U.S. Department of Justice Office of Justice Programs Bureau of Justice Statistics ************************************************************* Bureau of Justice Statistics Working Paper Series ************************************************************* Interviewing Conditions in the National Crime Victimization Survey, 1993-2013 ************************************************************* ************************************************************ ------------------------------------------------------------ This file is text only without graphics and many of the tables. A Zip archive of the tables in this report in spreadsheet format (.csv) and the full report including tables and graphics in .pdf format are available on BJS website at:http://www.bjs.gov/index.cfm?ty=pbdetail&iid=5725 ------------------------------------------------------------ ************************************************************** Shannan Catalano, Ph.D., BJS Statistician ************************************************************* ************************************************************* Bureau of Justice Statistics 810 Seventh Street, NW Washington, DC 20531 ************************************************************** ************************ NCJ 249682, August 2016 *********************** ************************************************************** ************ Abstract: ************ Interviewing Conditions in the National Crime Victimization Survey, 1993–2013, describes the circumstances under which interviews are conducted with persons in the National Crime Victimization Survey (NCVS) from 1993 to 2013. This BJS Working Paper focuses on two types of interview conditions: data collection mode and privacy. Data collection mode is how the interview is conducted--whether in person, during a face-to-face contact, or by telephone. Interview privacy describes whether a person is interviewed alone or in the presence of others during an in-person interview. The Working Paper describes the level and pattern of interviewing conditions in the NCVS over time, highlighting variations among select survey respondent characteristics. Findings cover the percentage of NCVS interviews conducted in person, the percentage of in-person interviews conducted privately, and how these conditions vary by characteristics of survey respondents. This BJS Working Paper provides information on the unweighted distribution of interviewing characteristics and the demographic characteristics of sample respondents who responded to NCVS interviews. In addition, weighted data are used to examine the association among interviewing conditions, crime rates, and household composition. ************************************************************ ****************** Table of Contents ****************** Abstract Introduction 1. Interviewing Conditions in the NCVS 1.1 Data collection mode and interview privacy 1.2 Changes in interviewing conditions over time 2. Data Collection Mode in the NCVS, 1993–2013 2.1 Respondent characteristics and data collection mode, 1993–2013 3. Interview Privacy in the NCVS, 1993–2013 3.1 Respondent characteristics and interview privacy, 1993–2013 4. Future Research Methodology ****************** List of Figures ***************** Figure 1. Average annual type of interview, data collection mode, and interview privacy in the NCVS, 2009–13 Figure 2. NCVS interviews, by type of interview and whether interview was conducted in person or by telephone, 2009–13 Figure 3. In-person interviews conducted with NCVS sample respondents, by whether interview was conducted under private or nonprivate interviewing conditions, 2009–13 Figure 4. Percent of NCVS interviews, by in-person or telephone interviewing conditions, 1993–2013 Figure 5. Percent of NCVS interviews conducted in person, by respondent’s sex, 1993–2013 Figure 6. Percent of NCVS interviews conducted in person, by respondent’s age, 1993–2013 Figure 7. Percent of NCVS interviews conducted in person, by respondent’s race and Hispanic origin, 1993–2013 Figure 8. Percent of NCVS interviews conducted in person, by respondent’s marital status, 1993–2013 Figure 9. Percent of NCVS interviews conducted in person, by household composition, 1993–2013 Figure 10. Percent of NCVS interviews conducted in person, by household income, 1993–2013 Figure 11. Proxy interviews conducted in the NCVS, 1993–2013 Figure 12. Percent of NCVS in-person interviews, by private or nonprivate interviewing conditions, 1993–2013 Figure 13. Percent of NCVS interviews conducted privately, by respondent’s sex, 1993–2013 Figure 14. Percent of NCVS interviews conducted privately, by respondent’s age, 1993–2013 Figure 15. Percent of NCVS interviews conducted privately, by respondent’s race and Hispanic origin, 1993–2013 Figure 16. Percent of NCVS interviews conducted privately, by respondent’s marital status, 1993–2013 Figure 17. Percent of NCVS interviews conducted privately, by household composition, 1993–2013 Figure 18. Percent of NCVS interviews conducted privately, by household income, 1993–2013 Figure 19. Rate of violent victimizations for single-person and multiperson households, by whether in-person interview was administered under private or nonprivate interviewing conditions, 2009–13 **************** List of Tables **************** Table 1.  NCVS interviews conducted in non-English language, 2011–2013 Table 2.  Rate of violent victimization per 1,000 persons age 12 or older, by in-person or telephone interviewing conditions, 2009–13 Table 3.  Rate of violent victimization, for in-person interview, by private or nonprivate interviewing conditions, 2009–13 ************************** List of Appendix Tables ************************** Appendix table 1. Estimates for figure 1: NCVS interviews, by type of interview and whether interview was conducted in person or by telephone, 2009–13 Appendix table 2. Estimates for figure 2: In-person interviews conducted with NCVS sample respondents, by whether interview was conducted under private or nonprivate interviewing conditions, 2009–13 Appendix table 3. Estimates for figure 4: Percent of NCVS interviews, by in-person or telephone interviewing conditions, 1993–2013 Appendix table 4. Estimates for figure 5: Percent of NCVS interviews conducted in person, by respondent’s sex, 1993–2013 Appendix table 5. Estimates for figure 6: Percent of NCVS interviews conducted in person, by respondent’s age, 1993–2013 Appendix table 6. Estimates for figure 7: Percent of NCVS interviews conducted in person, by respondent’s race and Hispanic origin, 1993–2013 Appendix table 7. Estimates for figure 8: Percent of NCVS interviews conducted in person, by respondent’s marital status, 1993–2013 Appendix table 8. Estimates for figure 9: Percent of NCVS interviews conducted in person, by household composition, 1993–2013 Appendix table 9. Estimates for figure 10: Percent of NCVS interviews conducted in person, by household income, 1993–2013 Appendix table 10. Respondent and household characteristics of NCVS sample members, by in-person or telephone interviewing conditions, 2009–13 Appendix table 11. Estimates for figure 11: Proxy interviews conducted in the NCVS, 1993–2013 Appendix table 12. Estimates for figure 12: Percent of NCVS in-person interviews, by private or nonprivate interviewing conditions, 1993–2013 Appendix table 13. Estimates for figure 13: Percent of NCVS interviews conducted privately, by respondent’s sex, 1993–2013 Appendix table 14. Estimates for figure 14: Percent of NCVS interviews conducted privately, by respondent’s age, 1993–2013 Appendix table 15. Estimates for figure 15: Percent of NCVS interviews conducted privately, by respondent’s race and Hispanic