13.63*** .09
0.99 0.49 1.50 .04
0.62 0.1 I 4.25: .06
I .86 1.85 2.19 .07
0.18 0.00 0.90 .03
Physical Abuse Vignettes 5. Son Hit With Belt Child age Prior relationship severity Intent SES R2
6: Infant Slapped Ethnicity Prior relationship R’ 7. Teen Beaten SES Recants R2
5.80: 0.00 84.13*** 0.79 0.47 .09
5.23* 0.35 65.03*** 3.65 I.17 .08
0.47 2.68 .I0
0.81 2.22 .Ol
3.37 77.17-e .I5
13.60*** 101.15*** .20
2.55 I .44 90.86*** I .05 0.60 .I0
I .80 0.26 .Ol 2.13 87. Is*** .I6
2.18 0.10 11.57*** 0.06 0.85 .03 I .56 0.42 .o I 11.31 16.83*** .06
4.43’ 0.37 1.36 1.24 2.04 .02 I .03 I.64 .Ol
4.35* 0.68 97.95*** 1.56 0.03 .I0 I .28 2.32 .o I
I I .96*‘* 5.22’ .04
7.57** 81.21** .I6
Sexual Abuse Vignettes 8. Teen Molested SES Recants Severitp R’ 9. Adult Molested as Child Access Risk R2 IO. Boy Molested Child age Perpetrator relationship Race R2 I I Gonorrhea Race Prior relationship R’ 12. Father Admits Fondling Child age Child gender Intent SES’ R2
4.82’
7.54** 93.80*** 9.05*** .22
4.75’ 98.76’*’ 14.29*** .26
0.57 26.04;‘; 2.13 .I0
8.73** I.81 0.65 .03
5.02’ 105.53*** 9.73”’ .25
1.13 0.59 .oo
0.56 42.63*** .08
I.10 I I .54*** .03
4.57’ 3 I .39*** .08
0.08 8.14” .02
0.37 40.13*** .08
3.73 2.36 0.10 .o I
3.13 10.81** 0.57 .04
2.16 I .95 I .oo .02
0.98 3.15 2.88 .Ol
0.87 27.14”’ 6.88” .05
0.75 2.47 2.04 .02
0.21 0.27 .oo
0.4 I 0.50 .oo
3.77 1.25 .03
0.02 0.52 .oo
0.80 0.12 00
3.72 I .40 .03
4.9 I *** 4. IO*** I.16 0.3 I .03
0.28 4.73-8 0.32 I .65 .02
8.80** 0.73 0.13 3.39 .03
I I3.06*** 22.7 I*** .28
10.45” 2.05 2.02 0.39 .03
9.93** 0.04 0.03 I .63 .03
Note. SES = socioeconomic status. ’ Indicates 3 levels of specified variable (all other case characteristics *1,
1992 copynght
(X45-2134/92 $5.00 + [email protected] Q 1992 Pergamon Press pie
THE IMPACI’ OF CASE CHARACTERISTICS ON CHILD ABUSE REPORTING DECISIONS GAILL.ZELLMAN RAND, Santa Monica, CA
Abstract-Drawing distinctions on the basis of at least some case characteristics in making reporting decisions makes considerable sense. Using data from a national survey ofmandated reporters that included vignettes in which case and personal characteristics were systematically varied, this paper examines the impact of selected characteristics while controlling for others. The data reveal that respondents noticed and responded to case characteristics. Abuse-relevant judgments and reporting intentions varied, often substantially, as a function of case characteristics. Three case cham~te~stics-previous abuse, severity of abuse, and recantation-were powerful predictors of vignette outcomes. A history of previous abuse led to judgments of greater seriousness, a more salutary impact of a report, and greater likelihood of reporting. More severe abuse was more likely to be labelled as abuse, and was more likely to be reported. When the alleged victim retracted her accusation upon questioning by an authority figure, respondents were significantly less likely to intend a report. Child age, perpetrator intent and family socioeconomic status also influenced abuse-relevant judgments and reporting intentions. Respondents were more likely to intend a report when younger children, lazy or angry perpetrators and children from poorer families were portrayed. ?&jr U;or&-Case
characteristics, Reporting, Reporting intentions~ Reporting decisions.
