MEMORANDUM
To: Rochelle Wilkie Martinez
From: Katrina Baum
Cc: Michael Rand, Jeremy K. Shimer, Kathryn Chandler, Jill DeVoe, Lynn Bauer
Date: December 24, 2008
Re: BJS Responses to OMB 2nd Passback for the 2009 School Crime Supplement
7. A 16 – the TITLE of A 16 should be “Publication Plans and Timetable” and should provide detail about the types of statistical tabulations and analyses planned, not just the names of publications. To the extent that the 2009 tabulation plans are similar to those for 2007, please provide the main table shells. The indicator data on the NCES website seems to use only a small subset of this information, so we want to understand how the other data will be tabulated and published. |
The following presents a revised section to be included in the SS.
The BJS and the NCES will be responsible for release of the data to the public (hereafter referred to as the “datafile”), the statistical analysis of the data, and the production of resultant web-based publications and tabulations. The datafile itself is released via the Inter-University Consortium for Political and Social Research (http://www.icpsr.umich.edu/) and includes a codebook, setup program in SAS language, text file of the raw data, as well as the datafile in SPSS, SAS, STATA data formats. An example of the SCS data release documentation and datasets can be found at ICPSR for the following years:
1989: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/09394.xml 1995: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/06739.xml 1999: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/03137.xml 2001: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/03477.xml 2003: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/04182.xml 2005: http://www.icpsr.umich.edu/cocoon/ICPSR/STUDY/04429.xml
SCS 2007 Datafile: The data file from the 2007 School Crime Supplement is scheduled to be similarly released via ICPSR in January 2009.
SCS 2007 web-based tabulations: NCES publishes approximately 20 web-based tables showing all results from the most recently released SCS. Web-based tables from the 2007 SCS will be released in the summer of 2009. Examples of these tables from the 2003 and 2005 SCS can be found at: http://nces.ed.gov/surveys/ssocs/ssocs_tables.asp Users must scroll to the bottom of the page and select either the SCS 2003 or SCS 2005 and their topic of interest.
SCS 2007 Publications: The following publications will be released using data from the SCS 2007 and a summary of each specific tabular presentation is provided below:
SCS 2009 Datafile: Interviewing for the 2009 supplement will be conducted during January through June 2009. Processing of the survey will take place between January and December 2009. Computer-based clerical editing and coding will be completed by June 2010 and the computer processing, editing, imputation, and weighting of the data will be completed by the end of November 2010. The Census Bureau will prepare a microdata user file with documentation which is scheduled to be sent to the Inter-University Consortium for Political and Social Research (ICPSR) by January 2011.
SCS 2009 web-based tabulations: NCES publishes approximately 20 web-based tables showing all results from the most recently released SCS. Web-based tables from the 2009 SCS will be released in the summer of 2011. Examples of these tables from the 2003 and 2005 SCS can be found at: http://nces.ed.gov/surveys/ssocs/ssocs_tables.asp Users must scroll to the bottom of the page and select either the SCS 2003 or SCS 2005 and their topic of interest.
SCS 2009 Publications: The following publications will be released using data from the SCS 2009 and a summary of each specific tabular presentation is provided below:
|
8.This answer is insufficient. b. Does NCES have a standard or policy on standard errors or minimum cell sizes? |
NCES does not have any specific standards or policies relating to standard errors and standards relevant to minimum cell size are discussed below. Each NCES team is responsible for determining thresholds appropriate to its survey design.
NCES practices the following when dealing with the sizes of standard errors for the SCS:
If the standard error associated with an estimate accounts for between 30 and 50 percent of its variance (calculated by dividing the standard error by the estimate), the estimate is noted with “! Interpret data with caution,” meaning the estimate is unstable. If the standard error associated with an estimate is greater than or equal to 50 percent of the estimate, the estimate is suppressed from the table with a “‡ Reporting standards not met.”
NCES also practices the following when dealing with minimum cell sizes:
In accordance with NCES’ standards pertaining to confidentiality, no cell with less than 3 unweighted cases is reported in NCES tabulations. Based on this, the minimum cell size for reporting estimates in a table is 3 or more cases, unweighted. If an estimate does not meet this criteria, it is reported as “‡.” For reference to confidentiality edits within surveys, refer to STANDARD 4-2-10 at http://nces.ed.gov/pubs2003/2003601.pdf
Both practices are followed when tabulating results from the supplement. |
8b. Making reasonable assumptions based on 2003 or 2005 results, are there any cross-tabulations that would have variances so large that they would not meet those standards or policies? |
Attached please find the report entitled: Student Victimization in U.S. Schools Results From the 2005 School Crime Supplement to the National Crime Victimization Survey. As you will see from tables B-1 through B-6, there are certain categories that readers are recommended to interpret with caution (“!”) or categories that are suppressed (“‡”) because the data do not meet reporting standards due to large variance or insufficient sample size. While NCES does not know the impact of the reduction in NCVS sample size on the SCS and cannot anticipate its impact, we do anticipate that categories currently in question can well be dealt with by collapsing categories (e.g. household income) or removing specific rows (other, non-Hispanic) to eliminate potential problems with reporting. |
16. It is misleading to place a PRA statement on the questionnaire if it is never read to or provided to the respondent in any way. For the CPS and similar collections, OMB has required that some of the information about burden, topic, and sponsor be included in core materials. Unless changes to incorporate this information into an advance letter, script or other appropriate location can be made for this round, we will provide a 1 year clearance and issue a terms of clearance that the materials must be revised to this end prior to the 2011 round of data collection. |
BJS proposes to add the following language as an introduction to be read in the CAPI instrument. Unfortunately, this language cannot be incorporated into the CAPI screen until February 1st. If this month delay is not acceptable, we can send a memorandum to field representatives and request they read this language when they begin the screen. If either of these scenarios are acceptable, we still request the full 3-year clearance.