origin, 1993–2013 Appendix table 16. Estimates for figure 16: Percent of NCVS interviews conducted privately, by respondent’s marital status, 1993–2013 Appendix table 17. Estimates for figure 17: Percent of NCVS interviews conducted privately, by household composition, 1993–2013 Appendix table 18. Estimates for figure 18: Percent of NCVS interviews conducted privately, by household income, 1993–2013 Appendix table 19. Respondent and household characteristics of NCVS sample, by whether in-person interview was conducted under private or nonprivate interviewing conditions, 2009–13 Appendix table 20. Standard errors for table 2: Rate of violent victimization, by in-person or telephone interviewing conditions, 2009–13 Appendix table 21. Standard errors for table 3: Rate of violent victimization for in-person interview, by private or nonprivate interviewing conditions, 2009–13 Appendix table 22. Rate of violent victimization for single-person and multiperson households, by whether in-person interview was administered under private or nonprivate interviewing conditions, 2009–13 Appendix table 23. Standard errors for appendix table 22: Rate of violent victimization for single-person and multiperson households, by whether in-person interview was administered under private or nonprivate interviewing conditions, 2009–13 *********************************************************** ************** Introduction ************** This report is part of the Bureau of Justice Statistics’ (BJS) working paper series, which is released to inform interested parties of research and to encourage discussion on the topic. It focuses on two types of interviewing conditions: data collection mode and privacy. Interviewing conditions are the circumstances under which interviews are conducted with persons in the NCVS. This report describes the distribution of two interviewing conditions-data collection mode and interview privacy-in the NCVS. Data collection mode is how the interview is conducted--either in person during a face-to-face contact or by telephone. Privacy describes whether a person is interviewed alone or in the presence of others during an in-person interview. Interviewing conditions vary for self-completed and proxy interviews (figure 1). These interviewing conditions may also vary by the sample characteristics and may be affected by changes in survey administration over time. They are important to understand because they can affect the quality of information collected from persons in a survey and, consequently, the survey estimates. This report focuses on self-completed interviews. For the period 2009 to 2013, 42% of National Crime Victimization Survey (NCVS) self-completed interviews were conducted in person and 54% were conducted by telephone (figure 2). In some instances, if the household member could not complete the interview, a proxy respondent was selected to respond to questions in the place of that person (4%). Of those self-completed interviews that were conducted in person, 47% were conducted privately and 53% were conducted in a nonprivate setting with others present (figure 3). Privacy can only be determined by an interviewer during an in-person interview and not for telephone interviews.***Footnote 1 When the interview is conducted by telephone, although FRs are instructed to encourage respondents to complete the interview in a private setting, FRs are unable to verify whether the respondent is alone or if anyone else other than the respondent is present for any part of the interview.*** *********************************************************** *************************************** 1 Interviewing Conditions in the NCVS *************************************** 1.1 Data collection mode and interview privacy ************************************************ The NCVS is a nationally representative, address-based survey in which eligible persons age 12 or older living in a sampled household are asked to self-report about the number and characteristics of victimizations experienced during the previous 6 months. Persons may be interviewed up to seven times over a 3-year period. The NCVS is an interviewer-administered survey conducted by the U.S. Census Bureau for the Bureau of Justice Statistics. Interviewers, also known as field representatives, read the survey questions out loud and enter the person’s answer onto the data collection form. A person selected to participate in the NCVS is referred to as an eligible sample member. When a person agrees to participate in the NCVS and completes an interview, he or she is referred to as a survey respondent. Respondent is used to refer to survey respondents of the NCVS in the remainder of the report. NCVS interviews are conducted either in person or over the telephone. An FR’s first contact with a sample household is always attempted in person at the eligible sample member’s home. The first interview with a household may be conducted during this initial face-to-face contact or sample members may request a telephone interview at a later time. Subsequent interviews with the household may be conducted either in person or by telephone. NCVS FRs are trained to conduct all in-person interviews under private conditions. For the NCVS, a private interview means that only the person being interviewed and the FR were present, while a nonprivate interview means that other persons (usually other members of the family or household) were also present. Because interviews are conducted with all eligible members of a household, it is not always possible to interview all household members under private interviewing conditions. Information on the presence of others during an interview is determined by an FR when an in- person interview is conducted (figure 1). 1.2 Changes in interviewing conditions over time ************************************************** Several key aspects of NCVS survey administration have changed over time, potentially affecting interviewing conditions. One aspect of survey administration is whether information collected during an interview is recorded using a paper and pencil interviewing (PAPI) method or a computer-assisted interviewing (CAI) method. From 1993 to 2006, the NCVS used a combination of a paper survey (FRs wrote the survey respondent’s information directly onto paper forms) and computer-assisted telephone interviewing (CATI) (Census Bureau telephone interviewers conducted interviews from centralized telephone facilities). However, from 2006 to the present, the NCVS has used only CAI methods, in which FRs enter the responses directly into a computer during both in-person and telephone interviews. In an interviewer- administered survey like the NCVS that relies on in-person and telephone interviews under CAI conditions, the survey questionnaire is programmed into a laptop and the FR reads the questions as they are presented on the computer screen. Over the 1993 to 2013 period, the NCVS implemented several changes that affected interviewing conditions. Beginning in January 2006, a new sample was introduced to the NCVS based on the 2000 Decennial Census to account for shifts in population that occur over time. This affected the percentage of in-person interviews, as first interviews with new sample households are always attempted in person.***Footnote 2 For more information on methodological changes to the NCVS in 2006 and their effect on victimization estimates, see Criminal Victimization, 2006 (NCJ 219413. BJS web, December 2007).