Methodological Reports
Articles on methodological issues in survey research specifically dealing with the GSS as well as more general problems.
Data Release 3a for the 2022 GSS Cross-section data, featuring a new multi-mode design, are now available. The additional data features new weighting recommendations, expanded household composition in 2022, and the inclusion of an oversample of Black, Hispanic, and Asian respondents in the 2022 single-year datafile. Release 3a additionally corrects issues with FUND and RELITEN. We encourage users to review the documentation and consider the potential impact of experiments and data collection approaches on survey estimates.
Wells, Brian M.; Sparkman, Rachel
MR138 GSS 2022 AmeriSpeak OversampleMethodological Reports
This report details the inclusion of AmeriSpeak® panelists as an oversample population in the 2022 General Social Survey (GSS) and the implications of including Black, Hispanic, and Asian oversample from this sample source. This report provides an overview of the AmeriSpeak sample and its properties relevant for the 2022 GSS. We examine how the AmeriSpeak oversample cases compare to the baseline GSS sample and how they impact estimates at the population and oversampled group levels.
The high-level findings are as follows:
• The AmeriSpeak cases exhibit some demographic differences from their baseline counterparts, but often improve representation, particularly for racial and ethnic subgroups (e.g., South American Hispanic groups, Chinese).
• Given the AmeriSpeak sample only completed the GSS on the web, there are some differences in substantive responses consistent with previous GSS work suggesting sensitivity to mode.
• U.S. population estimates should exhibit minimal differences between the existing 2022 estimates without the AmeriSpeak oversample as with the AmeriSpeak oversample.
• Including the Black and Hispanic oversamples minimally change the overall estimates for their respective subpopulations, but including the Asian oversample does produce large estimate changes for Asian subpopulation given the oversample accounts for a majority of the total Asian sample.
The AmeriSpeak oversample offers increased sample sizes for Black, Hispanic, and Asian respondents in the 2022 GSS Cross-section. In particular, the sample size for Asian respondents more than doubles with the inclusion of the oversample given their low prevalence in the population. While the Asian subpopulation estimates see more movement than their Black and Hispanic counterparts, we see improved representation for Asian subgroups, suggesting a potential improvement in estimation more broadly given the small initial sample size. Researchers are encouraged to conduct their own research to determine additional impacts of including the AmeriSpeak oversample.
GSS years: 2022
Wells, Brian M.; Seeskin, Zachary H.; Ihde, Amy
MR137 Post-stratification Weights for GSS 1972-2022Methodological Reports
This report describes a new set of post-stratification weights available for users of the 1972-2018 General Social Survey (GSS) cross-sectional surveys to help improve nonresponse bias adjustment. The weight derivation follows the approach applied to 2021 and 2022 GSS Cross-sections. Use of these weights results in weighted totals that, for each GSS cross-sectional sample, equal marginal control totals from the U.S. Census Bureau estimates for education, sex, marital status, age, region of the country, race, U.S. born status, and Hispanic origin when available. NORC recommends that GSS data users use this new weight for all analyses in the future. These weights also: (a) correct for the form assignment errors reported in GSS Methodological Report 36 for 1978, 1980, 1982, 1983, 1984, and 1985; (b) correct for the ballot-and-form assignment errors reported in GSS Methodological Report 134 for 2002, 2010, 2012, 2016, and 2018; and (c) support person-level analyses of the combined main and Black oversamples for 1982 and 1987. Given the global trend of declining response rates over the past several years, the use of auxiliary data, such as U.S. Census totals for nonresponse adjustment, is important for improving representativeness of estimates with respect to key demographic characteristics. In addition, this report examines the impact of using the poststratification weights across all GSS cross-sections. The majority of estimate differences observed include poststratification variables and their close correlates.
GSS years: 1972-2018
Wells, Brian M.; Davern, Michael; Ihde, Amy; Bautista, Rene
MR136 An Examination of Demographic and Substantive Differences between Early and Late Respondents in a Face-to-face Field SurveyMethodological Reports, NORC Working Paper
The General Social Survey (GSS), a biennial nationally representative survey of the U.S. adult population, has employed subsampling since 2004. Approximately halfway through the field period in years prior to 2020, half of the remaining cases are randomly subsampled for a more focused follow-up, while the other cases are dropped. Subsampling in the GSS has helped to improve response rates and to achieve cost and sample size efficiencies (O’Muircheartaigh and Eckman 2007). This paper explores the extent to which subsampled (or late) respondents vary from non-subsampled (or early) respondents in GSS 2014, 2016, and 2018. We first examine the demographic characteristics of early and late respondents. Second, we explore substantive differences between the two groups on key analytic variables (e.g., attitudes toward premarital sex, abortion, the death penalty, gun regulation, marijuana legalization, national spending priorities). Finally, we examine differences between early and late respondents on key GSS analytic variables controlling for demographic differences using multivariate logistic regression. Our investigations over three years of the GSS suggest that some demographic and
substantive differences between early and late respondents exist, consistent with previous GSS research (Smith 2006). Our results also suggest that most of the differences on key analytic variables do not persist after controlling for demographic characteristics in multivariate logistic regression models. This finding is consistent with past research on interviewer-administered surveys that find that late respondents are not different from early responders on most variables net of demographic characteristics (e.g., Keeter et al. 2006). Differences found between the 2014, 2016, and 2018 analyses emphasize the need for continued research related to subsampling in the GSS.
GSS years: 2014, 2016, 2018
MR135 GSS Post-Stratification Weights
Methodological Reports
This memo describes a new set of post-stratification weights available for users of the 2000–2018 GSS cross-sectional surveys. The weight derivation follows the approach applied to the previously released 2021 GSS Cross-section, for which post-stratification weights were developed to improve nonresponse bias adjustment, given the impact of the COVID-19 pandemic on survey operations and response rate. Use of these weights results in weighted totals of each GSS cross-sectional sample that equal marginal control totals from the U.S. Census Bureau estimates for education, sex, marital status, age, region of the country, race, Hispanic origin, and U.S. born status. These weights also correct for the ballot and ballot-and-form assignment errors reported in GSS Methodological Report 134 for 2002, 2010, 2012, 2016, and 2018. The use of auxiliary data such as U.S. Census totals for nonresponse adjustment is important for improving representativeness of estimates with respect to key demographic characteristics, given the global trend of declining response rates over the past several years.
GSS years: 2002, 2010, 2012,2016, 2018
Christian, Leah; Davis, Nicholas; Lancaster, Caroline; Paddock, Susan
MR 134: Assignment Error by Ballot and FormMethodological Reports
As previously reported, an unintended overlap between respondent selection and questionnaire assignment procedures in GSS surveys created an association between questionnaire version (ballot and form) and age order in some households in the historical GSS data. This assignment error occurred in data years 2002, 2010, 2012, 2016 and 2018. This methodological report describes the equivalence testing to compare original and corrected estimates for each category and each variable for all questions in the affected years. The analysis shows that the error differences due to the assignment error are overall relatively small and corrected weights have been provided for data users.
