BJS NCVS ID Theft Supplement Qs and As 12-07

Responses from Michael Rand, BJS, December 26, 2007

1. Why is Census/BJS unable to develop prevalence estimates from the current set of NCVS core ID theft questions?

Producing prevalence estimates is only one of the purposes of the proposed supplement. The supplement will produce person level prevalence estimates of identity theft, but will also provide information about the nature and impact of identity theft that the short battery of questions on the ongoing survey cannot address.

2. Is the purpose of the supplement to develop prevalence estimates of ever having been a victim of ID theft, having been a victim within the past 2 years, or both? 

   1. If both (which appears to be the case from the questions), are there other sets of NCVS “ever” estimates?

One of the purposes of the supplement is to obtain a one year prevalence estimate of identity theft. This is being accomplished using a two step process that first asks respondents to provide information about any identity thefts discovered during the previous two years. Subsequent questions ask when within the two year period the thefts were discovered. Using this data we will be able to construct a one year prevalence estimate of identity theft.

This two stage process has a number of benefits. First obtaining two year victimization, and then following up with identifying which identity thefts were discovered during the previous year reduces telescoping. This is a technique used in a number of retrospective surveys (but not the ongoing NCVS.) The initial two year period was used (rather than a longer period or a lifetime experience with identity theft) because it is short enough to reduce long term memory problems, but long enough to produce enough cases to allow analyses of the characteristics of victims and identity theft episodes, since the data for two years can be used for these analyses.

3. What is BJS’s assessment of the inadequacies of the existing estimates from cited studies?  How does this study correct for those inadequacies?

   1. The FTC conducted an additional survey in 2006, not mentioned in the clearance package.  Results were released recently, which are attached.  Please include this survey in responding to question 3.

As Katrina Baum stated in her response on Friday, December 21, one of the major benefits of this supplement is its large sample size. The sample size for the NCVS identity theft supplement is about 82,000 people, compared to about 5,200 for the 2006 FTC survey. By building a supplement to an ongoing survey, it is possible to reach many more people more economically than can be done by a separate survey effort. As Dr. Baum stated, the FTC has provided funding for this survey because of its potential to produce more precise estimates than that agency’s previous efforts.

Moreover, RDD surveys, such as the FTC’s Identity Theft survey have issues related to sampling frame that address based surveys such as the NCVS don’t have, and generally have much lower response rates. In RDD surveys, people who do not have land line telephones are excluded from the sample, and people who cannot be reached by telephone will be noninterviews. Because the NCVS incorporates both personal visit and telephone interviews, noninterview rates are much lower than in RDD surveys. For the NCVS, the overall response rate was 78.3% (90.1 percent of eligible households and 86.1 percent of persons in interviewed households.) By contrast, the FTC 2006 survey achieved a response rate of 26% (Synovate report page 72.)

4. Since this is a one time study in 2008, indicate how long an approval is required (e.g., 12 months)?

Addressed in Katrina Baum’s response.

5. Please clarify whether the supplement is administered during the same interview as the main NCVS. 

   1. If so, how are respondents informed about the supplement and its associated burden? 

   2. The procedures indicate that “The NCVS interview is always collected for each eligible respondent before the first supplement question is asked.”  To clarify, does that mean that if one eligible HH member’s interview is outstanding, another HH member already interviewed who is selected for the supplement cannot be interviewed.

Addressed in Katrina Baum’s response. It should be noted that the procedures to inform respondents about the burden of this supplement are the same as those used in other supplements conducted for the NCVS, such as the School Crime Supplement and the Police and Public Contact Supplement. Respondents are also informed that the supplement is voluntary, and it is expected that there will be some respondents who will provide information for the main NCVS survey but who will decline to participate in the supplement.

6. Part B indicates that the sample universe is all persons 16 years of age and older but the procedures indicate that the crime screen question is for each member 12 years or older and that each household member will provide the information by self-response.  Does this mean that for individuals between 12 and 16, we have limited proxy data that is not followed-up by self report?

Addressed in Katrina Baum’s response.

7. What are BJS’s nonresponse bias analysis plans for the ID Theft supplement? 

Addressed in Katrina Baum’s response. It should be noted that the expected response rate for this supplement will work to minimize any biases from nonresponse. However, post collection response bias analysis will be conducted prior to release of any estimates from the supplement.

8. Why does the paragraph at the top of the questionnaire cite the BJS confidentiality statute (in addition to Title 13) when BJS never receives any confidential data?  Wouldn’t this be confusing to respondents?