*** An additional change to the samples was implemented in 2010 and 2011 that potentially affected interviewing conditions: Between October 2010 and June 2011, sample cases that had been previously cut in the mid-2000s were reinstated. The reinstated sample cases amounted to about a 24% increase over the previous sample size, from about 8,500 households per month to 10,500. As in 2006, the introduction of this new sample was associated with an increase in first-time in-person interviews. A second change to NCVS survey administration involved the move in 2007 from centralized to decentralized telephone interviews. From 1993 through 2007, NCVS CATI interviews were initiated by Census Bureau interviewers located in a centralized call center. CATI from a centralized call center was discontinued for the NCVS in July 2007. Beginning in 2008, all NCVS CATI interviews were decentralized, meaning that, like in-person interviews, they are conducted by FRs working in the field. A third change that potentially affected interviewing conditions was refresher training of NCVS FRs. Beginning in August 2011, refresher training was conducted using an experimental split sample cluster design. Half of the FRs were trained beginning in 2011, while the second half of FRs were trained in 2012. During refresher training, special emphasis was placed on conducting NCVS interviews in private when possible. In addition, following refresher training, new FR performance metrics and enhanced field supervision of FRs were implemented in the NCVS. Refresher training, new performance metrics, and enhanced supervision may be associated with an increase in in-person interviews conducted privately. 2. Data Collection Mode in the NCVS, 1993–2013 *********************************************** From 1993 to 2013, telephone interviews consistently accounted for more than half of all completed NCVS interviews (figure 4). The percentage of in-person interviews during this period fluctuated between 21% and 43%. In part due to a shift away from centralized call centers, the percentage of NCVS in-person interviews increased in 2006, and 43% of interviews were conducted in person in 2013. About 58% of NCVS interviews were conducted by telephone in 2006, compared to 52% in 2013. 2.1 Respondent characteristics and data collection mode, 1993–2013 ***************************************************************** Although in-person interviewing in the NCVS increased from 1993 to 2013, the amount of the increase may have varied by the characteristics of the survey respondent. Because differences in interviewing conditions could result in differences in data quality and reported victimization rates, it is important to understand demographic differences in data collection mode over time. Sex ***** Following a slight decrease in the percentage of NCVS in-person interviews from 1993 to 1997, the percentage of in-person interviews generally increased at a similar pace for males and females from 1998 to 2013. Among both females and males, in-person interviews accounted for about 25% of interviews conducted in 1993, compared to about 45% in 2013 (figure 5). Age ***** From 1993 to 2000, the percentage of NCVS interviews conducted in person was generally stable for all age groups. The highest percentage was observed for persons ages 18 to 24 (32% to 36%), followed by persons ages 25 to 34 (28% to 30%) (figure 6). The percentages were nearly equal for those ages 12 to 17 and age 35 or older. Beginning in 2001 across all age groups, in-person interviewing accounted for an increasingly larger percentage of NCVS interviews. For example, from 2001 to 2013, the number of in-person interviews doubled among persons ages 12 to 17. Twenty-two percent of interviews were conducted in person with this age group in 2001, compared to 49% in 2013. During the same period the rate of in-person interviews increased 82% among persons age 35 or older, from 22% in 2001 to 42% in 2013. In-person interviews also increased among persons ages 18 to 24 and among those ages 25 to 34. Race and Hispanic origin ************************** From 1993 to 2013, white non-Hispanics consistently had the lowest rates of in-person interviews compared to black non-Hispanics and persons of other races (which includes American Indian, Alaska Native, Asian, and Pacific Islander) (figure 7). Although white non-Hispanics accounted for the smallest percentage of in-person interviews in 1993, in-person interviewing among this group nearly doubled between 1993 (18%) and 2013 (41%). Beginning in 2004, the percentage of interviews conducted in person among this group generally increased each year until 2011. From 2011 to 2013, the percentage of in-person interviews among whites remained stable at about 41%. In contrast, persons of Hispanic or Latino origin consistently had the highest percentage of in-person interviews, followed by blacks and African Americans and persons of other races. Although the highest percentage of in-person interviews was observed among Hispanics, a relatively smaller percentage increase was observed among this group from 1993 to 2013. Among Hispanics, in-person interviews decreased from 45% of all interviews in 1993 to 38% in 1995, and then fluctuated between 37% and 39% from 1996 to 2001. Beginning in 2002, in-person interviews among this group generally increased each year until 2013, when it reached 58% of interviews. Among blacks, following a period of general stability from 1993 to 2001, in-person interviewing generally increased each year, from 39% in 2004 to 53% in 2013. Among persons of other races, in-person interviews accounted for 29% to 34% of all interviews from 1993 to 2003. By 2013, in-person interviews had increased to account for 48% of NCVS interviews conducted with this group. Marital status **************** From 1993 to 2013, the highest percentage of in-person interviews was consistently observed among persons who were separated, followed by persons who were never married and those who were divorced or widowed (figure 8). From 1993 to 2001, a similar percentage of in-person interviews were conducted among persons who were never married and divorced or widowed. Beginning in 2002, in-person interviews conducted with persons who were never married began increasing at a faster rate than interviews with persons who were divorced or widowed. By 2013, a higher percentage of persons who were never married were interviewed in person than those who were divorced or widowed. From 1993 to 2005, the lowest percentage of in-person interviews was consistently observed for persons who were married. Beginning in 2006, in-person interviews of married persons generally increased each year, steadily converging with the percentage of in-person interviews observed for divorced or widowed persons. In 2013, the percentages of in-person interviews observed for married and divorced or widowed persons were about equal, accounting for 42% to 43% of all NCVS interviews conducted with these two groups. In-person interviews decreased slightly for persons in all marital status groups from 1993 to 1997, and remained generally stable from 1998 to 2000. Beginning in 2001, the percentage of in-person interviews conducted generally increased each year for all marital status groups. For those who were never married, in-person interviews increased from 29% of all interviews in 2001 to 51% in 2013. From 2001 to 2013, the percentage of NCVS interviews conducted in person also increased for persons who were separated and divorced or widowed. Household composition *********************** From 1993 to 2013, the highest percentage of in-person interviews was consistently observed for households composed of a single adult with children (figure 9). From 1993 to 2008, households composed of a single adult without children had the second largest percentage of in-person interviews. Beginning