Transcript for "Conducting Your Research Using NIS-4" Video by the National Data Archive on Child Abuse and Neglect (NDACAN).

SLIDE 1
National Data Archive on Child Abuse and Neglect.

SLIDE 2
Welcome to this tutorial: Conducting Your Research Using NIS-4, from the National Data Archive on Child Abuse and Neglect.

SLIDE 3
In this tutorial, we use these acronyms: NIS are the National Incidence Studies of Child Abuse and Neglect. NIS-4 is the fourth study. NCANDS is the National Child Abuse and Neglect Data System. PUF is the NIS-4 Public Use File. CPS is Child Protective Services. NDACAN is the National Data Archive on Child Abuse and Neglect. And PSU is the Primary Sampling Unit.

SLIDE 4
In this tutorial you will learn what the NIS studies are, the purpose of NIS-4, and the scope of child maltreatment in the US. We will review the NIS-4 sample and design, and the study findings. We will then go over how to analyze the NIS-4 PUF. The software packages that can be used to analyze NIS-4 include WesDaX, WesVar, SUDAAN, SAS, R, and Stata. We will learn to use the NIS-4 full case and paired jackknife replicate weights, and will identify resources to help you with your NIS-4 research.

SLIDE 5
The congressionally mandated NIS serve as the nation's needs assessment on child abuse and neglect. NIS collect data from community professionals to estimate the incidence of child maltreatment. NIS provide information on the nature and severity of the maltreatment; characteristics of children, perpetrators, and families; and changes in the distribution of maltreatment over time. NIS-4 data were collected in 2005-2006, by Andrea Sedlak, Westat, the Children's Bureau, and the Office of Planning, Research and Evaluation in the Administration for Children and Families, US Department of Health and Human Services. Previous NIS cycles were conducted in 1981, 1988, and 1996. 
The NIS design assumes CPS investigated cases represent only the "tip of the iceberg" of actual child maltreatment incidence.  In 2006, NCANDS data reported 3.3 million CPS referrals were made alleging the maltreatment of 6 million children. 905,000 children were determined to be victims of abuse or neglect. Looking deeper, in the one-year study period from 2005 to 2006, NIS-4 found 3 million children were endangered, and of them 1.25 million were harmed, by child maltreatment.

SLIDE 6
The NIS-4 sample was designed to be nationally representative. Data were collected within 110 PSUs at the county level. "Sentinels", professionals at community agencies such as health services, school and daycare, law enforcement, housing, shelter and other agencies, prospectively completed forms on children believed to be maltreated as they were encountered during two focused three-month reference periods. CPS data were retrospectively collected for the same time frame.
Data were unduplicated and non-maltreated children were removed from the Public Use File (PUF) distributed by NDACAN. The NIS-4 data has 29,424 child cases. The PUF contains 12,694 cases of countable maltreated children. Analyses are done using weights to make the dataset nationally representative.

SLIDE 7
Harm standard maltreatment is defined as an act or omission that results in demonstrable harm to the child. The endangerment standard is more broadly defined, and encompasses the harm standard. Endangerment standard maltreatment also counts children who were not harmed if thought to be placed in a potentially harmful situation, or if a CPS investigation substantiated or indicated maltreatment. Endangerment standard also includes perpetrators other than parents for certain maltreatment subtypes. Chemically-dependent newborn children with no form of countable harm or endangerment standard maltreatment after birth do not meet the NIS countability criteria under the harm or endangerment standard. 
There are several categories of children in the NIS-4 data. The NIS-4 data include children submitted by sentinels and those investigated by CPS whose maltreatment fit the harm or endangerment standard. Additionally, chemically-dependent newborns are included in the NIS-4 PUF data, as a special sub-group of interest to the field. 
The NIS-4 study does, but the distributed PUF data do not, include children who were classified "not countable" but whose maltreatment was described to the study by a sentinel or in a CPS agency investigation. The PUF data do not contain every data item on the collection forms.

SLIDE 8
NIS employs a common definitional framework classifying children according to types of maltreatment as well as the severity of maltreatment. The typology includes 60  categories: 10 forms of sexual abuse, 6 forms of physical abuse, 8 forms of emotional abuse, 12 forms of physical neglect, 4 forms of educational neglect, 11 forms of emotional neglect, and 9 other maltreatment subtypes. For example, subtypes for physical neglect include "Unwarranted delay or failure to seek needed care" and "Inadequate supervision."