INTRODUCTION MANDATED REPORTERS must weigh many factors when deciding whether to report suspected maltreatment that they encounter. ~h~a~te~stics of the child, the alleged perpetrator, and the circumstances surrounding a given case are potentially critical factors bearing on reporting decisions, as they may directly influence judgments about seriousness, use of the abuse label, and whether the law requires a report. These judgments in turn have been found to be closely related to reporting intentions (Zellman, 1990). Some case characteristics (e.g., visible injury) may also be important in reporting decisions because they provide professionals with “su~ci~nt evidence” that abuse or neglect has occurred. Professionals weigh lack of such evidence heavily in decisions not to report suspected abuse and neglect (Sat&bury & Campbell, 1985; Zellman, 1990a). Drawing distinctions on the basis of at least some case characteristics makes considerable sense both logically and empirically. For example, neglect may have different meanings and manifestations when a child is 4 versus 14 (Eckenrode, Powers, Doris, Munsch, & Bolger, 1988). Similarly, physical injuries inflicted by an enraged parent may provoke different reactions than would the same injuries inflicted by a parent who believes that physical punishment is an essential component of responsible child rearing. Moreover, increasingly overburThis research was funded by Grant No. 90XA- 12 I3 from the National Center on Child Abuse and Neglect, Administration for Children, Youth and Families, OHDS, DHHS. Received for publication March 12, 1990; final revision received November 2. 1990; accepted March 28, I99 1. Requests for reprints may be sent to Gail L. Zellman, RAND, I700 Main Street, P.O. Box 2 138, Santa Monica, CA 90406-2 138. 57
58
Gail L. Zellman
dened Child Protective Services (CPS) agencies appear to be doing more screening than in the past (U.S. Department of Health and Human Services, 1988) and at least some of this screening is done on the basis of case characteristics, e.g.. child age, type of abuse, relationship of child to alleged perpetrator. While such characteristics are generally used to determine how quickly a response must be made, in actual practice the lowest-priority cases may never receive attention as higher-priority cases continue to flow in (Zellman & Antler, 1990). Given this reality, some mandated reporters consider it to be both sensible and responsible to base reporting decisions on such characteristics, since it averts the potential costs of a report to the family when any benefit is highly unlikely (e.g., Zellman & Antler, 1990). A number of research studies indicate that potential reporters and other research subjects do consider case characteristics in making report-relevant judgments. For example, Giovannoni and Becerra ( 1979) found that a description of the injuries that resulted from parental behavior was a significant factor influencing professionals’ severity judgments. When a description of the consequences of the abuse was presented to professionals, most judged the act as more severe. In another study (Attias & Goodwin, 1985) recanting of an incest allegation led one-third of private practitioners to decide not to make a report to CPS. The familiarity of the potential reporter with the family was found to be a consideration in reporting decisions in another study (Morris, Johnson, & Clausen, 1985). Herzberger and Tennen ( 1988) found that college student respondents to their survey rated parental discipline as more severe and less appropriate when the child’s precipitating misbehavior was described than when it was not. More limited evidence suggests that parental intention to inflict harm or the forseeability of the injury that occurred as a result of parental action may also influence the attributions and judgments of reporters (Giovannoni & Beterra, 1979). Using data from the first National Incidence Study (U.S. Department of Health and Human Services, 198 l), Hampton and Newberger ( 1985) examined the effects of a range of case characteristics on the reporting behavior of hospital personnel. They found that hospital personnel appeared to consider family characteristics in making reporting decisions. Specifically, cases in which the child was white or relatively more affluent were less likely to be reported to CPS. These findings of apparent class and race bias, while disturbing, need to be examined in a context in which other case characteristics can be held constant, which was not possible in Hampton and Newberger’s investigation. Data from a national survey of mandated reporters that included vignettes in which case and personal characteristics were systematically varied enabled the author to examine the impact of selected case and personal characteristics while controlling for others. These data then were used to clarify the importance of personal and case characteristics on abuse-relevant judgments and reporting intentions among mandated child abuse reporters.