in 2007, the percentage of in-person interviews among households composed of a single adult without children stabilized, accounting for 43% to 45% of all their interviews until 2013. From 2007 to 2013, the percentage of in-person interviews among households composed of two or more adults with children steadily increased, from 38% in 2007 to 48% in 2013. From 2010 to 2013, a higher percentage of in-person interviews were conducted with households composed of two or more adults with children than with households composed of a single adult without children. In comparison, the percentages of in-person interviews conducted with households composed of two or more adults and two or more adults with children were about equal from 1993 to 2008. Beginning in 2008, in-person interviews increased at a faster rate for households composed of two or more adults with children than for two-adult households. By 2013, the two groups were no longer similar in their percentage of in-person interviews. Although in-person interviews increased for all household composition groups, the greatest percentage increase was observed for households composed of two or more adults and two or more adults with children. Among these households, the percentage of in-person interviews decreased slightly from 1993 to 1997 and then generally increased each year until 2006. After a slight decrease in 2007, the percentage of in-person interviews generally increased each year from 2008 to 2011 before stabilizing in 2012 and 2013. From 2001 to 2011, in-person interviews among two-person households increased, from 23% in 2001 to 42% in 2011. From 2011 to 2013, NCVS interviews conducted in person accounted for 42% of interviews among these households. During the same period, in-person interviews among households composed of two or more adults with children increased more than twofold, from 23% in 2001 to 48% in 2011. From 2011 to 2013, NCVS interviews conducted in person accounted for 48% of interviews among these households. In-person interviews conducted in households composed of a single adult without children decreased from 33% in 1993 to 27% in 1997. From 1998 to 2000, in-person interviews accounted for about 28% of all interviews conducted in these households, compared to 30% in 2001 and 45% in 2006. In-person interviews among households composed of a single adult with no children did not increase from 2007 to 2013, but fluctuated between 43% and 44%. Household income ******************* From 1993 to 2013, the lowest percentage of in-person interviews was consistently observed for households earning $50,000 or more annually while the highest percentage of in-person interviews was generally observed for households earning less than $24,999 annually (figure 10). Observed differences among household income categories were generally stable from 1993 to 2002, but began converging among some income category groups by 2003. For example, from 1993 to 2002, households earning $24,999 or less annually had a higher percentage of in-person interviews, compared to households of other income categories. From 2003 to 2005, similar annual percentages of in-person interviews were observed for households earning $24,999 or less and households with unknown annual income. Beginning in 2006, in-person interviews among households earning $24,999 or less annually generally increased each year and remained higher than households of other income categories through 2013. A similar pattern was observed among households whose income was unknown. From 1993 to 2005, these households had consistently higher percentages of in-person interviews than households earning $25,000 or more annually. However from 2006 to 2013, similar annual percentages of in-person interviews were observed among households with unknown income and households earning $25,000 to $49,999 annually. From 1993 to 2013, the greatest percentage increase was observed among households earning $50,000 or more annually. Among these households, in-person interviews increased nearly threefold, from 13% in 1993 to 38% in 2013. During the same period, in-person interviews among households earning $25,000 to $49,999 annually increased more than twofold, from 20% of all interviews conducted with these households in 1993 to 47% in 2013. The percentage of NCVS interviews conducted in person also increased among households earning less than $24,999 annually (up 54%). Compared to other household income groups, the smallest increase between 1993 and 2013 was observed among households whose income level was unknown (up 48%).***Footnote 3 Although more than half of NCVS interviews were conducted by telephone from 2009 to 2013, the type of interview varied by the survey respondent’s characteristics. Some persons were interviewed more frequently in person than by telephone. See appendix table 10 for more detailed information about demographic subgroups for 2009 to 2013.*** *********************************************************** ****************************** Proxy Interviews in the NCVS ****************************** The NCVS is a self-report survey in which respondents are asked directly about their experiences with criminal victimization. In some instances, a person in the household is allowed to answer questions for another household member in what is called a proxy interview. During data collection, a decision can be made to use proxy interviewing instead of direct interviewing for the following reasons: * A parent or guardian refuses to allow a child age 12 or 13 to be interviewed. In this case, any knowledgeable household member who is at least age 18 can be a proxy respondent for the child. * A household member is temporarily absent and will not return to the address until after the data collection period for the household has ended. * A household member is considered physically or mentally incapacitated. To qualify as physically or mentally incapacitated, the household member must have health or mental illness problems that are continuous throughout the interview period and make it impossible for the person to be interviewed. In a proxy interview, the interviewer asks a proxy respondent questions and the proxy respondent answers the questions in the place of the person for whom the interview is being conducted. The quality of data gathered from a proxy respondent is associated with the proxy’s knowledge of the event, the proxy’s relationship to the respondent, and the saliency of the event being reported. The percentage of proxy interviews conducted in the NCVS has remained generally stable over time, comprising 3% to 5% of all completed interviews from 1993 to 2013 (figure 11). Similar to self-report interviews, proxy interviews in the NCVS can be conducted in person or by telephone. For the period 2009 to 2013, 2% of all interviews conducted that were by proxy were in person and 3% were by telephone (appendix table 1). Proxy interviews are most common in larger households and among persons ages 12 to 17 and those age 65 or older (not shown). Proxy interviews are not allowed for non-English interviews in the absence of a suitable interpreter. Non-English Interviews in the NCVS, 2011–13 ********************************************* NCVS questionnaires are available in English and Spanish but interviews can be conducted in other languages if an interpreter is available. For all non-English interviews, field representatives (FRs) may use an interpreter if it is acceptable to the person being interviewed. An interpreter may be a family member, a neighbor of the person being interviewed, an official interpreter, or the FR if he or she is fluent in the language of the person being