GSS years: 2002, 2010, 2012, 2016, 2018,
Cowan, Sarah K.; Hout, Michael; Perett, Stuart
MR133 Updating A Time-Series of Survey Questions: The Case of Abortion Attitudes in the General Social SurveyLong-running surveys need a systematic way to reflect social change and to keep items relevant to respondents, especially when they ask about controversial subjects, or they threaten the items’ validity. We propose a protocol for updating measures that preserves content and construct validity. First, substantive experts articulate the current and anticipated future terms of debate. Then survey experts use this substantive input and their knowledge of existing measures to develop and pilot a large battery of new items. Third, researchers analyze the pilot data to select items for the survey of record. Finally, the items appear on the survey-of-record, available to the whole user community. Surveys-of-record have procedures for changing content that determine if the new items appear just once or become part of the core. We provide the example of developing new abortion attitude measures in the General Social Survey. Current questions ask whether abortion should be legal under varying circumstances. The new abortion items ask about morality, access, state policy, and interpersonal dynamics. They improve content and construct validity and add new insights into Americans’ abortion attitudes.
Morgan, Stephen L.; Lee, Jiwon
MR132 Six Alternative Weights that Adjust for Attrition in the 2006-2014 General Social Survey PanelsMR132
In this report, we present strategies for constructing weights to adjust for attrition in the GSS treble panel. We offer Stata code for the construction of the weights that we explain, as well as data files of weights that researchers may wish to adopt for their own use.
GSS years: 2006, 2008, 2010, 2012, 2014
Morgan, Stephen L.
MR131 Response Rates and Representativeness: A Benchmark Comparison of the General Social Surveys to the American Community Surveys, 2012-2018MR131
(no abstract provided)
GSS years: 2012, 2014, 2016, 2018
Smith, Tom W.
MR009a Sex and the GSS - Nonresponse Differences (2019)Methodological Report, Chicago, NORC , 2019
* Please note this is an updated version of MR009 (1979).
The problem of underrepresentation of males on the GSS reflects the nonresponse tendency of males, possibly exacerbated by female interviewers. Surveys using full probability sampling generally have an underrepresentation of males.
Smith, Tom W.; Son, Jaesok
MR130 Applying the Multi-level, Multi-Source (MLMS) Approach to the 2016 General Social SurveyMR130
(no abstract provided)
GSS years: 2016
Smith, Tom W.; Son, Jaesok
MR129 Transgender and Alternative Gender Measurement on the 2018 General Social Survey(no abstract provided)
Stephen L. Morgan
MR128 - Clergyman, Cleric, and Religious Leader: An Experiment on Alternative Reference Individuals for the Free Expression ItemsMethodological Reports
For the 2020 GSS, a review of the free expression items suggested revisions to “a Muslim clergyman who preaches hatred of the United States,” as part of a broader effort by the GSS Board to reassess all GSS items that are gender-specific in some way. Two gender-neutral alternatives were discussed, “an Islamic cleric who preaches hatred of the United States” and “an Islamic religious leader who preaches hatred of the United States.”
For the reasons detailed below, it is possible that a switch to “an Islamic cleric who preaches hatred of the United States” could prompt an undesirable discontinuity in response patterns, beyond what could be expected to result from a gender-neutral substitution. If some GSS respondents are more likely to suspect that the referenced Islamic cleric has a connection to terrorism, the elicited response may be a mixture of opposition to free expression and a perceived fear of physical violence, with more weighting on the latter. In contrast, “an Islamic religious leader who preaches hatred of the United States” may be preferable, if it is the case that GSS respondents are no more likely to infer a threat to their security than is the case for “a Muslim clergyman who preaches hatred of the United States.”
In this report, I offer two sets of results to inform decisions about the questionnaire for the 2020 GSS. First, to set the background, I use GSS data from 2008 through 2018 to summarize levels and changes in attitudes toward free expression for all six existing reference individuals. Second, I offer results from a three-armed experiment that compares “Muslim clergyman” to the two alternatives of “Islamic cleric” and “Islamic religious leader.” The experimental data were collected over the web in January and February of 2019 as part of the AmeriSpeak panel.
GSS years: 1972-2018
Smith, Tom W.
MR127 Question Wording Experiments: Job Satisfaction and the Co-residence of Adult Children and their Parentson the 2016 General Social Survey (GSS), two question-wording experiements were conducted testing variant versions of core GSS items on job satisfaction (SATJOB) and the co-residence of adult children and their parents (AGED). This report details the findings of these experiments.
GSS years: 2016
Smith, Tom W.; Kim, Jibum
MR126 A Review of Survey Data-Collection Modes: With a Focus on ComputerizationsSurveys are conducted using many different modes (e.g. face-to-face, mail, telephone, Internet). Because different modes have different error structures, it is very important to understand the advanctages and disadvantages associated with each mode. In recent years there have been major changes in the mdoes typically utilized in surveys. In particular, there have been increases in the use of computers in data collection, self-administration, and mixed-mode designs. The implications of these and future changes are considered.
Morgan, Stephen L.
MR125 A Coding of Social Class for the General Social SurveyThis paper documents an update to the Erickson, Goldthorpe & Portocarero social class schema, first proposed in 1987. Due to the backcoding of the 2010 US Census Occupational Classification, it is now possible to treat all GSS cases according to the same occupational codes, which can then be linked to updated EGP codes. Validity is ascertained using the 2012 American Community Survey’s occupational classification. Please note that this methodological report also includes a .do file for adding EGP codes to a STATA GSS datafile, as well as an OCC10 to EGP crosswalk, included below with download links
GSS years: 1972-2016
http://gss.norc.org/Documents/other/occ10-to-egp-class-crosswalk.csv http://gss.norc.org/Documents/other/code-for-egp-crosswalk.do
Hout, Michael Smith, Tom W. Marsden, Peter V.
MR124 Prestige and Socioeconomic Scores for the 2010 Census CodesMethodological Report, Chicago, NORC
The 2012 GSS included a popular prestige rating (Smith and Son 2014). A sample of 1,001 individuals, first interviewed in 2008 and included in the GSS panel, rated 90 occupations each; a rotation of occupations among respondents resulted in ratings for 860 occupational titles, most of which could be assigned to one of the 840 codes in the 2010 Standard Occupational Classification (SOC). This methodological report explains how we collected the ratings and converted them into prestige scores and a socioeconomic index for each of the 539 occupational categories of the Census Bureau's coding scheme now used in the GSS.
GSS years: 2012, 2014
Smith, Tom W. Laken, Faith Son, Jaesok
MR123 Gun Ownership in the United States: Measurement Issues and TrendsMethodological Report, Chicago, NORC , 1, 2014
Given the magnitude and seriousness of gun violence, it is important to have accurate and reliable information on the possession and use of firearms in the United States. This report examines one crucial element, the level of and trends in household and personal gun ownership. First, the report considers methodological issues concerning the measurement of gun ownership. Second, it examines trends in gun ownership. Third, it evaluates the nexus of these two factors, the impact of methodological issues on the measurement of trends gun ownership. Finally, it considers what ancillary trend data on crime, hunting, household size, and number of guns available suggest about trends in gun ownership.
GSS years: 1972-2014
Smith, Tom W. Son, Jaesok
MR122 Measuring Occupational Prestige on the 2012 General Social SurveyMethodological Report, Chicago, NORC , 1, 2014
(no abstract provided)
GSS years: 2012
Smith, Tom W. Kim, Jibum
MR121 The Multi-Level, Multi-Source (ML-MS) Approach to Improving Survey ResearchMethodological Report, Chicago, NORC
(no abstract provided)
GSS years: N/A
Kim, Jibum Son, Jaesok Kwok, Peter K. Kang, Jeong-han Laken, Faith Daquilanea, Jodie Shin, Hee-Choon Smith, Tom W.
