12.07.2015 Views

Fourth National Incidence Study of Child Abuse and Neglect (NIS–4)

Fourth National Incidence Study of Child Abuse and Neglect (NIS–4)

Fourth National Incidence Study of Child Abuse and Neglect (NIS–4)

SHOW MORE
SHOW LESS
  • No tags were found...

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

APPENDIX DWITHIN STUDY ANALYSES This appendix describes the analyses <strong>and</strong> presents the statistics for the within-studyanalyses <strong>of</strong> the NIS–4 results. This introduction discusses the computation <strong>and</strong> assessment <strong>of</strong>the t-statistics used to identify subgroup differences within the NIS–4. The tables that followprovide the specific results <strong>of</strong> all statistical tests that support the findings presented in the NIS–4Report to Congress.All within-study statistical differences were assessed by the t-statistic on ratemeasures. To address the fact that the NIS–4 data derive from a complex sample design(stratified <strong>and</strong> multi-stage), the computations were performed using WesVar (Westat, 2007).This program calculated the estimated rates (r 1 <strong>and</strong> r 2 ) <strong>and</strong> (using the replicate weights in theNIS–4 analysis data file) the st<strong>and</strong>ard error <strong>of</strong> these rates (SE r1 <strong>and</strong> SE r2 ). The t-statistic wasthen calculated as follows:√r 1 -r 2t =2SE 2 r1 + SE r2The analyses examined all subgroup differences in maltreatment rates in eachcategory <strong>of</strong> maltreatment <strong>and</strong> for each level <strong>of</strong> harm or injury resulting from maltreatment. Inorder to adjust for the higher likelihood <strong>of</strong> observing significant differences when conductingmultiple comparisons in this way, the significance <strong>of</strong> the resulting t-statistics was assessed usingthe Bonferroni critical value for t (Sankoh, Huque & Dubey, 1997). The appropriate significancelevels with their corresponding alpha levels (i.e., probability <strong>of</strong> observing the result by chance)are also presented for each series <strong>of</strong> tables in the next section.ReferencesSankoh, A.J., Huque, M.F., Dubey, S.D. (1997). Some comments on frequently used multipleendpoint adjustment methods in clinical trials. Statistics in Medicine, 16, 2529-2542.Westat, Inc. (2007). WesVar 4.3 User’s Guide. Rockville, MD. Available online athttp://www.westat.com/westat/statistical_s<strong>of</strong>tware/WesVar/pdf/WV_4-3_Manual.pdfD-i

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!