SLIDE 9
The NIS-4 Report to Congress published findings of an overall decrease in US maltreatment incidence. 1.25 million children (one in 58) were victims of harm standard maltreatment, demonstrating a decrease in maltreatment overall and in all abuse categories. This reflects a 32% decline in the US child maltreatment rate since NIS-3.
3 million children (one in 25) were victims of endangerment standard maltreatment, demonstrating no change in maltreatment rate overall, and decreases in abuse and all abuse subtype categories. However, emotional neglect more than doubled since NIS-3.
The report also cited strong and pervasive race differences; the influence of socioeconomic status, parent employment, family size, school attendance, and other factors; and policy issues around school reporting. For information on these findings, see the NIS-4 Report to Congress available from the child abuse and neglect Digital Library on the NDACAN website.
SLIDE 10
Use WesDaX online quickly and easily to analyze NIS-4. WesDaX automatically applies the NIS-4 weights correctly. In WesDax you can run cross tabulations, create tables for subsets of children, compute chi-square statistics, and export tabulated findings to Excel. Access NIS-4 WesDax at www.nis4.org/access_nis4.html

SLIDE 11
To conduct more in-depth analyses of NIS-4, order the dataset from the NDACAN website at www.ndacan.acf.hhs.gov. Click on: datasets, then NIS, and then the link for Dataset 147. 
From the NIS-4 NDACAN dataset page, you can: read The NIS-4 PUF Manual (User's Guide), read NIS publications in the NDACAN child abuse and neglect Digital Library, and browse the Codebook. 
NIS-4 includes variables on: countability, sources, home and family, child, perpetrator, population, weights, and maltreatment; and items from the CPS Screening Policies and Sentinel Definition Survey Supplemental Studies.
NIS datasets 1 - 3 are also available from NDACAN.

SLIDE 12
Before beginning NIS-4 analyses, read the READ_ME_FIRST.pdf that comes with your dataset and review the NIS-4 PUF Manual and Codebook available on the NDACAN website (and with your dataset). 
The PUF contains 1,844 variables on 12,694 child level cases of countable maltreated children (n = 12,408) and drug-affected newborns (n = 286). The PUF sample includes only the maltreated children. Use the Census denominator for noncontingent risk analyses or rates.
You MUST use the weights when working with NIS-4 data.
Because of the complex sample design and use of paired jackknife replicate weighting (JK2), users have been advised in the past to conduct analyses only with WesVar or SUDAAN (in particular, for significance tests or compute any statistic that relies on variance or standard error). Preliminary analyses (using logistic regression) demonstrate that R, SAS, or Stata (using the svr package) may also be used. The results among software packages may differ slightly due to differences in how the packages calculate standard error: some use the estimate based on the full data in the calculation of the standard error, whereas others use the average estimate over the replicate weights. The difference is negligible in tested cases.
You are encouraged to calculate known estimates from the NIS-4 Report to Congress to confirm correct use of the replicate weights. The PUF Manual provides step-by-step WesVar screenshot examples to learn by. 
Sample analysis instructions and sample syntax are available on our the NIS user support webpage at www.ndacan.acf.hhs.gov/user-support/user-support-nis.cfm
The NIS-4 PUF does not include the substantiation variables included in past NIS data files. See the NIS-4 PUF Manual for more information.

SLIDE 13
The weights in NIS-4 are designed to provide national estimates of the incidence of child abuse and neglect. With the complex sample design, children were selected for the study with unequal probability. There are design effects due to differential sampling, clustering, and stratification.
The base weight for each sample unit is the reciprocal of the probability for including that unit in the sample. Base weights were developed for the child, data form, sentinel, agency, and PSU. Sample weights were then developed to adjust for: PSU subsample selection probabilities, population mobility (due to natural disasters), nonresponse, multiplicity (through multiple reports or sources), and annualization. The adjusted base weights were used to calculate CHAWT, the full case weight (on each child level record).
The paired jackknife method was then used to provide replicate weights for computing variance for the complex multi-stage sample design. Using this method, repeated subsamples are drawn, and a statistic of interest calculated for each subsample. The method uses variability among the subsamples to estimate variance of the full sample statistic.
Any and all analyses must use CHAWT, the full case weight (on each record). Any significance tests or statistic that relies on variance or standard error must also use CHAWT1 to CHAWT62, the 62 replicate weights (on each record). The strata and PSU are not included in the PUF data. Many software programs now make it easy to insert these weights into your syntax as you run your analyses.

SLIDE 14
This concludes NDACAN's Conducting Your Research Using NIS-4 video tutorial.  Please email the Archive if you have any questions. We provide ongoing data user support. The email address is NDACANsupport@cornell.edu. Good luck with your NIS-4 research! Thank you.

SLIDE 15
The National Data Archive on Child Abuse and Neglect is a project of the Bronfenbrenner Center for Translational Research at Cornell University. Funding for NDACAN is provided by the Children's Bureau, An Office of the Administration for Children and Families.
4