A Comparison of Methods for Estimating Child Maltreatment Rates:

1
A Comparison of Methods for Estimating Child
Maltreatment Rates

• Evaluation Approaches for a Child Maltreatment
  Prevention Initiative

2

• Shayala Williams, MPH
• Katherine Rosanbalm, PhD
• Christina Christopoulos, PhD
• Kenneth A. Dodge, PhD
• http//www.childandfamilypolicy.duke.edu/

3
Background and Goals

• The Durham Family Initiative (DFI) began on
  January 1, 2002. It uses an ecological approach
  with an overarching goal of reducing the
  incidence of child maltreatment in Durham County,
  North Carolina by 50 within a decade.
• Official DSS reports of child maltreatment are
  widely believed to be biased and to underestimate
  the true prevalence of child maltreatment, so we
  sought supplementary methodologies to measure
  DFIs success.
• Two experimental methodologies, a neighborhood
  survey and a survey of professionals, were
  implemented to measure maltreatment rates from
  the perspectives of those who have close contact
  with children.
• The goal of this study was to compare these
  methodologies and present the advantages and
  disadvantages of each one.

4
The Neighborhood Survey

• The neighborhood survey was an in-person,
  door-to-door survey conducted between April and
  December of 2004.
• The survey design called for a stratified random
  sample based on income and population. All block
  groups in Durham County with 50 or more
  households were stratified into six clusters
  based on income level and population size, with
  the first cluster representing the most
  impoverished block groups and the sixth cluster
  representing the most affluent. The goal was to
  complete at least 25 interviews in each cluster.
• The survey contained sections intended to measure
  self- and neighbor- reported incidence of several
  types of negative and positive parental
  behaviors, including negative behaviors that
  could be classified as maltreatment.

5
The Neighborhood Survey Parenting Items

• Neighbor-Reported Parenting Items
• Respondents were asked to report on several
  parental behaviors based on their own
  observations of a neighboring mother of a 4-, 5-,
  or 6-year old child during the previous 30 days.
• Self-Reported Parenting Items
• The self-reported parenting items were
  self-administered and handed to the interviewer
  in a sealed envelope upon completion in order to
  maintain respondent anonymity.
• The youngest child in the household between the
  ages 1-10 was the focus of the self-reported
  parenting items. Only the responses for target
  children ages 4-6 are used in this analysis for
  comparison with the neighbor-reported items and
  the official rates for children ages 4-6 in
  Durham County.

6
The Neighborhood SurveyRespondent Characteristics
7
The Survey of Professionals

• The survey of professionals was conducted between
  November 2004 and January 2005.
• This mail and internet-based survey was conducted
  anonymously so that professionals could answer
  frankly without fear of any consequences.
• The study design called for a random sample of
  professionals in Durham County stratified by
  profession.
• Surveys were sent out to 1,686 professionals in
  Durham County who come into contact with children
  and families on a regular basis and are likely to
  have some experience with decisions about the
  detection and reporting of child maltreatment.
• Data were weighted so that each profession
  counted equally in analyses.

8
The Survey of ProfessionalsResponse Rate
9
Survey Results Maltreatment Items
10
Comparison of Baseline Rates Across Sources
11
Advantages/Limitations of Each Method
Method Advantages Disadvantages
Official Rates Data are readily available, least expensive surveillance method. Only a fraction of cases are reported to authorities, changes in social awareness or state guidelines may impact the rate of reports and substantiations.
Neighborhood Survey Self-Report High response rate, sampling methodology makes results generalizable to entire county, collected anonymously. Expensive and time consuming to conduct, must take care to assure respondents anonymity.
Neighborhood Survey Neighbor-Report High response rate, results are generalizable to county, neighbor reporter may be more honest than in self-reports. Expensive and time consuming to conduct.
Survey of Professionals Relatively inexpensive to conduct, anonymous. Low and inconsistent response rate across professions, only applicable to those children seen by professionals, does not provide a point prevalence.
12
Conclusions

• While just over one percent of children aged 4 to
  6 years old in Durham County received a
  substantiation for maltreatment by DSS, 55 of
  neighborhood survey respondents reported
  witnessing a nearby mother of a 4- to 6-year-old
  behave in a manner consistent with maltreatment,
  and 49 of respondents reported behaviors towards
  their own 4- to 6-year-old consistent with
  maltreatment. Professionals reported that over a
  third of children they encountered while working
  had been maltreated by a primary caregiver.
• Consistent with other studies, the self,
  neighbor, and professional reported rates of
  child maltreatment found in our surveys are many
  times higher than the official substantiated
  rates.
• Official rates underestimate the rate of child
  maltreatment in Durham County.
• Neighborhood and professional surveys may be
  useful supplemental data for research on the
  fluctuation of child maltreatment rates over time.

13
Future Directions

• The neighborhood and professional surveys will be
  repeated, allowing us to compare changes in rates
  over time among the methods. This will give us
  more information on the validity of each method.
• Collection of official statistics from the DSS
  Data Warehouse is ongoing and will be
  supplemented with
• Collection of information on all Emergency Room
  visits by children in Durham County with certain
  diagnostic codes.
• Collection of various child well-being indicators
  from the community such as infant mortality
  rates, percent of low birth weight infants, and
  percent of children involved in early
  intervention services.

14
For More Information

• Shayala Williams, MPH
• Statistician
• Email ShayalaW_at_duke.edu
• Office [phone]
• Center for Child and Family Policy
• Duke University
• Box 90539
• Durham, NC 27708-0539
• http//www.childandfamilypolicy.duke.edu/