Addressed in Katrina Baum’s response. The citation of both Census Bureau and Bureau of Justice Statistics confidentiality statutes is standard for all NCVS forms.

9. Please provide any written reports on the results of your cognitive or other testing of the interview questions.

1. Has the battery of questions on distress been used in NCVS before related to other types of crimes?  Please describe your experience with them as well as what input from health professionals you have received.  For example,

                                                               i.      Does anyone say “not at all distressing” when asked “how distressing” a crime was? Q27

                                                             ii.      Are people able to distinguish duration of distress as “a month or more?”  What is the significance of a month or more?  Q28

                                                            iii.      Where do the specific lists of symptoms (Q28 and Q 29) come from?

I have attached a copy of the Census Bureau report on cognitive testing for the supplement. The results of the cognitive testing informed the ultimate design of the questionnaire in that a number of questions were revised based on the testing.

Staff from BJS, the Office for Victims of Crime and the National Institute of Justice reviewed various batteries of questions used in public health and other social science related surveys. The battery that was developed was a synthesis of the various question sets used in other surveys because no set examined exactly met the requirements and nature of the harms that may be experienced by identity theft victims. As Dr. Baum stated, the set of trauma questions achieved in this synthesis was developed in consultation with Dr. Dean Kilpatrick of the Medical College of South Carolina and Dr. Kevin Becker of Harvard University.

As you can see from the cognitive testing results (report pages 63-70) the Census Bureau found that people understood the concepts being addressed and were able to provide reasonable answers to the questions. It appears that the primary issue with the questions is not comprehension, but whether people who have experienced more than one episode of identity theft can isolate their responses to the most serious episode. However, revisions to the questionnaire have addressed this issue to the extent possible, and it is BJS’s intention to address this issue in the analysis and discussion of the results of the survey.

i. In the cognitive testing, no respondents said that their victimization was “not at all distressing.” In developing questions it is necessary to account for a full range of possibilities, even if the response to a particular category is expected to be quite low.

ii. Addressed in Katrina Baum’s response.

iii. Addressed in Katrina Baum’s response.

BJS will be evaluating the responses to the trauma questions on the identity theft supplement to determine whether they can be added to the core NCVS incident report for crimes of violence and theft.

Responses from Katrina Baum, BJS, December 21, 2007

  
       
        1.      Why is Census/BJS unable to develop prevalence estimates from the current set of NCVS core ID theft questions?
       
        The current identity theft questions are only in the household-respondent portion of the questionnaire. They are not NCVS core questions, although we hope to secure more funding and have them there in the future. BJS has released two publications, Identity Theft, 2004 and Identity Theft, 2005. As you will see, we are only able to provide household-level prevalence. The concept of household is very difficult for the public to understand and individual-level estimates are what we want to collect beginning January 2008.
       
        2.      Is the purpose of the supplement to develop prevalence estimates of ever having been a victim of ID theft, having been a victim within the past 2 years, or both?
        
        No, the purposed is not to collect estimates of ever being victimized. The screener portion of the questionnaire will only collect data from those respondents who discovered identity theft during the 2-year reference period.
       
                a.      If both (which appears to be the case from the questions), are there other sets of NCVS ?œever??estimates?
       
                3.      What is BJS?™s assessment of the inadequacies of the existing estimates from cited studies?  How does this study correct for those inadequacies?
        
        The primary studies to data have been the FTC 2003 and 2005 studies that used random-digit dial methodology. The FTC agreed to partner with BJS because the NCVS has a much larger sample. Moreover, the existing studies do not have the breadth that the proposed supplement will cover. Each topical section has a header in the questionnaire, but some of the highlights that are not being collected elsewhere are emotional and physical response of the victim, law enforcement response, and breach notifications. This supplement has been jointly funded by several funding agencies because they each have a research interest that has not been addressed by the existing studies.
       
                a.      The FTC conducted an additional survey in 2006, not mentioned in the clearance package.  Results were released recently, which are attached.  Please include this survey in responding to question 3.
        
        BJS submitted the OMB package for this supplement in August 2007. The FTC 2005 study was not released until November 2007. The additional areas of emotional and physical response of the victim, law enforcement response, and breach notifications were not covered in the study released in November.
       
                4.      Since this is a one time study in 2008, indicate how long an approval is required (e.g., 12 months)?
        
        An approval for January 1 - June 30, 2007 (i.e. 6 months) is required.

                5.      Please clarify whether the supplement is administered during the same interview as the main NCVS.
        
        Yes, the supplement will be administered after the NCVS-1 and NCVS-2 are completed.
       
                a.      If so, how are respondents informed about the supplement and its associated burden?
        