METHODS This paper presents data from a national survey of mandated reporters about their child abuse reporting behavior. For coordination with a fieldwork component of the project (Zellman & Bell, 1990), the professionals were clustered in 15 states. Using data from the 1983 county and city data books, the 50 states were divided into four strata, defined by two dimensions: large urban versus smaller rural states, and states with relatively rich, well-educated populations versus ones with poorer, less well-educated populations. Within each stratum, states were selected using a form of stratified random sampling (Chromy, 1979) that took into account 1984 per-capita reporting rates (American Humane Association, 1986) and whether reports must be made to the police (this variable was selected because of substantial variation
Child abuse reporting decisions
59
across states and because we believed it might influence the inclination to report). Probability of selection increased with state population (proportional to population raised to the 0.4 power, except for two large states that were sampled with certainty). The resulting sample included equal numbers of states with populations under 3.0 million, 3.0-9.5 million, and over 9.5 million. General and family practitioners, pediatricians, child psychiatrists, clinical psychologists, and social workers were sampled from directories of their various professional organizations. Those clinical psychologists and social workers who indicated that they did not see children were excluded from the sample, which eliminated about 20% of these professionals. Surveys were also mailed to principals of public schools and heads of child care centers sampled from commercial lists of each. Pediat~cians, psychologists, social workers, and school principals were ove~mpled because we believed that they were likely to see the largest numbers of children and have the most reporting experience. The survey was completed by 257 pediatricians, 2 10 psychologists, 230 social workers, and 267 principals. The forms were returned by 104 general and family practitioners, 103 child psychiatrists, and 126 child care providers. Different versions of the surveys were developed for physical and mental health care providers, school principals, and child care providers. Most of the differences involved slight wording changes that reflected respondents’ different work environments. To improve response rate, a follow-up postcard and a second copy of the survey were sent. Phone calls to nonrespondents followed. A total of 1,196 professionals responded to the survey, an overall response rate of 59%. Response rates were only 38% for general and family practitioners; the response rate for all other groups was at least 55%. Of the public school principals contacted, 69% responded to the survey. While these response rates are not high, they are comparable or superior to response rates obtained in similar surveys of professionals (e.g., 32%, James, Womack, & Stauss, 1978; 63%, Kalichman, Craig, & Follingstad, 1990; 60%, Saulsbury & Campbell, 1985; 61%, Saunders, 1988). Although we were unable to collect data about nonrespondents, it is reasonable to assume that professionals who chose to complete the survey were more interested in and perhaps more concerned about child abuse and child abuse reporting than were nonresponders. Differences in response rates by profession, combined with the data reported below, support this assumption. At the very beginning of the survey form, before respondents described their own reporting behavior and something about the nature of their professional work, each respondent read and responded to five vignettes, each of which brietiy described a case of possible abuse or neglect. The vignettes provided common stimuli across respondents and thus controlled to some extent for differences in the types of children the respondents interact with, and the incidents that they see. The vignettes also permitted the exploration of the independent contributions of case characteristics to intended reporting and to other decisions that bear on reporting intentions. The vignettes were preceded by a statement acknowledging that the level of information was obviously limited and that in real life respondents would no doubt attempt to collect additional information before making a reporting decision, but that we would appreciate their making the best decision possible on the basis of the information provided. After each vignette, respondents answered six questions about the incident described (see Appendix A for the six questions). The first five questions asked respondents to make abuserelevant judgments about: ( 1) seriousness of the incident; (2) whether the incident should be labelled “abuse” or “neglect,” (as approp~ate); (3) whether the law would require a report in this instance; and the likeliho~ that (4) the child and, (5) the rest of the family would benefit from a report. The sixth item asked respondents to indicate how likely they would be to report the incident if they encountered it. On the basis of open-ended pretesting with mandated
Gail L. Zellman Table 1. Vignette Generation Design Respondent Group
Vignette
Gender
Ptimary Care Physicians’
Factors
Age
Mental Health Provider?
