UNDERSTANDING MAIL SURVEY RESPONSE BEHAVIOR A META-ANALYSIS

A meta-analysis of prior studies of techniques designed to inducemail survey response rates was conducted. Research encompassing184 effects (study outcomes) in 115 studies (articles) for 17predictors of response rate was examined. The average effectsize across all manipulations was r=.065, indicating an averageincrease of about 6.5 percent in response rates for manipulations.Effect sizes for specific predictors and two potential moderatorsof effects were examined. Results indicated that repeated contactsin the form of preliminary notification and follow-ups, appeals,inclusion of a return envelope, postage, and monetary incentives,were effective in increasing survey response rates. Significanteffect sizes for the predictors ranged from an increase in responseof 2 percent to 31 percent. Implications of the results forthe conduct of mail surveys and future research on mail surveyresponse behavior are discussed.     

Surname analysis for estimating local concentration of Hispanics and Asians

Surname analysis is a potentially useful technique for identifying members of particular racial, ethnic, or language communities within a population. We review the existing state of the art for identifying persons of Hispanic or Asian origin, based on surnames distinctive of each group. We describe the logic of surname analysis, profile several available surname dictionaries, and illustrate their applications in local redistricting. Results of our ongoing validation studies suggest promising future directions for improving accuracy and broadening applications.This article is based on a paper presented at the annual Population Association of America meetings at Cincinnati, 1 April 1993.     

Nonresponse weighting adjustment using estimated response probability

To reduce nonresponse bias in sample surveys, a method of nonresponse weighting adjustment is often used which consists of multiplying the sampling weight of the respondent by the inverse of the estimated response probability. The authors examine the asymptotic properties of this estimator. They prove that it is generally more efficient than an estimator which uses the true response probability, provided that the parameters which govern this probability are estimated by maximum likelihood. The authors discuss variance estimation methods that account for the effect of using the estimated response probability; they compare their performances in a small simulation study. They also discuss extensions to the regression estimator.