Small Groups Survey, 1991 (Individuals Not in Groups Sample)
CitationWuthnow, R. (2021, July 13). Small Groups Survey, 1991 (Individuals Not in Groups Sample).
SummaryThis national survey was part of a three-year research project conducted to understand the small-group movement. "The national survey screened a representative sample of the American public to identify persons who were currently involved in any small group that met regularly and provided caring and support for its members. This procedure yielded approximately 1,000 people who were asked a long list of questions about the nature of their group, why they became involved, what its activities were, how well they liked it, and what they had received from it. For comparative purposes, we also surveyed more than 900 people to find out why they had not become involved in a small group" (Wuthnow, 1994:9).
This data file is the second part of the national survey on small groups and contains a comparative sample of those not involved in small group activity. The sample containing those involved in small groups is also available at the ARDA.
The ARDA has added six additional variables to the original data set to enhance the users' experience on our site.
Data FileCases: 962
Weight Variable: WEIGHT1, WEIGHT2
Taken from Robert Wuthnow, Sharing the Journey, published 1994, p. 368-369. The sample of completed interviews was weighted in two steps to bring the demographic characteristics of the final sample into alignment with the demographic characteristics of the continental U.S. adult population. The first step eliminated the disproportionality of group members in the final sample. Group members were oversampled for this study to ensure that the final sample would contain a sufficient number of completed interviews (at least 1,000) in that category. This oversampling results in persons in groups being disproportionately represented in the final sample. In this weighting step, these people are weighted down to their correct proportion as determined through the screening process. The second step used demographic and regional parameters from the U.S. Census Bureau's Current Population Survey as target parameters to bring the final sample of completed interviews into alignment with the regional distribution of population and the age, sex, education, and race distributions of the continental U.S. population of adults. The weighting variables have also been multiplied by a constant to ensure that the total number of weighted cases and the number of completed interviews (1,983) are the same. For results based only on group members, a different constant is used to ensure that weighted cases and the number of completed interviews with group members (1,021) are the same.
The chief undersampling that occurred in the data collection process was among men in groups, people in groups living in small towns and rural areas, people both in groups and not in groups with high school educations or less, and people both in groups and not in groups between the ages of 18 and 34. Conversely, oversampling occurred mainly among persons age 50 and older, among persons with college educations, among blacks, and for people in groups, among women, people living in the Midwest, and people living in large cities. The weightings adjust these proportions to more nearly approximate their distribution in the U.S. population.
Although total numbers of weighted and unweighted cases are the same among both those in groups and those not in groups, the weighting procedure necessitates adjusting discrepancies within the various subcategories. Because of these discrepancies, statistical measures of significance (such as Chi-square) are not always accurate. These inaccuracies are likely to be greatest when variables are being examined that may be systematically related to differences in age or education. All analyses discussed in the text that involve inferences about statistical associations are based on comparisons of relevant statistics using both weighted and unweighted numbers.
Data CollectionAll interviews were completed between November 4 and November 18, 1991.
Funded ByThe Lilly Endowment, Inc.
Collection ProceduresIn-person interviews conducted by Gallup Institute
Sampling ProceduresTaken from Robert Wuthnow, Sharing the Journey, published 1994, p. 368-369. The sampling procedure used for the survey followed standard Gallup Organization practices for the selection of a general adult population sample. This procedure is similar to that used in in-person surveys such as The Gallup Poll, a description of which can be found in all reports published by the Gallup Organization and from which the following wording has been adapted. The sampling procedure is designed to produce an approximation of the adult civilian population, eighteen years and older, living in the United States, except those persons in institutions such as prisons or hospitals.
The design of the sample is that of a replicated, probability sample down to the block level in the case of urban areas and to segments of townships in the case of rural areas. The sample design includes stratification by these seven size-of-community strata, using 1990 census data: (a) incorporated cities of population 1,000,000 and over; (b) incorporated cities of population 250,000 to 999,999; (c) incorporated cities of population 50,000 to 249,999; (d) urbanized places not included in (a-c); (e) cities over 2,500 population outside of urbanized areas; (f) towns and villages of less than 2,500 population; and (g) rural places not included within town boundaries. Each of these strata are further stratified into four geographic regions: East, Midwest, South, and West. Within each city size-regional stratum, the population is arrayed in geographic order and zoned into equal sized groups of sampling units. Pairs of localities are selected in each zone, with probability of selection and each locality proportional to its population size in the most current U.S. census, producing two replicated samples of localities.
Within each subdivision so selected for which block statistics are available, a sample of blocks or block clusters is drawn with probability of selection proportional to the number of dwelling units. In all other subdivisions or areas, blocks or segments are drawn at random or with equal probability. In each cluster of blocks and each segment so selected, a randomly selected starting point is designated on the interviewer's map of the area. Starting at this point, interviewers are required to follow a given direction in the selection of households until their assignment is completed.
A total of 160 representative sampling locations was drawn for the present survey. Interviewers assigned to each location were instructed to screen households until seven persons in groups and six persons not in groups had been interviewed. This procedure yielded a total of 1,021 interviews with individuals in groups and 962 interviews with individuals not in groups. All interviews were completed between November 4 and November 18, 1991.
Because the sample is stratified, each of the 160 sampling locations is assigned equal weight in determining the composition of the overall sample. The incidence of small-group members in each sampling location is determined by dividing the number of interviews completed with small-group members in that location by the total number of interviews completed in that location (interviews with group members, interviews with nonmembers, and interviews terminated as a result of the screening process). Across all sampling locations, the average incidence of group members is 40 percent.