Child Care Providers
Elementary Principals
Secondary Principals
X
Neglect No
F
5
F/M
4 or 8
3. Ingested pills
F/M
18months
4. Left alone/ dws
M
13
I.
medication
2. Latchkey
child
Perpetrator intent: lazy/poor/ incompetent Previous incidents: yes/ “o SES: clinic/office Prior relationship: yes/no Age: 4 or 8 Gender Previous incidents: yes/
X’
X
X
X
X
X
X
X
IlO
SES: clinic/office Perpetrator intent: retarded/upset with child gender Gender Previous incidents: yes/ “* Severity: mother selling drugs to kids/not selling Gender Race: black/white
X
X
Physical Abuse 5. Son hit with belt
M
6or 14
M
6 months
7. Teen beaten
F
14
8. l-lx”
F
14
F
24
6.
Infant slapped
Age: 6 or 14 Prior relationship: yes/no Severity: moderate/ severe Perpetrator intent: anger/learned SES: welfare/ middle class Ethnicity: Hispanic/white Prior relationship: yeslno SES: attorney/ carpenter Recants: yes/no
X
X
X
X
X
X
X
X
X
X
X
X
x
Sexual Abuse molested
9. Adultabu~d as a childd
SES: attorney/ carpenter Recants: yes/no Severity: rubbed breasts/exposed self/intercourse Risk: not rema~ed/two young stepchitdren Access: known address-lives nearby/left state -no known address
X
X
Child abuse reporting
decisions
61
Table 1. (Continued) Respondent
Vignette
Gender
Age
Factors
Primary Care Physicians”
Mental Health Provide&’
Child Care Providers
Group
Elementary Principals
Secondary Principals
Sexual Abuse (conlinued) 10. Boy molested
M
3 or 8
11. Gonorrhea
F
6
12. Father admits fondling
F
6or
14
Age 3 or 8 Perpetrator relationship: babysitter/ estranged father Race: black/white Race: black/white Prior relationship: yes/no Age:6or14 Child gender SES: accountant/ machinist/ accountant seen socially Perpetrator intent: father drunk/ child provocative
X
X
X
X
X
X
X
X
X
X
Note. SES = socioeconomic status. a The category includes family/general
practitioners and pediatricians. b The category includes child psychiatrists, clinical psychologists, and social workers. ’ An X indicates that the professional group identified in the column heading received the vignette described in that row. d This vignette has its own metric and is not included in many analyses.
reporters in a variety of professions, we determined that the first five items captured issues that professionals often consider in deciding whether or not to report suspected maltreatment, a hypothesis borne out in our data (Zellman, 1990). The sixth item-likelihood of reporting the case described-was the ultimate outcome variable, and measures the respondent’s behavioral intention to report. While behavioral intentions are obviously not the same as actual behavior, they have been found to be significant predictors of actual behavior in a number of studies across a broad range of behaviors (e.g., Ajzen & Fishbein, 1980; Sheppard, Hartwick & Warshaw, 1988). Each respondent received five vignettes selected from a set of 12 core vignettes that we devised. These core vignettes included four cases of possible neglect, three of physical abuse, and five of sexual abuse (see Appendix B for one version of each of the 12 core vignettes). We included only five vignettes per survey to limit respondent burden. Each core vignette itself varied several factors ( 12 in all) found to be important in the child abuse literature, e.g., severity, history of previous abuse, and family socioeconomic status. The number of factors that were varied within each core vignette ranged from two to five. (See Table 1 for presentation of these factors and their levels by core vignette.) Given the fairly large number of core vignettes ( 12) and the staggering number of possible combinations of vignette variations (in the millions), we had to be sure that each respondent got a unique subset and that presentation of variations was balanced across respondents. A series of computer programs used arithmetic sequences to get us substantially closer to achieving perfect balance in assignments than would have been possible with random assignment (Zellman & Bell, 1987). These programs provided each respondent with a unique subset of vignettes, appropriate to his or her profession, which varied the twelve factors in a fairly unobtrusive manner. Since two vignettes were very similar (#7 and #8), the program ensured that no respondent received both of them. (See Zellman & Bell, 1990 for further discussion of these programs.)
