NISMART-3 is the third of a series of studies undertaken in response to the mandate of the 1984 Missing Children's Assistance Act (Pub. L. 98-473) that requires the Department of Justice Office of Juvenile Justice and Delinquency Prevention to conduct periodic national incidence studies to determine the number of children reported missing and the number recovered in a given year.
NISMART identifies five types of children who experienced episodes that could lead them to become missing. These include episodes in which children (1) were abducted by a family member, (2) were abducted by a nonfamily perpetrator, (3) had run away or were thrownaway, (4) were missing because they were lost or injured, and (5) were missing for benign reasons (i.e., misunderstandings).
All NISMART cycles have used a household survey of adult parents/caretakers to obtain data for all episode categories. Law enforcement surveys provided supplementary data on the most serious and rarest non-family abductions (called "stereotypical kidnappings").
NISMART-3 intended to replicate the NISMART-2 design that unified the results from all component studies to provide estimates of five categories of episode children (family abducted, non-family abducted, runaway/thrownaway, missing injured or lost, and missing for benign reasons) and two categories of missing children (caretaker missing and reported missing). However, the NISMART-3 household survey of youth and Juvenile Facilities Survey were not usable, due to the extremely low case yields.
The NISMART-3 adult household survey used 3 instruments: a paper household
eligibility screener form that was mailed to sampled households, a web survey, and a CATI survey. The adult web and CATI surveys asked about all the children who lived in the household during the past year; the
youth web and CATI surveys asked about the youth's individual experiences in the past year. The
web survey was an abbreviated version of the CATI interview. Because this was the first NISMART in the series to use a multi-mode approach--mail, web, and CATI--the design of the mailings, reminders, and the project website was incrementally refined during initial data collection to maximize response rates based on experiences with a series of small subsamples.
Law Enforcement Study (LES)
The NISMART-3 LES was designed to measure the national incidence of stereotypical kidnappings that occurred between October 1, 2010 and September 30, 2011 and closely replicates the methodology of the NISMART-2 LES. The definition of stereotypical kidnappings is limited to cases where children were abducted under especially ominous circumstances. This definition includes cases in which a child (age 0 to 17) was taken by a stranger or slight acquaintance, moved at least 20 feet or held for at least one hour, and one or more of the following serious circumstances applied: the child was kept overnight or longer, killed, taken 50 miles or more, held for ransom, or the perpetrator intended to keep the child permanently.
The LES obtained a sample of stereotypical kidnapping cases through a survey of all law enforcement agencies (n=4,644) located in a nationally representative sample of 433 counties.
For the initial phase of data collection, a mail survey was used to ascertain whether each agency in the sample of counties had investigated stereotypical kidnappings between October 1, 2010 and September 30, 2011. The mail survey reproduced the survey used in NISMART-2. It included two main sections, one that asked whether the agency had investigated any child abductions that met the definition of a stereotypical kidnapping and one that asked whether it had investigated any child homicides that qualified as a stereotypical kidnapping.
For the second phase of data collection effort extensive follow-up telephone interviews were conducted to capture key features of cases reported in the mail surveys, including information about victim-perpetrator relationships and characteristics of victims, perpetrators, crimes and investigations. Trained and experienced telephone interviewers conducted the interviews using a computer-assisted telephone interviewing system, WINCATI. The NISMART-3 LES telephone interview instrument largely replicated the paper-and-pencil instrument used for telephone interviews in the NISMART-2 LES.
Addresses were obtained from the United States Postal Service (USPS) through the Delivery Sequence File (DSF) which lists all addresses that receive mail deliveries. Prior to sampling the address frame was sorted and a sample of 130,000 addresses from all 50 states and the District of Columbia was selected with equal probabilities. The address sample was fielded in five cohorts. The first four cohorts used small subsamples, totaling 15,000 addresses, to pretest different versions of the mailed screener, reminders, and invitations to improve response rates; the last cohort comprised the remaining 115,000 addresses and received the final version of these materials.
Upon receiving completed screener questionnaires, eligible households were sent invitations to complete an online survey. Invited parents/caretakers who did not complete the online survey were interviewers called respondents to complete a computer-assisted telephone interview (CATI).
Youth surveys for NISMART-3 were created and administered, similar to NISMART-2, but not used in analysis due to low case yield.
Law Enforcement Survey
The LES obtained a sample of stereotypical kidnapping cases through a survey of all law enforcement agencies (n=4,644) located in a nationally representative sample of 433 counties. Law enforcement agencies were located in the sampled counties using the 2011 National Directory of Law Enforcement Administers (www.safetysource.com), which provided a database listing all of the municipal and county law enforcement agencies and state criminal investigation agencies in the United States by address, county and other details.
Agencies in the sample were sent a mail survey that asked whether the agency investigated any stereotypical kidnappings in their jurisdiction between October 1, 2010 and September 30, 2011.
When agencies reported such investigations, telephone interviews were conducted with investigating officers to obtain details of the episodes.
Household survey weights for NISMART-3 were assigned to account for both probability of selection and potential bias in the estimates. The goal of the raking adjustment was to inflate the sample data up to the dimension of the American Community Survey (ACS) 2013 population totals, providing estimates of statistics such as proportions and totals that would have been obtained if the entire population of U.S. Households had been surveyed. This adjustment also compensated for coverage issues in the sampling frame, to the extent possible. For additional information on weighting please see the Household Methodology Report included in this release.