interviewed. If it is difficult to find a suitable interpreter, field supervisors may help FRs obtain assistance in completing a non-English interview. Information on interview language was added to the NCVS in 2011 but is not available in the public-use files collected before 2011. For the period 2011 to 2013, an average of about 6,700 non-English interviews (4.3% of all NCVS interviews) were conducted annually (table 1). About 85% of non-English NCVS interviews were conducted in Spanish and 5% were conducted in Chinese, while Vietnamese (2%), Russian (1%), and Korean (1%) accounted for smaller percentages of non-English interviews. Six percent of non-English interviews were conducted in other languages. *********************************************************** 3. Interview Privacy in the NCVS, 1993–2013 ********************************************** NCVS interviewing protocols require FRs to interview respondents privately when possible to ensure that respondents are able to provide what is often personal and sensitive information about their experiences. However, a private interview is not always feasible, such as when a person refuses to be interviewed privately, when multiple persons reside in a small space such as a studio apartment, or when others (such as the parent of a 12-year-old child) refuse to allow the respondent to be interviewed privately. In other cases, a parent or another person may be caring for a young child or an older relative, or may have other caregiver responsibilities. From 1993 to 2011, 41% to 45% of in-person interviews were conducted privately (figure 12). The percentage increased in 2011, and in 2012 and 2013 the percentages conducted under private and nonprivate interviewing conditions were about equal. The increase in interviews conducted in private may be attributed to NCVS refresher training, new FR performance metrics, and enhanced field supervision of FRs implemented in the NCVS beginning in 2011. In 2013, 51% of in-person interviews were conducted privately with the respondent and 49% were conducted under nonprivate interviewing conditions, with the person being interviewed in the presence of others. 3.1 Respondent characteristics and interview privacy, 1993–2013 **************************************************************** Overall interview privacy was generally stable from 1993 until 2006. Beginning in 2007, the percentage of in-person interviews conducted privately generally increased each year until 2013. In addition, privacy during NCVS interviews varied over time according to the characteristics of the survey respondent. Sex ***** From 1993 to 2010, 41% to 44% of NCVS interviews with males and females were conducted privately (figure 13). Private interviews increased for males and females between 2010 and 2013. Among males, 43% of interviews were conducted private in 2010, compared to 52% in 2013. Similarly among females, private interviews increased 18% during the same period (from 42% in 2010 to 50% in 2013). Age ***** From 1993 to 2013, the highest percentage of private interviews was observed for persons age 35 or older, followed by persons ages 18 to 24 and those ages 25 to 34 (figure 14). Private interviews accounted for similar percentages of in-person interviews conducted with persons ages 18 to 24 and those ages 25 to 34. In comparison, the lowest percentage of private interviews was consistently observed for persons ages 12 to 17. Among all age groups examined, the percentage of in-person NCVS interviews conducted privately was generally stable from 1993 to 2009, and generally increased each year among all age groups from 2010 to 2013. The largest increase in private interviews was observed among persons ages 12 to 17, from 16% of all interviews in 2008 to 26% in 2013. Private in-person interviews also increased among persons ages 18 to 24 (from 39% in 2008 to 49% in 2013) and among those ages 25 to 34 (from 42% in 2008 to 51% in 2013). Private interviews also increased among persons age 35 or older. Between 2008 and 2010, the percentage of NCVS interviews conducted privately with this group increased, from 46% of all interviews in 2008 to 54% in 2013. Race and Hispanic origin ************************* From 1993 to 2013, black and white survey respondents had the highest percentage of private interviews (figure 15). During the same period, persons of other races had a lower percentage of private interviews than whites or blacks but a higher percentage than Hispanics. From 1993 to 2013, the lowest percentage of in- person interviews was consistently observed for Hispanic survey respondents. From 1993 to 2011, the percentage of private interviews among whites was generally stable, fluctuating from 42% to 47% of all in-person interviews conducted with this group. However, private interviews increased among whites, from 47% in 2010 to 53% in 2013. The percentage of privately conducted interviews among blacks increased in 1997 to be similar to whites. From 1997 to 2001, blacks and whites generally had similar percentages of privately conducted interviews annually. The percentage of private interviews among blacks increased from 43% in 1993 to 49% in 2002. From 2002 to 2010, private interviews among blacks fluctuated, accounting for 46% to 49% of all in-person interviews conducted with this group. From 2011 to 2013, private interviews among blacks increased from 52% of all in-person interviews in 2011 to 58% in 2013. Private interviews also increased among persons of other races and Hispanics. For persons of other races, the percentage of private interviews increased from 31% in 1993 to 35% in 1994. From 1994 to 2001, private interviews among persons of other races accounted for about 33% of all in-person interviews conducted with this group. Between 2001 and 2013, private interviews among persons of other races increased from 33% in 2001 to 49% in 2013. For Hispanics, private interviews increased from 24% in 1993 to 29% in 1995. From 1996 to 2000, private interviews among Hispanics ranged from 25% to 27% of in-person interviews conducted with this group. However, between 2000 and 2005, private interviews among Hispanics generally increased each year, with private interviews accounting for 29% of all interviews in 2001 and 32% in 2005. From 2005 to 2011, 29% to 34% of all in-person interviews with Hispanics were conducted privately, increasing to 39% in 2012 and 42% in 2013. Marital status ***************** From 1993 to 2013, the highest percentage of private interviews was observed for survey respondents who were divorced or widowed, followed by persons who were separated and those who never married. The lowest percentage of private interviews was consistently observed for persons who were married (figure 16). From 1993 to 2011, the percentage of private interviews was generally stable among persons who never married and those who were married. During the same period, 41% to 44% of all interviews for persons who never married were conducted privately, which increased to 51% in 2013. For married persons, 30% to 37% of all interviews were private from 1993 to 2011. This increased to 43% in 2013. Among persons who were separated, the percentage of in-person interviews conducted privately increased steadily from 1993 to 2013, from 50% in 1993 to 55% in 2001 and 59% in 2009. In 2013, 62% of in-person interviews with separated persons were conducted under private interviewing conditions. From 1993 to 2011, 63% to 68% of NCVS interviews were conducted privately among persons who were divorced or widowed. The percentage of private interviews among this group increased to about 71% in 2012 and 2013. Household composition ********************** From 1993 to 2013, the lowest percentage of private interviewing was consistently observed for households composed of two or more adults with children (figure 17). Among households composed of two or more adults with children, private interviews accounted for 23% to 27% of in-person interviews from 1993 to 2005, before increasing to 30% in 2009. By 2013, private interviews accounted for 38% of in-person interviews conducted among these households. Lower percentages of private interviewing were also observed for households composed of a single adult with children. From 1993 to 1998, the percentage of privately conducted interviews observed for this group was generally stable, ranging from 29-31%. Between 1999 and 2010, the percentage of privately conducted interviews conducted with households composed of a single adult with children ranged from 35% to 42%. Beginning in 2010, private interviewing generally increased each year until 2013 when the percentage of privately conducted interviews among these households increased to 48%. From 1993 to 2013, the highest percentage of private interviews was observed for survey respondents who resided in households composed of a single adult without children. During this period, about 90% of in-person interviews among these households were conducted under private interviewing conditions, a substantially higher percentage than observed for any other household group. From 1993 to 2010, among households composed of two or more adults, 33% to 39% of in-person interviews were conducted privately, increasing to 46% in 2012 and 2013. From 1993 to 1998, among households composed of a single adult with children, 29% to 31% of in-person interviews were conducted privately, which increased to 38% in 1999. From 1999 to 2006, private interviews accounted for 35% to 38% of in-person interviews among these households, and increased to 42% in 2009 and 48% in 2013. Household income ****************** From 1993 to 2004, the highest percentage of private interviews was observed for households with an unknown income level (figure 18). Beginning in 2005, the percentage of private interviews among households earning less than $24,999 annually was about equal to that observed for households with an unknown income. From 2005 to 2013, households in both of these income categories had the highest percentages of in-person private interviews. From 1993 to 1999, households with incomes of less than $24,999, $25,000 to $49,999, and$50,000 or more had similar percentages of private interviews. From 2000 to 2013, the percentage of private interviews was lowest among households earning $25,000 to $49,999 and $50,000 or more, while the percentage of private interviews during this period either increased or remained stable among households in other income categories. Among households earning less than $24,999 annually, 40% to 42% of in-person interviews were conducted privately from 1993 to 1999. From 1999 to 2007, privately conducted interviews generally increased each year in this income category, accounting for 48% of in-person interviews conducted in 2007. Between 2007 and 2010, 46% to 49% of interviews were conducted privately, increasing to 54% in 2013. From 1993 to 2011, among households earning $25,000 to $49,999, 37% to 42% of interviews were conducted privately. In 2012 and 2013, private interviews accounted for 48% of in-person interviews in this income category. Among households earning $50,000 or more annually, the percentage of privately conducted interviews declined from 44% in 1993 to 36% in 2007. From 2007 to 2013, the percentage of private interviews generally increased each year in this income category (40% in 2009, 42% in 2011, and 49% in 2013). Between 1993 and 2011, among households with an unknown income, 44% to 49% of in-person interviews were conducted privately, increasing to 53% in 2013. ***Footnote 4 Although 53% of NCVS interviews were conducted nonprivately from 2009 to 2013, the mix of private and nonprivate interviews varied according to the characteristics of the survey respondent. When examined by respondent characteristics, privately conducted interviews accounted for the majority of interviews among certain subgroups. In addition, among some subgroups, private and nonprivate interviews were conducted at about similar percentages. See appendix table 19 for more detailed information about demographic subgroups for 2009 to 2013.*** ******************** 4. Future Research ******************** The variables examined in these analyses may be associated with each other and with other variables not included in the analyses. Additional research could investigate how these complex relationships are associated with survey estimates of criminal victimization. For example, from 2009 to 2013, differential rates of criminal victimization were associated with the data collection mode. From 2009 to 2013, the rate of serious violent crime associated with in-person interviews (11 victimizations per 1,000 persons 12 or older) was more than two times higher than the rate of serious violent crime associated with telephone interviews (5 per 1,000) (table 2). These observed differences may be due to the interviewing conditions or the composition of respondents in each of the mode groups. Similar associations were found when persons were interviewed alone versus with others. During the period 2009 to 2013, the rate of serious violent crime associated with private interviews (12 per 1,000) was 1.4 times higher than the rate associated with nonprivate interviews (9 per 1,000) (table 3). Higher rates of violent victimization associated with private interviews may occur because respondents choose not to reveal victimizations when others are present, particularly if the offender is another household member or if the victimization reveals sensitive information about the respondent or causes embarrassment. To examine this in more detail, comparisons between victimization rates are made by whether a person lives alone or in a multiperson household with others and whether the interview took place privately or in the presence of others. If interview privacy and household composition are jointly associated with a reduction in the level of reported victimizations on the survey, there should be a difference between victimizations among persons living with others when they are interviewed alone compared to when others are present, but this difference should not be observed between private and nonprivate interviews conducted among persons who live alone. Among persons living in multiperson households, the overall rate of serious violent victimization was 10 victimizations per 1,000 persons age 12 or older (figure 19). However when examined by interview privacy, rates of serious violence were significantly higher among persons who were interviewed privately and significantly lower among persons who were interviewed non-privately. Among persons residing in multiperson households who were interviewed privately, the rate of violent victimization was 12 per 1,000, or 1.4 times higher than the observed rate of serious violent victimization among persons residing in multiperson households who were interview non-privately (9 per 1,000) (appendix table 22). In contrast, among respondents who lived alone, no significant differences were observed between in-person private interviews (35 per 1,000) and in-person nonprivate interviews (34 per 1,000). The absence of observed differences in victimization rates for private and nonprivate interviews conducted with persons who live alone can be important because it suggests that household composition accounts for some portion of the