MR120 Trends and Correlates of Income Nonresponse: Forty Years of the General Social SurveyMethodological Report, Chicago, NORC , 2012
Using the 40 years of the General Social Survey (GSS), we investigate the long-term trend and the correlates of family and personal income nonresponse. Family and personal income nonresponse has increased slightly by about 5 percentage points from 1974 to 2010 (9% to 13% in family income; 7% to 12% in personal income). While family income nonresponse was equivalently attributed to “Don’t Know” and “Refused,” personal income nonresponse was mainly attributed to “Refused.” We found very similar correlates of family and personal income nonresponse, such as being older, female, married, self employed, those not answering the number of earners, uncooperative respondents, people living in the East, and those surveyed in recent periods. In addition, based on the interviewer’s evaluation, uncooperative respondents are less likely to response “Don’t Know” than “Refused” and respondents with poor comprehension are more likely to respond “Don’t Know” than “Refused.” Our findings suggest that we need to distinguish “Refused” from “Don’t Know” if we aim to better understand income nonresponse and to consider paradata to evaluate the cognitive processing of income nonresponse.
GSS years: 1972-2012
Hout, Michael Hastings, Orestes P
MR119 Reliability and Stability Estimates for the GSS Core Items from the Three-wave Panels, 2006–2010Methodological Report, Chicago, NORC , 6, 2012
We assess the reliability and stability of core items in the General Social Survey using Alwin’s (2007) implementation of Heise’s (1969) model. Of 265 core items examined we find mostly positive results. Eighty items (over 30 percent) have reliability coefficients greater than 0.85; another 84 (32 percent)
have reliability coefficients between 0.70 and 0.85. Facts are generally more reliable than other items. Stability was slightly higher, overall, in the 2008-2010 period than the 2006-2008 period. The economic recession of 2007-09 and the election of Barack Obama in 2008 altered the social context in ways that may have contributed to instability.
GSS years: 2006, 2008, 2010
Smith, Tom W. Son, Jaesok
MR118 An Analysis of Panel Attrition and Panel Change on the 2006-2008 General Social Survey PanelMethodological Report, Chicago, NORC , 5, 2011
(no abstract provided)
GSS years: 2006, 2008
Smith, Tom W.
MR117 An Analysis of Computer Assisted Recorded Interviews (CARI) on the 2008 General Social SurveyMethodological Report, Chicago, NORC , 5, 2010
(no abstract provided)
GSS years: 2006, 2008
Smith, Tom W. Kim, Jibum
MR116 An Assessment of the Multi-level Integrated Database ApproachMethodological Report, Chicago, NORC , 11, 2009
(no abstract provided)
GSS years: 2002, 2004, 2006, 2008
Smith, Tom W.
MR115 An Analysis of New Vocabulary Items on the 2008 General Social SurveyMethodological Report, Chicago, NORC , 2010
(no abstract provided)
GSS years: 2006,2008
Smith, Tom W.
MR114 Response Order and Response Balance Effects Experiments on the 2008 General Social SurveyMethodological Report, Chicago, NORC, 2009
(no abstract provided)
GSS years: 2008
Smith, Tom W.
MR113 2006-2008 General Social Survey Panel ValidationMR113 2006-2008 General Social Survey Panel Validation
(no abstract provided)
GSS years: 2006, 2008
Smith, Tom W. Sokolowski, John
MR112 Using Audio-Visuals in SurveysMethodological Report, Chicago, NORC , 2008
(no abstract provided)
GSS years: 2002,2008
Malhotra, Neil Krosnick, Jon A Haertel, Edward
MR111 The Psychometric Properties of the GSS Wordsum Vocabulary TestMethodological Report, Chicago, NORC , 8, 2007
Social scientists in many disciplines have used the GSS's ten-item Wordsum vocabulary test to study the causes and consequences of vocabulary knowledge and related constructs. In adding up the number of correct answers to yield a test score, researchers have implicitly assumed that the ten items all reflect a single, underlying construct and that each item deserves equal weight when generating the total score. In this paper, we report evidence suggesting that extracting the unique variance associated with each word and measuring the latent construct only with the variance shared among all indicators strengthens the validity of the index. We also report evidence suggesting that Wordsum could be improved by adding words of moderate difficulty to accompany the existing questions that are either quite easy or quite difficult. Previous studies that used Wordsum should be revisited in light of these findings, because their results might change when a more optimal analytic method is used.
GSS years: 1974-2004
Smith, Tom W. Kim, Seokho
MR110 Counting Religious Nones and Other Religious Measurement IssuesMethodological Report, Chicago, NORC , 4, 2007
The new Baylor Religion Survey (BRS) is an important addition to the available data on religion in contemporary America ('American Piety,' 2006). Few national surveys have included so many valuable questions on the religious background, beliefs, and behaviors of adult Americans. The BRS is a fruitful source for expanding our knowledge about religion and the initial analysis that accompanied the release of the data last Fall has already made a major contribution to the sociology of religion ('American Piety' 2006; 'American Piety 2005,' 2006; Dougherty, Johnson, and Polson, 2006; Dougherty, Johnson, and Polson, 2007; 'Losing My Religion,' 2006).
GSS years: 2006
Smith, Tom W.
MR109 An Evaluation of Spanish Questions on the 2006 General Social SurveyMethodological Report, Chicago, NORC , 2007
(no abstract provided)
GSS years: 2006
Kim, Jibum Smith, Tom W. Kang, Jeong-han Sokolowski, John
MR108 Community Context and Survey CooperationMethodological Report, Chicago, NORC , 2006
(no abstract provided)
GSS years: 1984-2002
Smith, Tom W.
MR107 Wording Effects on the National Spending Priority Items Across Time, 1984-2004Methodological Report, Chicago, NORC , 2006
(no abstract provided)
GSS years: 1984-2004
Smith, Tom W.
MR106 The Subsampling of Nonrespondents on the 2004 General Social SurveyMethodological Report, Chicago, NORC , 2006
(no abstract provided)
GSS years: 2006
Smith, Tom W.
MR105 A Revisit to 'The Vanishing Protestant Majority' and Expanded Religious Probes on the 2004 General Social SurveyMethodological Report, Chicago, NORC , 2005
Several probes were added to the 2004 GSS to see if the declining Protestant population was due to the data being hidden in Christian or Inter/non-denominational categories. Very few cases were found, but the 2006 GSS will attempt to resolve the status of several dozen cases.
GSS years: 1972-2004
Smith, Tom W.
MR104 An Experimental Comparison of Methods of Measuring EthnicityMethodological Report, Chicago, NORC , 2005
The large rise in multiple ethnic mentions was due to the change in mode to CAPI. Although this change was noted, there was no significant change on the distribution of ethnicities.
GSS years: 2002,2004
Smith, Tom W.
MR103 The Impact of Alternative Response Scales on Measuring Self-ratings of HealthMethodological Report, Chicago, NORC , 2005
Switching from a 4-category to 5-category health scale doesn't change explanatory power and would prohibit trends in these categories from being reliably estimated across scales. Changing from 4 to 5 also shifts distribution at the positive end, but not at the negative end.
GSS years: 1972-2004
NHIS, FQES, National Health and Nutrition Examination Survey II
Smith, Tom W.
MR102 Discrepancies in Gender CodesMethodological Report, Chicago, NORC , 2005
Using CAPI (introduced in 2002) allows researchers to compare HEF variables and reconcile conflicting information while still in the field. Cleaning and consistency checks will be made of HEF variables to obtain more accurate data.