Law Enforcement Survey (LES)
A weight was created for each agency and case to allow the responding agencies and case interviews to represent all law enforcement agencies in the United States. A set of 80 jackknife replicate weights was also created for each agency and case for estimating variances. For additional information on weighting please see the LES Technical Report included in this release.
Mode of Data Collection:
computer-assisted telephone interview (CATI),
Description of Variables:
Adult Household Survey
The adult household portion of NISMART3 data is comprised of 12 relational data files.
Part 0001 - Adult Case: [135 variables, 3603 cases] includes the number of household children, episode screener questions, episode screener confirmations and number of occurrences, respondent's characteristics, household demographics, household adult gender, and household own/rent.
Part 0002 - Adult Episode Screeners: [249 variables, 319 cases] includes time that the confirmed episode events occurred and responses to the episode.
Part 0003 - AI AC Child Roster: [108 variables, 6865 cases] includes child demographics and living situation, child's involvement in candidate episodes, child's parents' status and characteristics.
Part 0004 - AF Episode: [103 variables, 66 cases] includes episodes of family member abductions and abduction episode characteristics, such as duration of episode and month/year of episode.
Part 0005 - AF Episode Child: [145 variables, 105 cases] includes child-specific events of family abduction episodes, such as harm to child during the episode.
Part 0006 - AN Episode: [103 variables, 28 cases] includes nonfamily abduction episode characteristics.
Part 0007 - AN Episode Child: [174 variables, 29 cases] includes child-specific events of non-family abduction episodes, such as harm to child during the episode.
Part 0008 - AR Episode: [49 variables, 97 cases] includes runaway/thrownaway episode characteristics.
Part 0009 - AR Episode Child: [234 variables, 104 cases] includes child-specific events of runaway/thrownaway episodes, such as harm to child during the episode.
Part 0010 AG Episode: [37 variables, 245 cases] includes general missing episode characteristics.
Part 0011 - AG Episode Child: [99 variables, 258 cases] includes child-specific events of general missing episodes, such as harm to child during the episode.
Part 0012 - Household Analysis: [256 variables, 6865 cases] includes demographic variables, evaluative coding results, such as episode children, missing children, and reported episode/missing children; full sample child weights (crkwt) and replicate weights; adjusted child weights (acrkwt) for adult case measures, derived characteristics for nonresponse bias analyses.
Law Enforcement Survey (LES)
The LES consisted of 3 datasets; a case file, a victim file, and a perpetrator file. The datasets are comprised of information from one or more of the six categories described below.
Part 0013 - LES Cases [986 variables, 145 cases] includes all sections listed below and contains.
Part 0014 - LES Victim [803 variables, 98 cases] includes information about the case regarding the victim, primarily from section B, D, and E.
Part 0015 - LES Perpetrator [278 variables, 131 cases] includes information about the case regarding the perpetrator, primarily from sections C, and D.
Preliminary Screening: This section included questions on the stereotypical kidnapping criteria, case status, agency role in the investigation, and involvement of other agencies. Cases screened out as ineligible if they failed to meet the criteria for a stereotypical kidnapping or if the investigation was closed or suspended for all of the study timeframe, October 1, 2010 to September 30, 2011.
Section A - Screening Information: This section contained questions about victim-perpetrator relationships. Eligible cases had at least one victim-perpetrator pair in which the perpetrator was a stranger or slight acquaintance. The NISMART-3 LES included a new relationship category, online meeting.
Section B - Victim Information: This section included information on victim demographics, life changes or events, supervision, and problems such as drug use and criminal activity by either victims or parents at the time of the abduction. There were loops to record information for up to three child victims. Interviewers recorded information about additional child victims on paper.
Section C - Perpetrator Information: This section included information about perpetrator demographics, life circumstances at the time of the crime, prior offenses, and current status in the criminal justice system. This section included loops to record information for up to three perpetrators.
Section D - Crime Information: This portion of the interview included questions about the discovery and report of the crime; recovery of the victim; the site where the victim was last seen; initial contact between the victim and perpetrator(s); the movement, detainment, or concealment of the victim; maltreatment or injuries to the victim; criminal motive and plan; and characteristics of key sites in the abduction. In cases with multiple perpetrators, respondents were asked whether perpetrators' roles differed and, if so, how.
Section E - Investigation Information: This section gathered data on the investigative activities and tasks undertaken by other law enforcement agencies in cases where multiple law enforcement agencies were involved. This section included loops to record information for up to four additional agencies. Finally, this section asked respondents what evidence or leads were key to recovering the victim or identifying the perpetrator. For NISMART-3, an item was added to ask respondents whether technology such as cell phones or the Internet played a role in the investigation.
Screener response rate was 41% (41,484).
Interview response rate (of eligable screened respondents) was 37% (3,603).
The overall response rate equaled 15%, this incorporates all eligibility criteria across both screener and interview phases.
Law Enforcement Survey
The response rate for the mail survey was 86%. Interviews were completed for 91% of the targeted cases.
Extent of Processing: ICPSR data undergo a confidentiality review and are altered when necessary to limit the risk of
disclosure. ICPSR also routinely creates ready-to-go data files along with setups in the major
statistical software formats as well as standard codebooks to accompany the data. In addition to
these procedures, ICPSR performed the following processing steps for this data collection:
Performed consistency checks.
Created variable labels and/or value labels.
Standardized missing values.
Checked for undocumented or out-of-range codes.