association between victimization rates and privacy. These findings suggest that privacy in multi-person households is important and that additional compositional variables, such as the demographic variables examined in these analyses, could also be associated with interviewing conditions and survey response. To examine the association between privacy and single person and multiperson households in more detail, comparisons between victimization rates among respondent who lived with others but were interviewed privately are compared to overall in-person rates of victimization among respondents who lived alone. The overall in-person rate for live alone households was examined because the focus of the comparison is on ascertaining whether victimization rates observed in private interviews with respondents in multiperson households approximate rates observed among respondents in single- person households, regardless of whether the single person household interview was conducted privately. Among respondents who lived with others, the rate of rape and sexual assault (2 per 1,000 persons age 12 or older) was identical to the observed rate of rape and sexual assault among in-person interviews conducted with respondents who live alone (appendix table 22). Further, no significant differences were observed when comparing rates of robbery, aggravated assault, or simple assault for respondents from multiperson households who were interviewed privately and respondents residing in single-person households. This finding suggests that additional factors, such as household composition, can affect the association between interviewing conditions and rates of victimization. Additional research is needed to better understand the effect of interview privacy on victimization rates in these different contexts. Further investigations could also examine whether the type of person who is present during a nonprivate interview is associated with differential survey estimates under private and nonprivate interviewing conditions. This report focused on whether someone other than the FR and survey respondent was present during the interview, but the NCVS collects far more detailed information on the exact nature of the other persons who are present, including whether it was the respondent’s spouse, other household members, or nonhousehold members. For example, the presence of a parent may influence the responses provided during a teenager’s interview, and the presence of a spouse may influence the responses provided by a male or female survey respondent. In addition, the association between interviewing conditions and NCVS- specific design features deserves further investigation. Survey respondents in the NCVS are interviewed seven times over the course of 3 years, and interviewing conditions (such as mode and privacy) are associated with the interview number in which a person is responding. This report did not examine these types of complex relationships between interviewing conditions, respondent characteristics, survey rates, estimates of crime, and survey design aspects of the NCVS. A final avenue of research should investigate the quality profile of NCVS data collected under different interviewing conditions. Survey questionnaires can be administered in a number of ways, including face to face, telephone, mail, and internet. Each mode has unique advantages and disadvantages related to cost, data quality, nonresponse, flexibility, and timeliness. An analysis of the data derived from different interviewing conditions can shed light on whether NCVS data quality varies across different interviewing conditions. ************** Methodology ************** Survey coverage ****************** The National Crime Victimization Survey (NCVS) is an annual data collection conducted by the U.S. Census Bureau for the Bureau of Justice Statistics (BJS). The NCVS is a self-report survey in which interviewed persons are asked about the number and characteristics of victimizations experienced during the previous 6 months. The NCVS collects information on nonfatal personal crimes (rape or sexual assault, robbery, aggravated assault, simple assault, and personal larceny) and household property crimes (burglary, motor vehicle theft, and other theft) both reported and not reported to police. In addition to providing annual level and change estimates on criminal victimization, the NCVS is the primary source of information on the nature of these incidents. Persons provide information about themselves (e.g., age, sex, race and Hispanic origin, marital status, education level, and income) and whether they experienced a victimization. For each victimization incident, the NCVS collects information about the offender (e.g., age, sex, race and Hispanic origin, and victim–offender relationship), characteristics of the crime (e.g., time and place of occurrence, use of weapons, nature of injury, and economic consequences), whether the crime was reported to police, reasons the crime was or was not reported, and victim experiences with the criminal justice system. The NCVS is administered to persons age 12 or older from a nationally representative sample of households in the United States. The NCVS defines a household as a group of persons who all reside at a sampled address. Persons are considered household members when the sampled address is their usual place of residence at the time of the interview and when they have no usual place of residence elsewhere. Once selected, households remain in the sample for 3 years, and eligible persons in these households are interviewed every 6 months either in person or over the phone, for a total of seven interviews. All first interviews are conducted in person, and subsequent interviews are conducted either in person or by phone. New households rotate into the sample on an ongoing basis to replace outgoing households that have been in the sample for the 3-year period. The sample includes persons living in group quarters (such as dormitories, rooming houses, and religious group dwellings) and excludes persons living in military barracks and institutional settings (such as correctional or hospital facilities) and persons who are homeless. Describing the NCVS sample **************************** The data presented in Sections 1–3 of this BJS Working Paper describe the unweighted distribution of interviewing conditions in the NCVS and the demographic characteristics of sample respondents who responded to NCVS interviews. Because these descriptive analyses are limited to the NCVS sample, they do not represent a national estimate. As a result, tables describing the NCVS sample are based on unweighted data and all differences are actual. Nonresponse and weighting adjustments ************************************** Estimates in Section 4 of this BJS Working Paper use data from the 1993 to 2013 NCVS data files that were weighted to produce annual estimates of victimization for persons age 12 or older living in U.S. households. Because the NCVS relies on a sample rather than a census of the entire U.S. population, weights are designed to inflate sample point estimates to known population totals and to compensate for survey nonresponse and other aspects of the sample design. The NCVS data files include both person and household weights. Person weights provide an estimate of the population represented by each person in the sample. Household weights provide an estimate of the U.S. household population represented by each household in the sample. In 2013, a weighted total of 90,630 households and 160,040 persons age 12 or older were interviewed for the NCVS. Each household was interviewed twice during the year. The response