GSS years: 1980-2004
Hout, Michael
MR101 Getting the Most out of the GSS Income MeasuresMethodological Report, Chicago, NORC , 2004
(no abstract provided)
GSS years: 1972-2002
Kim, Jibum Smith,Tom W. Kim, Seokho Kang, Jeong-han Berktold, Jennifer
MR100 Comparison between SAQ and CASI for Supplemental Survey NonresponseMethodological Report, Chicago, NORC , 2004
(no abstract provided)
GSS years: 2000, 2002
Smith, Tom W. Michael, J. Michael
MR099 Comparing the Knowledge Networks Web-Enabled Panel and the In-Person 2002 GSS, Experiments with Mode, Format, and Question WordingsMethodological Report, Chicago, NORC , 2003
(no abstract provided)
GSS years: 1984, 2000, 2002
Smith, Tom W.
MR098 A Review of CAPI Effects on the 2002 GSSMethodological Report, Chicago, NORC , 2003
(no abstract provided)
GSS years: 2002
Smith, Tom W.
MR097 A Methodological Analysis of HIV Risk Behavior from the 1988-2002 General Social SurveyMethodological Report, Chicago, NORC , 2003
(no abstract provided)
GSS years: 1988-2002
Smith, Tom W.
MR096 Measuring Racial and Ethnic discrimination" Methodological Report, Chicago, NORC , 2002(no abstract provided)
GSS years: 1972-2000
Smith, Tom W.
MR095 An Experimental Comparison of Internet and In-Person SurveysMethodological Report, Chicago, NORC , 10, 2001
(no abstract provided)
GSS years: 2001
Marsden, Peter V.
MR094 Interviewer Effects Measuring Network Size Using a Single-Name GeneratorMethodological Report, Chicago, NORC , 6, 2001
Name generators, used measuring egocentric networks in surveys, are complex questions that make substantial demands on respondents and interviewers alike. They are therefore
vulnerable to interviewer effects, which arise when interviewers administer questions differently in ways that affect responses-in particular, the number of names elicited. Van Tilburg (1998) found significant interviewer effects on network size in a study of elderly Dutch respondents; that study included an instrument with seven name generators, the complexity of which may have accentuated interviewer effects. This article examines a simpler single-generator elicitation instrument administered in the 1998 General Social Survey (GSS). Interviewer effects on network size as measured by this instrument are smaller than those found by Van Tilburg, but only modestly so. Variations in the network size of respondents within interviewer caseloads (estimated using a single-item "global" measure of network size and an independent sample of respondents) reduce but do not explain interviewer effects on the name generator measure.
Interviewer differences remain significant after controls for between-interviewer differences in the sociodemographic composition of respondent pools. Further insight into the sources of interviewer effects may be obtained via monitoring respondent-interviewer interactions for differences in how name generators are administered.
GSS years: 2000
Smith, Tom W.
MR093 Aspects of measuring Race: Race by Observation vs. Self-Reports and Multiple Mentions of Ethnicity and RaceMethodological Report, Chicago, NORC , 7, 2001
(no abstract provided)
GSS years: 2001
Smith, Tom W.
MR092 A Methodological Analysis of HIV Risk Behavior from the 1988-1998 General Social SurveyMethodological Report, Chicago, NORC , 1999
(no abstract provided)
GSS years: 1988-1998
Smith, Tom W.
MR091 Measuring Inter-racial Friendships: Experimental ComparisonsMethodological Report, Chicago, NORC , 12, 2002
(no abstract provided)
GSS years: 1998
Smith, Tom W.
MR090 Examining the Relationship Between Educational Attainment, Age/Cohort, and Dependent VariablesMethodological Report, Chicago, NORC , 1998
The relationship between educational attainment and age/cohort is curvilinear, and not negative as the historical trend might indicate, for two reasons: the advent of associate degrees and the time required to obtain graduate degrees. Control variables (for age/cohort) straighten out otherwise confounding relationships.
GSS years: 1990, 1991, 1993, 1994, 1996
ANES 1984-1991
Smith, Tom W.
MR089 measuring race by observation and self-identificationMethodological Report, Chicago, NORC , 1997
(no abstract provided)
GSS years: 1994, 1996
Smith, Tom W.
MR088 A Review of Church Attendance MeasuresMethodological Report, Chicago, NORC , 2, 1997
(no abstract provided)
GSS years: 1996
Smith, Tom W. Shin, Hee-Choon Xiaoxi, Tong
MR087 A Report on Sample Frame Comparisons and Design Effects on the 1993 General Social SurveyMethodological Report, Chicago, NORC , 2, 1996
There were few sample frame effects associated with NORC's shift from the 1980 to the 1990 Census. Also, design effects based on the 1990 sample frame are of the same nature and magnitude as indicated by previous research.
GSS years: 1993
Smith, Tom W.
MR086 The Impact of the Presence of Others on a Respondents Answers to QuestionsMethodological Report, Chicago, NORC , 12, 1995
(no abstract provided)
GSS years: 1972-1994
Smith, Tom W.
MR085 A Review of Ethno-Racial Measures on the General Social SurveyMethodological Report, Chicago, NORC , 6, 1995
This article reviews questions in the GSS asking a respondent's race or ethnicity and proposes several methods in which these measures could be refined.
GSS years: 1972-1994
Smith, Tom W.
MR084 Context Effects Resulting from the Revision of the GSS Core in 1994Methodological Report, Chicago, NORC , 1995
The GSS recently shortened the length of its core and, as a result, the context of many items changed. Few context effects were caused by these shifts.
GSS years: 1994
Smith, Tom W.
MR083 Some Aspects of Measuring EducationMethodological Report, Chicago, NORC , 1995
(no abstract provided)
GSS years: 1972-1993
Smith, Tom W.
MR082 Trends in Non-Response RatesMethodological Report, Chicago, NORC , 8, 1994
(no abstract provided)
GSS years: 1975-1993
Smith, Tom W.
MR081 A Comparison of Two Government Spending ScalesMethodological Report, Chicago, NORC , 3, 1994
Response differences in two GSS and one ISSP spending scale were slight, except for spending on the environment, which showed a context effect. In all, the spending scales are generally answered in a consistent and meaningful manner.
GSS years: 1990
ISSP 1990
Smith, Tom W.
MR014 House Effects and the Reproducibility of Survey Measurements: A Comparison of the 1980 General Social Survey and the 1980 American National Election StudyMethodological Report, Chicago, NORC , 1980
The 1980 General Social Survey and the American National Study by SRC are compared to determine house effects. The difference on frequency of don't know categories between the two surveys is the most significant house effect. Also, the difference in time of interview and training of interviewers causes variation in data.
GSS years: 1980
ANES 1980
Smith, Tom W.
MR015 Non-Attitudes: A Review and EvaluationMethodological Report, Chicago, NORC , 1985
Studies of voting behavior and other political matters in the fifties developed a picture of the American electorate that was startlingly at odds with the basic assumption of a rational citizenry as formulated in classic democratic theory. In general, the low or defective levels of conceptualization, information, participation, attitude constraint, and consistency were seen as indicating a very underdeveloped level of political thought and weak or disorganized political attitudes. In particular, inconsistency in attitudes over time was interpreted as indicating an abundance of non-attitudes. In this paper, we will review the literature on non-attitudes. We will examine how the concept of non-attitudes compares with rival explanations of mass belief systems and evaluate the conceptual and evaluate the conceptual and empirical appropriateness of competing formulations. We will then consider the implications of these findings on survey design and analysis in general.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978
Schaeffer, Nora Cate
MR013 A General Social Survey Experiment in Generic WordsMethodological Report, Chicago, NORC , 1982
Variations in the wording of the child qualities items were examined in order to determine the degree of male bias present. This bias is present in both variations but to a lesser degree in the child item than the he item.