Gail L. Zellman
62
RESULTS In order to simplify the interpretation of the vignette analyses, responses to the vignettes were converted to a scale of 0 to 100, using a linear transformation to retain original scale properties. Higher numbers are associated with increased propensity to see an event as serious and reportable. On our 5-point scales we reassigned values so that 1 (e.g., extremely serious) = 100, 2 (e.g., very serious) = 75, 3 (e.g., somewhat serious) = 50, 4 (e.g., not very serious) = 25, and 5 (e.g., not at all serious) = 0. On the 4-point scales, values were reassigned so that 1 = 100, 2 = 66.66, 3 = 33.33, and 4 = 0. To capture the definitional subjectivity implicit in child abuse reporting (e.g., Herzberger & Tennen, 1988; Meriwether, 1986) we did not provide survey respondents with definitions of abuse or neglect. Instead, we used language similar to that found in the reporting laws. Respondents appeared to notice and respond to differences in vignettes. As shown in Table 2, abuse-relevant judgments varied, often substantially, as a function of selected case characteristics such as severity. Differences in vignette responses as a function of respondent profession did emerge, particularly on the sexual abuse vignettes, but they are not presented here because they would have detracted from this paper’s main points. The reader is referred to Zellman and Bell ( 1990) Chapter 7 for presentation of these data.
Predicting Vignette Outcomes We used analysis of variance to test for the effects of the factors varied on each vignette. Although employing analysis of variance with ordinal data is not its classic use, it represents a valid test of the hypothesis that a factor is unrelated to a given outcome. Separate analyses were run for each of the six outcome measures. Preliminary tests for two-way interactions revealed that very few were significant. Of all 2-way interactions, only 5% were statistically significant at a < .05, which is the percentage that would be expected by chance. Thus, we focus our analyses on main effects. While most ANOVAS were significant due to large sample size, the amount of variance accounted for (R2 values) is generally small, with more than 80% of R’s less than .20, a pattern similar to that found by Kalichman, Craig, & Follingstad ( 1990). Nevertheless, some of the relationships shown in Table 3 warrant discussion. As shown in Table 3, case and family characteristics were generally more important in predicting ratings of seriousness, use of the abuse label, and likelihood of reporting than they were in predicting ratings of the anticipated benefits of a report for the child or the rest of the family.
Eflticts
ofCase Characteristics
Three case characteristics-previous 3. Each consistently predicts variations
Previous abuse. A history of previous
abuse, severity, and recantation-stand in vignette outcomes across vignettes.
out in Table
abuse, which was varied on three of the four neglect vignettes, had a significant impact on all five abuse-relevant judgments and on behavioral intention in virtually every instance, as shown in Table 2. Vignettes in which an earlier, similar incident was noted were rated significantly more serious and were more likely to be described as neglect. The law was perceived to more consistently require a report in these cases, and a report was believed to have a more salutary impact on both the child and the rest of the family. In each vignette, previous abuse was a significant predictor of intention to report, with likelihood of reporting higher when abuse had occurred in the past. The most
Childabut reporting decisions
63
striking eikts of previous incidents appear on Vignette #3 (Table 2), in which a child ingests a prescription drug. In this case, the knowledge that this had occurred before apparently allowed respondents to rule out the reasonable hypothesis that the ingestion was an accident unlikely to occur again.