rate was 84% for households and 88% for eligible persons. Victimizations that occurred outside of the United States were excluded from this report. In 2013, less than 1% of the unweighted victimizations occurred outside of the United States and were excluded from the analyses. Victimization weights used in this analysis account for the number of persons present during an incident and for high-frequency repeat victimizations (i.e., series victimizations). Series victimizations are similar in type but occur with such frequency that a victim is unable to recall each individual event or describe each event in detail. Survey procedures allow NCVS interviewers to identify and classify these similar victimizations as series victimizations and to collect detailed information on only the most recent incident in the series. The weight includes series incidents as the actual number of incidents reported by the victim, up to a maximum of 10 incidents. Including series victimizations in national rates results in large increases in the level of violent victimization. However, trends in violence are generally similar, regardless of whether series victimizations are included. In 2013, series incidents accounted for about 1% of all victimizations and 4% of all violent victimizations. Weighting series incidents as the number of incidents up to a maximum of 10 produces more reliable estimates of crime levels, while the cap at 10 minimizes the effect of extreme outliers on rates. Additional information on the series enumeration is detailed in the report Methods for Counting High-Frequency Repeat Victimizations in the National Crime Victimization Survey (NCJ 237308, BJS web, April 2012). Standard error computations ***************************** When national estimates are derived from a sample, as with the NCVS, it is important to use caution when comparing one estimate to another estimate or when comparing estimates over time. Although one estimate may be larger than another, estimates based on a sample have some degree of sampling error. The sampling error of an estimate depends on several factors, including the amount of variation in the responses and the size of the sample. When the sampling error around an estimate is taken into account, the estimates that appear different may not be statistically different. One measure of the sampling error associated with an estimate is the standard error. The standard error can vary from one estimate to the next. Generally, an estimate with a small standard error provides a more reliable approximation of the true value than an estimate with a large standard error. Estimates with relatively large standard errors are associated with less precision and reliability and should be interpreted with caution. To generate standard errors around numbers and estimates from the NCVS, the Census Bureau produced generalized variance function (GVF) parameters for BJS. The GVFs take into account aspects of the NCVS’s complex sample design and represent the curve fitted to a selection of individual standard errors based on the Jackknife Repeated Replication technique. The GVF parameters were used to generate standard errors for each point estimate (e.g., counts, percentages, and rates) in this report. For data presented in Section 4, BJS conducted tests to determine whether differences in estimated numbers, percentages, and rates in this report were statistically significant once the sampling error was taken into account. Using statistical programs developed specifically for the NCVS, all comparisons in the text were tested for significance. The primary test procedure was the Student’s t-statistic, which tests the difference between two sample estimates. Differences described as higher, lower, or different passed a test at the 0.05 level of statistical significance (95% confidence level). Differences described as somewhat, slightly, or marginally different, or with some indication of difference, passed a test at the 0.10 level of statistical significance (90% confidence level). Caution must be taken when comparing estimates not explicitly discussed in this report. Data users can work with the estimates and the standard errors of the estimates provided in this report to generate a confidence interval around the estimate as a measure of the margin of error. The following example illustrates how standard errors can be used to generate confidence intervals: According to the NCVS, for the period 2009 to 2013, the rate of violent crime for in-person self-interviews was 30.1 per 1,000 persons (see table 2). Using the GVFs, it was determined that the estimated victimization rate has a standard error of 1.5 (see appendix table 20). A confidence interval around the estimate was generated by multiplying the standard errors by ±1.96 (the t-score of a normal, two-tailed distribution that excludes 2.5% at either end of the distribution). Therefore, the 95% confidence interval around the 30.1 estimate is 30.1 ± (1.5 × 1.96) or (27.1 to 33.0). In other words, if different samples using the same procedures were taken from the U.S. population, 95% of the time the violent crime rate for in-person interviews would fall between 27.1 and 33.0 per 1,000 persons. In this report, BJS also calculated a coefficient of variation (CV) for all estimates, representing the ratio of the standard error to the estimate. CVs provide a measure of reliability and a means to compare the precision of estimates across measures with differing levels or metrics. In cases in which the CV was greater than 50%, or the unweighted sample had 10 or fewer cases, the estimate was noted with a “!” symbol (interpret data with caution; estimate based on 10 or fewer sample cases, or the coefficient of variation is greater than 50%). Methodological changes to the NCVS in 2006 ********************************************* It was originally thought that methodological changes implemented in 2006 might have affected the crime estimates for that year to such an extent that they are not comparable to estimates from other years. BJS and the Census Bureau conducted an evaluation of 2007 and later data from the NCVS and found a high degree of confidence that estimates for 2007 through 2013 are consistent with and comparable to estimates for 2005 and previous years. See the Criminal Victimization series of reports for 2006 to 2014 on the BJS website. *********************************************************** The Bureau of Justice Statistics of the U.S. Department of Justice is the principal federal agency responsible for measuring crime, criminal victimization, criminal offenders, victims of crime, correlates of crime, and the operation of criminal and civil justice systems at the federal, state, tribal, and local levels. BJS collects, analyzes, and disseminates reliable and valid statistics on crime and justice systems in the United States, supports improvements to state and local criminal justice information systems, and participates with national and international organizations to develop and recommend national standards for justice statistics. Jeri M. Mulrow is acting director. This report was written by Shannan Catalano, BJS Statistician. Erika Harrell, BJS Statistician verified the report. Irene Cooperman and Brigitte A. Coulton edited the report. Barbara Quinn and Tina Dorsey produced the report. August 2016, NCJ 249682 *********************************************************** ************************************************ Office of Justice Programs Innovation * Partnerships * Safer Neighborhoods www.ojp.usdoj.gov ************************************************ ******************** 8/4/16 JER 10:50am ********************