Nonresponse Effects on Relationships Between Variables

To examine nonresponse effects on variable relationships, datafrom a panel study of 1,870 older Iowa residents are reported.When data are accumulated by wave of response, differences areminor on variable distributions (percentage, mean, standarddeviation), bivariate relationships (correlation), and multivariaterelationships (regression), suggesting that the assumed impactof nonresponse bias on variable relationships has been overstated.     

Further consideration of indicators of community attachment

Two studies of contradictory results concerning models of community attachment are briefly reviewed, one of which reported strong support for a ‘systemic’ model while the other placed greater emphasis on the relationship of size of place with attachment. Using data from 27 communities located in Iowa, measures used in both studies are replicated to the extent possible. Length of residence, age, and local social ties, important elements in the ‘systemic’ model of community attachment, are more strongly related to attachment than is size of place in this restudy.     

Older men in employed and retired families

Although much research on household work has focused on young dual-career couples, there has been speculation that retirement may have a significant effect on the arrangement of household activities in later life. Furthermore, involvement in the household is believed to influence psychological well-being. Available literature, however, provides contradictory evidence regarding the extent to which integration into household activities may either reduce or contribute to stress. The degree to which male involvement in the household in later life was stressful was examined in this research. Data were analyzed from structured interviews with 1165 older, married men who were employed full-time, part-time, or retired. Employed men were significantly less involved in the household than were the retired; however, men participated more in both masculine and feminine tasks if their wives were employed. Involvement in the household was not stressful although the benefits derived from participation varied somewhat by type of task.     

Community Attachment in a Rural Region1

Abstract Two models of community attachment in mass society—the linear development model and the systemic model—are reviewed. Based on data from respondents living in 27 communities in north-central Iowa, population size and density, key variables in the linear-development model, relate only weakly to community attachment as measured by social bonds and local sentiments. Length of residence, income, and age, which are variables in the systemic model, generally are more strongly related to the bonds and sentiments. Because samples were sufficiently large that data for each community could be analyzed independently, additional tests of the systemic model were possible; results indicate that differences occur among communities within an area thought to be relatively homogeneous.     

Interim Response to a Mail Questionnaire: Impacts on Variable Relationships*

Nagging doubts continue on the utility of the mail questionnaire, although advances in this technique provide response rates of 70 percent or higher. Return bias, a factor in attacks on the mail questionnaire, has been depicted as a necessary evil of such magnitude that the technique should be abandoned. But few studies using mail questionnaires have examined the impact of interim response on variable relationships. To test for return bias, data are reported from 931 residents of five small Iowa communities. Four mail requests elicited a 78.2 percent completion rate; interviews were attempted with terminal mail nonrespondents, yielding a final response rate of 92.8 percent. In a statistical sense, only minor differences are evident on correlation coefficients and unstandardized regression coefficients for the first and second response waves compared with data accumulated for all respondents. But when substantive conclusions are examined, the impact of interim response does not subside until the first three response waves are accumulated. Thus, attention must be given to the potential influence of return bias in mail-questionnaire studies, but calls for discontinuing use of the technique are unwarranted.