GSS years: 1980
Smith, Tom W.
MR012 The Subjectivity of EthnicityMethodological Report, Chicago, NORC , 1985
Objective measures of ethnicity and nationality may be inaccurate because substantial numbers of people do not know where their ancestors were born, have multiple nationalities, or do not adopt the ethnicity suggested by place of birth. Generally a combination of behavioral, natal, and subjective approaches is most effective, and even a simple subjective measure may be more effective than an objective one.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1980
Smith, Tom W.
MR011 Self-Employment: An Analysis of General Social Survey Measures of Employment StatusMethodological Report, Chicago, NORC , 1980
Conflicts found in the work supervision and self-employment items stem from: (1) borderline cases which include both elements, (2) answering the question for one's spouse rather than self, and (3) misinterpretation of the supervision question.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1980
Smith, Tom W.
MR080 A Comparison of Two Confidence ScalesMethodological Report, Chicago, NORC , 12, 1993
Response differences in a GSS scale and the ISSP scales result from measurement differences and group descriptors. Differences in questions wording, in particular, concerning descriptors for the government, produced different responses for subjects suggesting that the two labels were connoting different meanings.
GSS years: 1991
Smith, Tom W.
MR079 Notes on John Brehm, The Phantom Respondent: Opinion Surveys and Political Representation (Ann Arbor, MI: University of Michigan Press, 1993)Methodological Report, Chicago, NORC , 7, 1993
John Brehm incorrectly weighted the data when comparing the CPS and the GSS. This article shows that the GSS, when properly weighted, is very similar to the CPS in all demographics except for gender (See also Brehm, 324).
GSS years: 1978-1988
CPS 1978, 1980, 1982, 1984, 1986, 1988; NES
Smith, Tom W.
MR078 Little Things Matter: A Sampler of How Differences in Questionnaire Format Can Affect Survey ResponsesMethodological Report, Chicago, NORC , 7, 1993
Changes in the alignment of categories or the shift in scale can have significant response effects for surveys. The physical layout of surveys can also affect interviewers and hence data quality negatively through confusing skip patterns or the amount of physical space available corresponding to how much open ended detail is recorded.
GSS years: 1972-1993
ISSP 1987; Gallup 1954; Wirthlin; Gordon Black; ORC; Yankelovich and Hart-Tetter 1990; NORC 1954
Smith, Tom W.
MR077 The Relationship of Age to Education Across TimeMethodological Report, Chicago, NORC , 1992
(no abstract provided)
GSS years: 1972-1991
Smith, Tom W.
MR076 An Analysis of Response Patterns to the Ten-Point ScalometerMethodological Report, Chicago, NORC , 1992
Respondents with non-attitudes and middle-of-the-road attitudes are attracted to the +1 rather than the -1 response category on a ten-point scalometer. Endpoints, especially the negative endpoint, disproportionately attract responses. Changes in the scalometer, such as including 0 as a midpoint and pointing out all ten or eleven response categories, might cause fewer response effects and more accurate ratings.
GSS years: 1974-1991
Gallup 1953-1973
Smith, Tom W.
MR075 The 1992 General Social Survey ReinterviewMethodological Report, Chicago, NORC , 1992
This article compares the differences in the respondents to the 1991 GSS and the 1992 reinterview, and describes three differences. First, respondents to the reinterview were more upscale than the original interviewees. Second, non-response was higher among those who were uncooperative in the first interview. Third, non-response was higher among non-voters.
GSS years: 1991
Nakao, Keiko Treas, Judith
MR074 The 1989 Socioeconomic Index of Occupations: Construction from the 1989 Occupational Prestige ScoresMethodological Report, Chicago, NORC , 1992
Following Duncan's procedures new socioeconomic indexes are calculated based on the GSS/NORC study of occupational prestige.
GSS years: 1989
Census 1980
Smith, Tom W.
MR073 A Report on the GSS Household Enumeration VariablesMethodological Report, Chicago, NORC , 1992
This article describes the addition of 93 variables from the HEF for all full-probability surveys from 1975-1992.
GSS years: 1972-1991
Marsden, Peter V.
MR072 The Reliability of Network Density and Composition MeasuresMethodological Report, Chicago, NORC , 1991
Ethno-religious composition, political composition, density, and composition of a network by friends or co-members of organizations are measured with relatively high reliability, and some, such as sex composition, remain problematic, even when the number of alters grows quite large. The sensitivity of reliability estimates to differences in instrument design is examined using design variations in the surveys studied.
GSS years: 1985, 1987, 1988
Northern California Community Study 1977-1978
Smith, Tom W.
MR071 An Analysis of Missing Income Information on the General Social SurveysMethodological Report, Chicago, NORC , 1991
Missing information is a greater problem for income than for any other demographic and non-response has increased over time. However, non-response in the GSS is less than in many other studies
GSS years: 1972-1991
Census 1960, 1970; CPS 1973; Income Survey Development Program 1978; Survey Income and Program Participation 1983; NES 1988; ISSP 1989
Nakao, Keiko Treas, Judith
MR070 Computing 1989 Prestige ScoresMethodological Report, Chicago, NORC , 1990
The authors construct a new prestige scale for the 1980 Census Occupational classification. Using 1989 GSS data, 740 occupations were ranked according to social prestige.
GSS years: 1990
Nakao, Keiko Hodge, Robert W. Treas, Judith
MR069 On Revising Prestige Scores for All OccupationsMethodological Report, Chicago, NORC , 10, 1990
Prestige scores for all occupations developed from the national surveys in the 1960's have been widely used by researchers in the social sciences. The change in the 1980 Census classification of occupations necessitates updating the prestige scale accordingly. New scores can be obtained either by reworking the old scores or by collecting new data. In this paper, we argue for the latter choice based on the methodological, substantive, and theoretical considerations. The plan to collect occupational assessments from a nationally representative sample of 1500 Americans in the 1989 NORC General Social Survey will also be outlined.
GSS years: 1990
Smith, Tom W.
MR068 Discrepencies Between Men and Women in Reporting Number of Sexual Partners, A Cross-National ComparisonMethodological Report, Chicago, NORC , 9, 1990
(no abstract provided)
GSS years: 1988, 1989
Smith, Tom W.
MR067 An Analysis of Missing Data in the Study of Intergenerational MobilityMethodological Report, Chicago, NORC , 12, 1989
Missing data on father's occupation may have a small impact on intergenerational comparisons with intergenerational associations weaker for missing data. Possible corrections for missing data may include using mother's and spouse's work information.
GSS years: 1972-1988
Smith, Tom W.
MR066 Thoughts on the Nature of Context EffectsMethodological Report, Chicago, NORC , 9, 1989
Difficulty in studying order effects stems from the number of potential causal influences, competing explanations, interaction effects with question type, question specificity, question vagueness, question centrality, response type, history, administration, conflicting attitudes, and other effects. Promising solutions include split ballots, think aloud procedures, follow-up questions, probes on other dimensions, and various experimental designs.