Severity ofabuse. Severity, defined in the physical and sexual abuse vignettes in terms of the injury sustained (e.g., Son Hit With Belt, Vignette #5, Table 2) or perpetrator behavior (Teen Molested, Vignette #8, Table 2) was a significant predictor of seriousness ratings, use of the abuse label, the law’s requirements, and likelihood of reporting. In each instance, the more severe condition was more likely to be rated serious, and to be labelled abuse. The law was perceived as more iikely to require a report in the serious condition and respondents were more likely to intend one. Severity was not a significant predictor of impact on the rest of the family in either vignette, but was a significant predictor of impact on the child in the physical abuse vignette, where a report was perceived as more likely to benefit a child in the serious condition. On the single neglect vignette in which severity was varied (#4, in which a teen is left alone by a drug-using mother who is described as also selling drugs in the “severe” condition), severity proved to be unimportant in predicting any vignette outcome. We suspect that the severity variation meant little in the context of a case that many would argue was at the margins of reportable neglect.
~~~~~~~~~~~.Whether or not an overheard allegation of physical or sexual abuse was later recanted by a teenager had a significant impact on ratings of seriousness, use of the abuse label, beliefs about what the law required, and likelihood of reporting (Vignettes #7 and #8, Table 2). In every case, an indication that the alleged victim had retracted her accusation when questioned by an authority figure led to ratings that substantially reduced the likelihood of reporting. Recanting had a much smaller effect on the likely benefits of a report, perhaps because respondents believed that even if the abuse did not occur, such an accusation indicated that the family was troubled and in need of help. Child age. While less powerful or consistent than the above factors, child age was a significant contributor to vignette outcomes, as shown in Table 2. In three of the four vignettes in which age was varied (Vignettes #S, #IO, # 12) incidents alleged to have happened to younger children were rated as more deserving of a report on all six outcome variables. In the neglect vignette in which child age was varied (#2, Table 2), leaving a child of four alone at home for long periods during the day was seen as more problematic than when the child was eight. Hitting a six-year-old with a belt (Vignette #5> was consistently viewed as more problematic than when a I4-year-old was hit. Such judgments appear to be institutionalized in some CPS agencies. Interviews in CPS agencies in six states during the field-based portion of this study revealed that in several agencies, incidents of alleged physical abuse in which a parent is the perpetrator and the victim is an adolescent are defined as out of scope and are not accepted n priori as reports (Zellman & Antler, 1990). On one of the two sexual abuse vignettes, the six-year-old variation was rated more prohlematic on each vignette outcome than when the same vignette portrayed a 14-year-old (Vignette #f 1). In the fourth vignette in which age was varied, however, the direction of the effect was reversed and weaker. In this vignette (#9), in which a boy is described as fondling other children as a result of his own sexual abuse, respondents were somewhat more likely to rate the abuse as serious, and to label the incident as abuse when the child was eight as opposed to three. There was no age effect on ratings of what the law required, on the predicted efficacy of a report, or on the Iikelihood of reporting.
