Assessing child maltreatment prevention via administrative data systems: A case example of reproducibility
Abstract
Critical issues about scientific reproducibility have been raised about biomedical research,
including the reliability of data and analyses within a given study. The case example in this
article examined a reproducibility issue pertaining to the use of administrative data systems
for evaluation of child maltreatment (CM) prevention, making use of a prevention study
conducted over a decade ago that provided a unique opportunity. The place-randomization
study, which randomized counties to condition, found that community-wide implementation
of a parenting and family support intervention produced positive impact on
county-wide rates for substantiated CM cases and out-of-home placements, documented
through a state information system. The key consideration is whether and to what extent
the administrative record data re-examined retroactively a decade later for the original
study’s time period would yield comparable results to those based on data acquired at the
time of the study. The results indicated that despite small changes over time, the same
data patterns and statistical effects were reproducible for the two archival outcome variables.
For substantiated CM, the reproduced analyses reflected higher effect sizes and a
clear pattern of reduction as a function of intervention. For out-of-home placements, effect
sizes were quite comparable to the original ones, reflecting preventive impact. Overall, this
case study illustrated the verifiability of data reproducibility in the context of a population
outcome evaluation, which underscores the importance of reliable population-prevalence
measurement as an essential part of a comprehensive public health strategy aimed at the
prevention of CM.
© 2016 The Author
Description
item.page.type
Article
item.page.format
Keywords
Administrative records, child abuse and neglect, parenting, intervention, child maltreatment, Place randomized trial, prevention, reproducability
Citation
Prinz, R. J. (2017). Assessing child maltreatment prevention via administrative data systems: a case example of reproducibility. Child Abuse & Neglect, 64, 13-18.