        An introduction to the supplement will be read after the NCVS-2. The language of this introduction can be found on page 1 of the questionnaire. If respondents ask the field representative at the beginning of the overall NCVS interview or at the beginning of the supplement, they are instructed to inform the respondent that from time to time the Justice Department collects information on special topics like school crime or police public contact. For the next 6 months, there is a special topic collection on identity theft that will last, on average, 10 minutes.

                b.      The procedures indicate that ?œThe NCVS interview is always collected for each eligible respondent before the first supplement question is asked.??amp;nbsp; To clarify, does that mean that if one eligible HH member?™s interview is outstanding, another HH member already interviewed who is selected for the supplement cannot be interviewed.
        
        NO. Each respondent is treated individually. For example, the household respondent completes the household questionnaire (the only place presently where there are identity theft questions), then the NCVS-1, the NCVS-2 if any crimes were reported in NCVS-1, then finally the identity theft supplement (ITS). The field representative then moves to the next person in the household age 12 or older. This person completes a NCVS-1, NCVS-2 if applicable, and then a ITS if he or she is 16 or older. And so on through all household members.
        
        6.      Part B indicates that the sample universe is all persons 16 years of age and older but the procedures indicate that the crime screen question is for each member 12 years or older and that each household member will provide the information by self-response.  Does this mean that for individuals between 12 and 16, we have limited proxy data that is not followed-up by self report?
        
        The ITS will only be administered to household members who are 16 or older. The CAPI technology will automatically prompt the interviewer with the ITS if the respondent meets the age criterion. If the respondent is 15, the ITS will not load for this repondent. There will be a screen at the conclusion of the NCVS-1 (if no crimes are reported), or the NCVS-2 (if crimes are reported), that says the interview is complete. Proxy interviews are allowed for the NCVS1 and NCVS-2, but proxy interviews are not allowed for the ITS due to the complex nature of the crime.

                7.      What are BJS?™s nonresponse bias analysis plans for the ID Theft supplement?
        
        BJS will evaluate nonresponse and consult with the Census Bureau to develop a nonresponse bias analysis and how the estimates may be impacted.

                8.      Why does the paragraph at the top of the questionnaire cite the BJS confidentiality statute (in addition to Title 13) when BJS never receives any confidential data?  Wouldn?™t this be confusing to respondents?
        
        The confidentiality statute is cited to communicate to respondents that their responses will not be shared with anyone outside of the Census Bureau. In fact, BJS as the sponsor only works with public-use data to ensure the confidentiality of the respondent. Crime victimization is a very sensitive topic and this statute has been included in all NCVS data collection since the survey's inception.

                9.      Please provide any written reports on the results of your cognitive or other testing of the interview questions.
        
        The cognitive report may be obtained from Ms. Theresa DeMaio in the Center for Survey Methods and Research at the Census Bureau. She may be on leave this week, but her email address is [email protected] (there may be a middle initial j in the middle of her name).
       
                a.      Has the battery of questions on distress been used in NCVS before related to other types of crimes?  Please describe your experience with them as well as what input from health professionals you have received. 
        
        The battery of questions on distress have not been administered previously in the NCVS. These questions were designed in consulatation with Dr. Dean Kilpatrick from the Medical College of South Carolina and a trauma expert at Harvard (whose names escapes me now, I am on annual leave today). These questions were adapted from existing trauma questions that are administered in other surveys. Answers to individual questions are provided below.
        
        For example,
           i.      Does anyone say ?œnot at all distressing??when asked ?œhow distressing??a crime was? Q27
        
        Unfortunately, the text of this question did not come through entirely in the transmission. I am unable to answer it now.
       
          ii.      Are people able to distinguish duration of distress as ?œa month or more???amp;nbsp; What is the significance of a month or more?  Q28
        
        Yes, the respondents were able to distinguish the duration of stress of a month of more. Cognitive interviewers included this as a probe during the interviews and the experts at CSMR recommended that this month period was working cognitively with the respondent. The rationale for using a month or more was adapted from the DSM-4 criteria that requires a month or more to meet a PTSD diagnosis. We will not label this distress as PTSD, but this is the rationale as to why we are not asking about the emotional and physical impact for those respondents who did not experience distress for less than a month.


            iii.      Where do the specific lists of symptoms (Q28 and Q 29) come from?
       
        The symptoms come from the standardized questions that were used to design these questions and in consultation with the Office for Victims of Crime, and the two trauma experts cited above.