64
Gail L. Zellman Table 2. Mean Vignette Outcomes
Seriousness
by Vignette Variations
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Family
Likelihood of Report
Neglect Vignettes I. No Medication Intent L=Y No money Child effectively resists Previous incidents Yes No 2. Latchkey Child SES Clinic Your practice Prior relationship Yes No Child age 4 8 Child Gender Male Female Previous Incidents Yes No 3. Ingested Pills SES Clinic Your office Intent Upset re child gender Retarded Chtld gender Male Female Previous incidents Yes No 4. Left Alone/Drugs Severity Mother using drugs Mother using and selling to teens Gender Male Female Race Black White Neglect Means
16
74 63
78 64 51
61 55 39
62 59 45
47 47 37
61 57 39
75 68
70 59
47 57
60 51
41 40
59 46
71 76
81 17
72 71
64 61
49 45
74 69
76 16
71 80
70 73
63 63
49 46
71 72
86 66
88 69
83 59
70 55
52 42
83 60
71 15
79 78
71 71
64 61
49 46
74 70
79 14
71 81
75 68
66 60
48 47
75 68
80 85
64 65
54 56
54 60
46 48
56 59
81 84
63 66
56 54
54 60
43 50
55 60
83 82
64 64
54 56
55 59
47 46
57 58
89 76
84 45
79 31
71 43
58 35
84 31
76 80
83 85
75 78
64 60
50 42
80 82
79 77
85 83
77 77
63 61
50 42
81 81
76 80 78
84 84 73
74 80 63
57 61 60
42 50 47
19 83 65
Physical Abuse Vignettes 5. Son Hit With Belt Child age 6 I4 Prior relationship Yes No Severitv Bruised thigh in past Observed welts on neck Intent Anger Learned SES Welfare Middle class
65
71 68
74 71
71 68
56 54
46 42
61
69 69
13 72
71 68
55 55
45 44
64 62
64 15
66 79
61 78
52 58
43 4s
54 13
70 69
74 71
71 69
55 55
4s 43
64 62
69 70
72 74
70 69
54 56
43 46
63 63
65
Child abuse reporting decisions Table 2. (Continued)
Seriousness
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Family
Likelihood of Report
Physical Abuse Vignettes (continued) 6. Infant Slapped Ethnicity Alvakdo Greenberg Prior relationship Yes No 7. Teen Beaten SES Prominent attorney Carpenter Recants Yes No Physical Abuse Means
73 75
74 71
60 64
60 64
50 46
61 57
72 16
71 75
63 61
61 63
46 50
51 62
79 82
11 78
16 79
56 65
38 48
72 79
73 89 75
65 85 73
66 89 70
54 28
40 46 47
65 87 65
Sexual Abuse Vignettes 8. Teen Molested SES Prominent attorney Carpenter Recants Yes No Severity Rubbed breasts Exposed himself Intercourse 9. Adult Molested as Child Access Father out of state Father in state Risk No other children Father remarried, stepchildren 10. Boy Molested Child age 3 8 Perpetrator relationship Male sitter Estranged father Race Black White 1 I. Gonorrhea Race Black White Prior relationship Yes No 12. Father Admits Fondling Child age 6 14 Child Gender Male Female SES Machinist Accountant Accountant seen socially Intent Drunk, lost control Child provocative Sexual Abuse Means
14 78
71 1-l
70 76
53 55
32 41
66 72
66 86
64 84
60 85
41 61
35 39
55 83
72 72 85
71 12 81
68 68 82
53 51 58
37 39 34
64 66 11
71 73
89 88
31 34
49 43
38 39
33 35
71 73
96 81
27 31
38 54
32 45
25 44
18 81
71 74
75 78
65 62
56 53
75 71
81 78
15 69
18 15
66 61
61 48
78 74
80 79
73 72
18 75
66 62
58 51
78 14
95 95
92 91
98 95
18 78
53 57
99 96
95 96
91 92
96 98
17 79
54 55
97 98
91 88
93 90
92 86
68 63
50 49
88 83
89 90
92 91
89 89
63 67
47 53
85 86
90 89 89
92 93 90
91 89 87
65 66 64
51 52 46
88 85 83
89 90 83
92 92 83
89 89 83
66 64 47
51 49 50
88 85 83
Note. All entries were converted from questionnaire propensity to see an event as serious and reportable.
responses to a scale of O-100. Higher numbers SES = socioeconomic status.
are associated
with increased
66
Gail L. Zellman Table 3. Vignette Outcomes as a Function of Vignette Variation (F Values)
Seriousness
Abuse?