“Now, I would like to ask some questions about any contacts you may have had with the police. We estimate the survey will take between 2 to 10 minutes depending on your responses. The law authorizes the collection of this data and requires us to keep all information about you and your household strictly confidential.”
A brochure describing the School Crime Supplement was sent with an advance letter to all households included in sample. A copy of this brochure was submitted with this original OMB package. Although the sponsor is not explicitly stated in this brochure, NCES is listed on the front of brochure as well as the Department of Education. We believe that NCES is implied as the sponsor. Topics covered in this brochure sent with the advance letter include:
If the information covered in this brochure is sufficient, we can instruct field representatives to have this brochure available to be read to respondents who may not have received an advance letter. |
21a. Please correct the statement in item 21 that says “FERPA regulations require that prior consent be given….” There is a specific exemption from the consent provisions for the Secretary of Education (and therefore NCES) for data from schools for “evaluations and assessments” (that the preamble specifically indicated included statistical activities) that OMB worked with NCES to clarify during the recent updates to the regulations. For clarification, please speak to Marilyn Seastrom. |
We misattributed the consent requirement to FERPA. We should have attributed it to PPRA. We will delete this reference from the response memo. From our NCES Standard 1-1-1 (http://nces.ed.gov/statprog/2002/std1_1.asp):
Surveys that involve interviewing students in elementary and secondary schools must adhere to the requirements of the Protection of Pupil Rights Act and related amendments (see 20 US Code Section 1232h and amendments included in Section 1061 of the No Child Left Behind Act).
Specifically, without written consent from a student's parent, questions may not be asked about the following: a. Political affiliations or beliefs of the student or the student's parent; b. Mental or psychological problems of the student or the student's family; c. Sex behavior or attitudes; d. Illegal, anti-social, self-incriminating, or demeaning behavior; e. Critical appraisals of other individuals with whom respondents have close family relationships; f. Legally recognized privileged or analogous relationships, such as those of lawyers, physicians, and ministers; g. Religious practices, affiliations, or beliefs of the student or the student's parent; or h. Income (other than that required by law to determine eligibility for participation or for receiving financial assistance under such a program). |
21b. Please elaborate on the point that “Information contained in the supplement is richer because of its link to the data in the larger NCVS.” While this in undoubtedly true, how specifically does NCES leverage this richness in its reports? Does it report on school victimization in the context of family victimization, for example? |
NCES has limited resources to put forth publications and tabulations from the SCS. Since January of 2007, NCES released almost 230 products representing 36 NCES surveys. The SCS data products and publications account for 7 of these products. While NCES does not report or plan to report data in the context of the wealth of information available from the larger NCVS, the data are prepared by NCES and made widely available for distribution for such secondary analysis. As such, NCES has not examined school victimization in the context of family victimization, per se. Title 1 of the Education Sciences and Reform Act mandates that NCES collect, report, analyze, and disseminate statistical data related to education in the U.S. In particular, it requires the examination of the characteristics and nature of criminal incidents at school, including but not limited to, the frequency and seriousness of violence affecting students, the relationship between victims and perpetrators, demographic characteristics of victims and perpetrators, and weapons used in incidents of victimization at school. Given this mandate in light of available resources, NCES must prioritize its report publication effort to focus on its requirement of making available statistical data regarding the state of school crime nationwide. |
|
The question originally posed was “If the NCVS were not able to collect this supplement in the future (i.e., after its redesign scheduled for implementation in 2013), what alternatives would be available to NCES?” We interpreted this question as “What alternatives would NCES have to distribute similar data?” The answer is that we would have no alternatives to collect the data we derive from this supplement. That is, the NCVS is the nation’s only source of victimization data. We might be able to propose or append student-based questions on specific school characteristics to another survey (e.g. National Household Education Survey), or derive student reports of various school contexts (security guard presence, drug availability, fear etc.) from past surveys (Education Longitudinal Study of 2002), but linking these data to the actual victimization data are where the strength of these data lie. |
| File Type | application/msword |
| Author | baumk |
| Last Modified By | Lisa Price-Grear |
| File Modified | 2008-12-31 |
| File Created | 2008-12-31 |