GSS years: 1976, 1978, 1980
SRC 1979-80; NORC 1987; Greater Cincinnati Surveys 1983-84; DAS 1971
Smith, Tom W.
MR065 A Methodological Review of the Sexual Behavior Questions on the 1988 and 1989 GSSMethodological Report, Chicago, NORC , 10, 1990
Though most of the data on sexual behavior and attitudes from the 1988 and 1989 GSS appear valid and reliable, caution still needs to be use when examining the 1988 responses of male homosexuals and the gender gap in reported number of partners. (See also No. 2954??)
GSS years: 1988, 1989, 1990
Ligon, Ethan
MR064 The Development and Use of a Consistent Income Measure for the General Social SurveyMethodological Report, Chicago, NORC , 1989
Two new GSS income measures (REALINC and RINCOME) constructed from current GSS variables for household and respondent income correct for changes in the price level across years.
GSS years: 1972-1988
BLS 1983; CPS 1980
Alwin, Duane F.
MR062 Problems in the Estimations and Interpretation of the Reliability of Survey DataMethodological Report, Chicago, NORC , 2, 1989
In this paper I discuss several of the difficulties involved in estimating the reliability of survey measurement. Reliability is defined on the basis of classical true-score theory, as the
correlational consistency of multiple measures of the same construct, net of true change. This concept is presented within the framework of a theoretical discussion of the sources of error in survey data and the design requirements for separating response variation into components representing such response consistency and measurement errors. Discussion focuses on the
potential sources of random and nonrandom errors, including "invalidity" of measurement, the term frequently used to refer to components of method variance. Problems with the estimation
of these components are enumerated and discussed with respect to both cross-sectional and panel designs. Empirical examples are given of the estimation of the quantities of interest, which are the basis of a discussion of the interpretational difficulties encountered in reliability estimation. Data are drawn from the ISR's Quality of Life surveys, the National Election Studies and the NORC's General Social Surveys. The general conclusion is that both crosssectional and panel estimates of measurement reliability are desirable, but for the purposes of isolating the random component of error, panel designs are probably the most advantageous.
GSS years: 1973, 1974
Alwin, Duane F. Krosnick, Jon A.
MR061 The Reliability of Attitudinal Survey Data: The Influence of Questions and Respondent CharacteristicsMethodological Report, Chicago, NORC , 1989
(no abstract provided)
GSS years: 1973, 1974
Smith, Tom W.
MR060 Trends in Voluntary Group Membership Comments on Baumgartner and Walker – Measurement Validity and the Continuity of Results in Survey ResearchMethodological Report, Chicago, NORC , 2, 1989
(no abstract provided)
GSS years: 1973 -1988
Smith, Tom W.
MR059 Measuring Occupation: A Comparison of 1970 and 1980 Occupation Classification Systems of the Bureau of the CensusMethodological Report, Chicago, NORC , 12, 1988
Open ended coding errors for occupation are frequent, but fortunately mistakes do not differ from correct coding by much, and thus do not greatly affect analysis. More attention is needed in training coders and devising coding schemes.
GSS years: 1988
CPS
Smith, Tom W.
MR058 A Methodological Review of the Sexual Behavior Questions on the 1988 GSSMethodological Report, Chicago, NORC , 11, 1988
For the sexual behavior items on the 1988 GSS, there is little evidence of non-response bias and attitudes and behaviors appear somewhat consistent. Though reports of sexless marriages can reasonably be explained, the data on number of partners and male homosexuals is questionable. (See also No. 2953??)
GSS years: 1988
Ligon, Ethan
MR057 Rationale and Construction of Poverty Measures in the General Social SurveyMethodological Report, Chicago, NORC , 9, 1988
By the recoding of various demographics, GSS respondents can be classified according to the government definition of poverty.
GSS years: 1972-1988
Smith, Tom W.
MR056 Timely Artifacts: A Review of Measurement Variation in the 1972-88 GSSMethodological Report, Chicago, NORC , 9, 1988
Changes in GSS measurement procedures have distorted some trends in variables across time. These effects are identified, and in cases of extreme distortion, corrections are suggested.
GSS years: 1972-1988
Gallup 1976
Smith, Tom W.
MR055 Context Effects in the GSSMethodological Report, Chicago, NORC , 8, 1988
(no abstract provided)
GSS years: 1988, 1989
Rasinski, Kenneth A
MR054 The Effect of Question Wording on Public Support for Government SpendingMethodological Report, Chicago, NORC , 5, 1988
Past attempts at explaining the effect of question wording on responses to survey questions have stressed the ability of question wording to persuade and influence the respondent, resulting in attitude change. This paper promotes an alternative view, which is that even small changes in wording often shift the meaning of the question and thus affect the way the respondent things about the issue. Analyses of question wording experiments on the 1984, 1985, and 1986 General Social Surveys were conducted to examine the effect of wording changes on public support for various types of government spending. Consistent wording effects were found across the three years. An examination of the effects of wording changes and of their interaction with respondent individual differences led to two conclusions: (1) even minor wording changes can alter the meaning of a survey question, and (2), this effect is not limited to individuals with lower levels of education or with less stable attitudes
GSS years: 1984, 1985, 1986
Schuman, Howard Scott, Jacqueline
MR053 Replication of Two GSS Question Form Effect Experiments: 1972 and 1982Methodological Report, Chicago, NORC , 5, 1989
Two split-ballot experiments, one on DK filtering and one on agreeing response set, were included in the GSS in 1974 and replicated in 1982. Response effects occurred in each experiment in 1974 and were generally replicated in 1982, but the effects do not interact with time.
GSS years: 1972, 1982
Smith, Tom W.
MR052 Rotation Designs of the GSSMethodological Report, Chicago, NORC , 2, 1988
The GSS's switch from a rotation to a split ballot design offers advantages of maintaining one year intervals for variables, ease in judging rate of change in items, and applying econometric time series analysis and testing for context effects. Disadvantages include more sampling variability, complications in representation of time in analysis, and the possibility of introducing new context effects.
GSS years: 1972-1988
Smith, Tom W.
MR051 Random Probes of GSS QuestionsMethodological Report, Chicago, NORC , 2, 1988
(no abstract provided)
GSS years: 1984
Smith, Tom W.
MR050 Phone Home, An Analysis of Household Telephone OwnershipMethodological Report, Chicago, NORC , 8, 1987
(no abstract provided)
GSS years: 1973-1987
Smith, Tom W.
MR049 Inconsistent PeopleMethodological Report, Chicago, NORC , 1980
Test/retest consistency varies by attributes of the respondent, and this variation is largely a function of reliability differentials between groups.
GSS years: 1972, 1973, 1974, 1978
Smith, Tom W.
MR048 The Use of Public Opinion Data by the Attorney General's Commission on PornographyMethodological Report, Chicago, NORC , 8, 1987
The Commission's analysis of public opinion is methodologically unsound and therefore substantively suspect.
GSS years: 1972-1986
Gallup 1977, 1985; Yankelovich 1975 1976 1977 1982; Commission on Obscenity and Pornography 1970; Newsweek/Gallup 1985
Duncan, Greg J. Groskind, Fred
MR047 Some Methodological Aspects of Responses to the 1986 GSS Welfare Entitlement VignettesMethodological Report, Chicago, NORC , 5, 1987
Low benefit responses to the 1986 Welfare Vignette supplement are probably not due to misunderstanding the questions. It is also a mistake to assume that respondent-designated incomes were intended to be net benefits.
