Advance Correspondence and Error in Mail Surveys

The results of this experimental study suggest that advance correspondence produces a higher response rate, but at a higher cost per usable return, than using no advance correspondence. Two forms of advance correspondence, postcard and letter, produced approximately equal response rates but the advance postcard had a lower cost per usable return. The findings are discussed in the context of mean square error to illustrate that the researcher ordinarily must be concerned about both response rate and number of returns when designing a survey.