Law Requires a Report
Report Benefits Child
Report Benefits Rest of Familv
Likelihood of Reoort
Neglect Vignettes I. No Medication Intenta Prewous incidents R’
2. Latchkey Child SES Prior relationship Child age Child gender Previous incident R2 3. Ingested Pills SES Intent Child gender Previous incidents Rl 4. Left Alone/Drugs Severity Gender Race RI
62.22”* 27.X0*** .2l
28.21*** 14.99*** .I I
17.01*** 15.99*** .08
7.89*** 9.95*** .04
28.40*** 26.82*** .03
0.33 0.0 I I27.06*** I .69 7.58** .24
4.53+ 2.13 107.81*** 0.07 5. IX* .22
0.1 I I .62 88.18*** 0.02 7.07** .I9
I .64 0.04 33.5 I *** 1.38 4.65* .I0
2.48 I.12 15.35*** 0.98 0.32 .0X
4. I?’ 0.23 82.18*** 2.35 7.84*’ .I8
5.158 1.26 0.29 25.94*** .I4
0.10 1.07 0.00 224.85’** .49
0.65 0.70 0.16 204.97*** .47
2.34 2.24 1.30 64.68*** .25
0.17 3.13 0.07 34.52*** .I5
0.46 I .72 0.06 23 I .20*** .50
0.14 0.31 0.02 .Ol
0.33 0.0 I 1.35 .02
20.98*** 13.63*** .09
0.99 0.49 1.50 .04
0.62 0.1 I 4.25: .06
I .86 1.85 2.19 .07
0.18 0.00 0.90 .03
Physical Abuse Vignettes 5. Son Hit With Belt Child age Prior relationship severity Intent SES R2
6: Infant Slapped Ethnicity Prior relationship R’ 7. Teen Beaten SES Recants R2
5.80: 0.00 84.13*** 0.79 0.47 .09
5.23* 0.35 65.03*** 3.65 I.17 .08
0.47 2.68 .I0
0.81 2.22 .Ol
3.37 77.17-e .I5
13.60*** 101.15*** .20
2.55 I .44 90.86*** I .05 0.60 .I0
I .80 0.26 .Ol 2.13 87. Is*** .I6
2.18 0.10 11.57*** 0.06 0.85 .03 I .56 0.42 .o I 11.31 16.83*** .06
4.43’ 0.37 1.36 1.24 2.04 .02 I .03 I.64 .Ol
4.35* 0.68 97.95*** 1.56 0.03 .I0 I .28 2.32 .o I
I I .96*‘* 5.22’ .04
7.57** 81.21** .I6
Sexual Abuse Vignettes 8. Teen Molested SES Recants Severitp R’ 9. Adult Molested as Child Access Risk R2 IO. Boy Molested Child age Perpetrator relationship Race R2 I I Gonorrhea Race Prior relationship R’ 12. Father Admits Fondling Child age Child gender Intent SES’ R2
4.82’
7.54** 93.80*** 9.05*** .22
4.75’ 98.76’*’ 14.29*** .26
0.57 26.04;‘; 2.13 .I0
8.73** I.81 0.65 .03
5.02’ 105.53*** 9.73”’ .25
1.13 0.59 .oo
0.56 42.63*** .08
I.10 I I .54*** .03
4.57’ 3 I .39*** .08
0.08 8.14” .02
0.37 40.13*** .08
3.73 2.36 0.10 .o I
3.13 10.81** 0.57 .04
2.16 I .95 I .oo .02
0.98 3.15 2.88 .Ol
0.87 27.14”’ 6.88” .05
0.75 2.47 2.04 .02
0.21 0.27 .oo
0.4 I 0.50 .oo
3.77 1.25 .03
0.02 0.52 .oo
0.80 0.12 00
3.72 I .40 .03
4.9 I *** 4. IO*** I.16 0.3 I .03
0.28 4.73-8 0.32 I .65 .02
8.80** 0.73 0.13 3.39 .03
I I3.06*** 22.7 I*** .28
10.45” 2.05 2.02 0.39 .03
9.93** 0.04 0.03 I .63 .03
Note. SES = socioeconomic status. ’ Indicates 3 levels of specified variable (all other case characteristics *1,