GSS years: 1986
Krosnick, Jon A. Alwin, Duane F.
MR046 Satisficing: A Strategy for Dealing With the Demands of Survey QuestionsMethodological Report, Chicago, NORC , 3, 1987
Respondents often choose merely satisfactory answers to survey questions when the cognitive and motivational demands of choosing optimal answers are high. Satisficing is more prevalent among people with less cognitive sophistication, though it is no more prevalent among people for whom the topic of a question is low in salience and/or personal importance.
GSS years: 1980
Smith, Tom W.
MR044 A Study of Non-Response and Negative Values on the Factorial Vignettes on WelfareMethodological Report, Chicago, NORC , 1986
Disinterest, lack of reading ability, difficulty in judgment, and comprehension lead to nonresponse to the welfare vignettes. Bias was small and related to nonresponse associated variables. Respondents also made marking mistakes. Despite these problems, the vignettes worked well with a small amount of error.
GSS years: 1986
Smith, Tom W.
MR044 A Study of Non-Response and Negative Values on the Factorial Vignettes on WelfareDisinterest, lack of reading ability, difficulty in judgment, and comprehension lead to nonresponse to the welfare vignettes. Bias was small and related to nonresponse associated variables. Respondents also made marking mistakes. Despite these problems, the vignettes worked well with a small amount of error.
GSS years: 1986
Smith, Tom W.
MR043 Classifying Protestant DenominationsMethodological Report, Chicago, NORC , 1987
(no abstract provided)
GSS years: 1972-1986
Smith, Tom W.
MR042 Attrition and Bias on the International Social Survey Program (ISSP) SupplementMethodological Report, Chicago, NORC , 1986
This paper discusses the use of survey supplements, factors influencing supplement attrition and nonresponse bias, and attrition and nonresponse bias on the 1985 ISSP Supplement. Overall supplement attrition was moderate and not random. Attrition was higher among those who are politically uninterested and less educated, less likely to discuss problems and socialize with others, Northeasterners, isolationists, and dislike foreign countries.
GSS years: 1984, 1985
OCG I & II 1962, 1973; BSA 1983-1985; Opinion and ISV 1976; Civil Liberties Survey 1978; NORC 1964
Burt, Ronald S.
MR041 A Note on Missing Network Data in the GSSMethodological Report, Chicago, NORC , 6, 1987
(no abstract provided)
GSS years: 1985
Burt, Ronald S.
MR040 A Note on the GSS's Ersatz Network Density ItemMethodological Report, Chicago, NORC , 6, 1987
(no abstract provided)
GSS years: 1985
Burt, Ronald S. Marsden, Peter V. Rossi, Peter H.
MR039 A Research Agenda for Survey Network DataBuilding on the GSS network items, the authors propose several changes and improvements which could be used to make a standard set of network items for survey research. This set would be efficient, reliable, and valid.
GSS years: 1985
DAS 1966; Northern California Communities Study 1977
Burt, Ronald S. Guilarte, Miguel G.
MR038 A Note on Scaling the General Social Survey Network Item Response CategoriesMethodological Report, Chicago, NORC , 1985
The idea of structural balance is used to suggest quantitative intervals between relationship strength response categories in the GSS network data. In contrast to an assumption of equal intervals between the categories of relationship strength, the intervals appear quite uneven.
GSS years: 1985
Burt, Ronald S.
MR037 A Note on Sociometric Order in the General Social Survey Network DataMethodological Report, Chicago, NORC , 6, 1986
The people identified as important discussion partners in the GSS network data were cited in order of strength of relationship with respondent: the first cited person having the strongest relation, the second having the next strongest, and so on. Order effects on closeness and contact frequency are described in the context of network size and relation content.
GSS years: 1985
Smith, Tom W. Peterson, Bruce L.
MR036 Problems in Form Randomization on the General Social SurveysMethodological Report, Chicago, NORC , 1986
An unintended overlap between respondent selection and form assignment procedures in GSS surveys from 1978 to 1985 created an association between form and age order in some households. This led to an association between form and various variables linked to age order. A weight was developed to compensate for the assignment bias and achieve random distribution of affected variables across forms.
GSS years: 1973-1985
Smith, Tom W.
MR035 An Analysis of the Accuracy of Spousal ReportsMethodological Report, Chicago, NORC , 1985
Overall proxy reports for spouses were as accurate as self-reports, probably because attributes measured (religion education, occupation, etc.) were major, basic demographics. Significantly higher levels of non-response were found for proxy reports, but a level of missing data was nevertheless negligible.
GSS years: 1972-1978, 1980, 1982-1985
Smith, Tom W.
MR034 Unhappiness on the 1985 GSS - Confounding Change and ContextMethodological Report, Chicago, NORC , 1985
Alteration of the GSS content by the addition or deletion of items, by the switching of items from permanent to rotating status, or by switching items from one rotation to another hampers keeping measurement conditions constant and therefore increases the possibility that true change will be confounded with measurement effects.
GSS years: 1972-1985
Smith, Tom W.
MR033 That Which We Call Welfare by Any Other Name Would Smell Sweeter: An Analysis of the Impact of Question Wording on Response PatternsMethodological Report, Chicago, NORC , 1985
The term welfare consistently produces more negative evaluations than does the term poor, illustrating the major impact different words can have on response patterns.
GSS years: 1984, 1985
SRC 1972, 1974, 1976, 1982; MAP 1968, 1982; Harris 1972 1976; Gallup 1976; Yankelovich
Burt, Ronald S.
MR032 Network Items and the General Social SurveyMethodological Report, Chicago, NORC , 1984
This is an argument for obtaining network data in the General Social Survey.
GSS years: 1984
Peterson, Bruce L.
MR031 Confidence: Categories and ConfusionMethodological Report, Chicago, NORC , 1985
This report on the 1984 GSS experiment comparing the effect of varying the number of response categories, concludes that the inter-item correlations are not appreciably different in the seven-point version of the confidence question than in the traditional three-point item.
GSS years: 1985
Smith, Tom W.
MR030 A Preliminary Analysis of Methodological Experiments on the 1984 GSSMethodological Report, Chicago, NORC , 1984
The report is a preliminary analysis of eight methodological experiments and adaptations in the 1984 General Social Survey: New Denominational codes; intra-item order effects, child qualities; sex of child, child qualities; spend priorities; confidence variation in response categories; bible fundamentalism, two trends; Images of God, two scales; and order effect of grace.
GSS years: 1972-1984
Gallup; ANES; NORC
Dempsey, Glenn R.
MR029 Measuring Political Ideology and Social Status: A Comparison of U.S. and European Scales Used on the 1983 General Social SurveyMethodological Report, Chicago, NORC , 9, 1984
The purpose of these two experiments on the 1983 GSS was to determine whether U.S. and European scaling techniques could measure political ideology and social status in the U.S. in similar ways. POLVIEWS tends to have stronger correlations with political and social attitudes than does POLVIEWX (European scale). CLASS, the standard GSS question also correlates higher than the European counterpart RANK.
GSS years: 1984
Smith, Tom W.
MR028 A Comparison of Telephone and Personal InterviewingMethodological Report, Chicago, NORC , 1984
(no abstract provided)
GSS years: 1972-1982
Smith, Tom W.
MR027 Children and Abortion: An Experiment in Question OrderMethodological Report, Chicago, NORC , 1983
When question order was reversed so that questions on valued qualities of children came before those on abortion, support for abortion decreased. Although a split ballot in 1983 failed to confirm the effect of altered question order, that may be the result of lower overall support of abortion.
