| Skip Navigation Links | ||
 |
 |
 |
|
The NLS Annotated Bibliography - User Submission Form
BRANDEN, LAURA PERGAMIT, MICHAEL R. Response Error in Reporting Starting Wages Presented: Danvers, MA, American Association of Public Opinion Research Annual Conference, May 1994 Cohort(s): NLS General, NLSY79 ID Number: 2766 Publisher: American Association of Public Opinion Research Permission to reprint the abstract has not been received from the publisher. Human capital models in labor economics emphasize, among other things, the returns to tenure on a job. While longitudinal data improve these measures compared with cross-sectional data, complete wage profiles for an individual in any household data set do not exist. Generally, the available data consist of a series of contemporaneous wage observations gathered at infrequent intervals, usually once each year. This is the standard in the Panel Study of Income Dynamics (PSID) and in the various National Longitudinal Surveys, the primary longitudinal data sets in labor economics. Since it is unlikely that we observe a person exactly when they begin their job, we must retrospectively ask their starting wage. Retrospective questions tax people's memories in different ways depending on the nature of the information to be retrieved, how it is stored in memory, the length of recall required, the saliency of the event, etc. Starting wages are expected to be perhaps the most easily recalled wages other than the current wage because the starting wage is connected with a specific event, i.e. beginning work for a given employer. Therefore, an investigation of individuals' reports of starting wages are probably the most accurate of any wage reports other than their current wage. In this paper, we use the National Longitudinal Survey of Youth (NLSY), taking advantage of a skip pattern error which resulted in re-asking most of the sample about the starting wage for their employer at two consecutive interviews. Because we never know the true starting wage, this paper examines the consistency in response between the two answers given at two different interviews, roughly one year apart. ERNST, MICHELLE PERGAMIT, MICHAEL R. Data Quality and the Use of Standardized Child Assessments in Survey Research Presented: Miami Beach, FL, Annual Meeting of the American Association For Public Opinion Association, May 2005. Cohort(s): Children of the NLSY79 ID Number: 5850 Publisher: American Association of Public Opinion Research Permission to reprint the abstract has not been received from the publisher. Beginning with the Children of the National Longitudinal Survey of Youth/1979 cohort (CNLSY79) in 1986, large-scale surveys began to incorporate standardized assessments. Formerly used only in clinical settings or schools, these assessments are now administered in a household setting by lay field interviewers. The administration of standardized assessments to children by lay field interviewers raises data quality concerns. Standardized child assessments have rigid administration protocols. Deviation from procedure can greatly affect a child's response. Furthermore, administrative complexity varies across assessments. While some assessments consist of a very simple and straight-forward administrative protocol, other assessments rely much more on the skills of the individual conducting the administration. It is hypothesized that an administratively complex assessment with strong published psychometric properties may not maintain those properties when administered by interviewers in large-scale studies. This paper proposes examining the published psychometrics for three assessments (the Woodcock-Johnson, the PPVT, and the PIAT) and comparing the published psychometrics with its reliability and validity within single, longitudinal studies (the NLSY79 and the PSID). By using multiple years of assessment data from the Children of the NLSY79 (PPVT/PIAT) and the Panel Study of Income Dynamics Child Supplement (Woodcock-Johnson), we have access to a large number of assessments conducted by a large number of interviewers. We can compare distributions between interviewers as well as looking at the same interviewer over time. These two data sets provide a rich source of data on assessments that allows us to examine many differences in administration. It will also be possible to examine how the psychometric properties of different assessments stand up in a large-scale survey as a function of the complexity of the assessment. If interviewer variability is greater in administrations of the complex tests, this argues for greater consideration of administrative procedures when choosing assessments for large-scale survey research. O'MUIRCHEARTAIGH, COLM PEDLOW, STEVEN Combining Samples versus Cumulating Cases: A Comparison of Two Weighting Strategies in NLSY97 Presented NYC: Annual Meeting of the American Statistical Association, August 11-152002; Also presented Portland OR: American Association for Public Opinion Research (AAPOR) 55th Annual Conference Meetings, May 2000 Cohort(s): NLSY97 ID Number: 3576 Publisher: American Association of Public Opinion Research Permission to reprint the abstract has not been received from the publisher. AAPOR SESSION E: Impact of Telephone Sampling Design on Sample Efficiency and Bias -- Friday 5/19/2000 WANG, YONGYI PARVATI, KRISHNAMURTY Interview Mode Effects in NLSY97 Round 4 and Round 5 Presented: Phoenix, AZ, American Association of Public Opinion Research Annual Meeting, May 2004 Cohort(s): NLSY97 ID Number: 4570 Publisher: American Association of Public Opinion Research Permission to reprint the abstract has not been received from the publisher. The incidence of telephone interviewing has been increasing in successive rounds of NLSY97. There are concerns about the accuracy of responses to sensitive questions when the interview is conducted by telephone compared to when these questions are self-administered as part of an in-person interview. This study explores the impact of interview mode on respondents' willingness to reveal sensitive information in NLSY97 round 4 and round 5. The dependent measures for this study include sex behavior, smoking, drug use, destroying, stealing, attacking and arrest. Within each round, controlling for the differences in demographic characteristics, respondents tend to underreport negative behaviors on most SAQ items when interviews are conducted by telephone. They are also less willing to respond to these sensitive questions, resulting in more missing data. We also linked the two rounds together by looking at how individual respondents responded to the same questions in round 4 and round 5. The results show that for respondents who did not switch interview mode across rounds, the distributions of response differences do not differ much regardless of whether the interviews were conducted consistently in-person or by phone. If the respondents did switch interview modes across rounds, the distribution of response differences are significantly different for some sensitive items, depending on whether the switch is from in-person to phone or the other way round. This evidence also supports the existence of interview mode effects. Search returned 4 items. Search Start: 00:02:05 Search Finish: 00:02:05
Last Modified Date: June 13, 2002 - 05:49 AM | |||||
|
|
|
|
URL: http://www.nlsinfo.org |
|