GSS years: 1977, 1978, 1980, 1983
Alwin, Duane F. Krosnick, Jon A.
MR026 The Measurement of Values for Children: A Comparison of Ratings and RankingsMethodological Report, Chicago, NORC , 4, 1983
Ranking and rating techniques for measuring parental socialization values are found to be similar with respect to ordering aggregate preferences. However, ranked measures account for appreciably more variance in the latent variable, self-direction versus conformity.
GSS years: 1980
Durall 1946; Schuman and Presser, 1981
Smith, Tom W.
MR025 Using Temporary Refusals to Estimate Nonresponse BiasMethodological Report, Chicago, NORC , 10, 1984
(no abstract provided)
GSS years: 1980, 1982
Smith, Tom W.
MR024 Attitude Constraint as a Function of Non-Affective DimensionsMethodological Report, Chicago, NORC , 1982
Results from the 1982 GSS experiment show that non-affective dimensions such as importance, information, firmness, and open-ended questions added to issues like support/opposition to the ERA and abortion, and can discriminate the attitude constraint between two related measures.
GSS years: 1972-1982
Smith, Tom W.
MR023 An Experimental Comparison of Clustered and Scattered Scale ItemsMethodological Report, Chicago, NORC , 6, 1983
Clustering scale items together increases inter-item correlations, but has no clear impact between the scale and independent variables.
GSS years: 1973, 1974, 1975, 1976, 1977, 1978, 1980, 1982
Smith, Tom W.
MR022 Recalling Attitudes: An Analysis of Retrospective Questions on the 1982 General Social SurveyMethodological Report, Chicago, NORC , 1984
Clustering scale items together increases inter-item correlations, but has no clear impact between the scale and independent variables.
GSS years: 1972, 1973, 1982
Smith, Tom W.
MR021 Discrepancies in Past Presidential VoteMethodological Report, Chicago, NORC , 7, 1982
Voter turnout and candidate voted for are difficult variables to reliably measure. Voting is consistently over-reported and votes for winners are usually exaggerated.
GSS years: 1972-1982
ANES 1968, 1972, 1976, 1980; CPS 1968 1972, 1976, 1980
Smith, Tom W.
MR020 Conditional Order EffectsMethodological Report, Chicago, NORC , 5, 1982
Order-effects are an ill-known phenomenon in survey research. There are many different types with distinct causes. Conditional order effects in which the variation occurs mostly or completely among those giving a particular response to the antecedent question are examined in depth.
GSS years: 1976, 1978, 1980
Smith, Tom W.
MR019 Contradictions on the Abortion ScaleMethodological Report, Chicago, NORC , 1981
Respondents who contradict themselves on abortion items actually disapprove of abortion. The approving response to the general item is best considered an error in grasping the connection between the general and the situational items.
GSS years: 1977, 1978, 1980
Smith, Tom W.
MR018 Problems in Ethnic Measurement: Over-, Under-, and MisidentificationMethodological Report, Chicago, NORC , 1983
Ethnicity is the most difficult of all background variables to measure, as language, religion, race, nationality and culture must be pieced together. About one quarter of Americans are either over- or under-identifiers of their ancestors. The ability of ethnicity to explain attitudes drops with immigrant generation, though it remains significant even after several generations.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1980
SRC 1978; ANES 1978; Census of Canada 1971
Smith, Tom W.
MR017 Educated Don't Knows: An Analysis of the Relationship Between Education and Item NonresponseMethodological Report, Chicago, NORC , 1981
In general, item nonresponse is higher for the less educated. The reverse is true however on obscure and fictive questions without filters. With filters, the obscure and fictive questions show no association between item nonresponse and education.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978, 1980
ANES 1956, 1958, 1960, 1980
Smith, Tom W.
MR016 The Hidden 25 Percent: An Analysis of Nonresponse on the 1980 General Social SurveyMethodological Report, Chicago, NORC , 1981
Various methods of measuring the impact of non-response bias on survey estimates are examined. It is concluded that there is no simple, general, or accurate way of measuring it. Further research is encouraged.
GSS years: 1980
Smith, Tom W.
MR010 Situational Qualifications to Generalized Absolutes: An Analysis of 'Approval of Hitting' Questions on the General Social SurveysMethodological Report, Chicago, NORC , 1980
There is an apparent contradiction between the disapproving responses to the general hitting question and the more specific subquestions. This contradiction is due in part to differences in education and achievement.
GSS years: 1973, 1975, 1976, 1978
Smith, Tom W.
MR009 Sex and the GSS - Nonresponse Differences (1979)Methodological Report, Chicago, NORC , 1979
* Please note there is a version updated in 2009 (MR009a).
The problem of underrepresentation of males on the GSS reflects the nonresponse tendency of males, possibly exacerbated by female interviewers. Surveys using full probability sampling generally have an underrepresentation of males.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978
Census 1970, 1972-78; CPS 1975-77; CNS 1973-74; ANES 1972-78
Smith, Tom W. Bruce, C. Stephenson
MR008 Analysis of Test/Retest Experiments on the 1972, 1973, 1974 and 1978 General Social SurveysMethodological Report, Chicago, NORC , 1979.
The authors explain various techniques to determine measurement error in opinion surveys. Focusing on test/retest experiments, they conclude that the problems of distinguishing measurement error from true change are sufficiently fundamental and sufficiently complex that they must be attacked with various techniques.
GSS years: 1972, 1973, 1974, 1978
Stephenson, C. Bruce
MR007 Probability Sampling with QuotasMethodological Report, Chicago, NORC , 4, 1979.
Probability sampling with quotas (PSQ) overrepresents large households. Both PSQ and full probability sampling (FP) underrepresent people from large households. Also, PSQ underrepresents men who are working full-time. Finally, difficult respondents may be underrepresented more seriously in the PSQ sample. However, FP underrepresents men and urbanites.
GSS years: 1972, 1973, 1974, 1975, 1976
Smith, Tom W.
MR006 Ethnic Measurement and IdentificationMethodological Report, Chicago, NORC , 3, 1980.
Ethnicity is a difficult attribute to measure. It can be determined for about 78 percent of all non-blacks when measured subjectively and for about 85 percent when determined subjectively and natally. A lack of ethnic affiliation is related to being a member of the old stock, host culture; having low education and social standing; and poor transmission of family information between generations.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977
CPS 1972; SRC 1972, 1974, 1978
Smith, Tom W.
MR002 In Search of House Effects: A Comparison of Responses to Various Questions By Different Survey OrganizationsWhile house effects are not an insurmountable and pervasive survey problem, they do affect survey response particularly in the area of the don't know response level.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977
Stouffer 1954; NORC 1960; Gallup 1971-76 (14); Roper 1971, 1973; SRC 1972, 1974-76
Christian, Leah; Paddock, Susan; Blumerman, Lisa; Bautista, Rene; Davern, Michael
MR001 Can we have confidence in confidenceMethodological Report, Chicago, NORC , 1981.
Differences in survey procedures, i.e., format, wording placement, and order, artificially increase the variation of responses to questions on institutional confidence. Also, the concept of confidence is somewhat vague and allows for fluctuations that complicate an analysis of opinions on confidence. All in all, much of the inter- and intra-survey changes in trends are true fluctuations.
GSS years: 1972, 1973, 1974, 1975, 1976, 1977, 1978
