Showing papers by "Don A. Dillman published in 1998"

Development of a Standard E-Mail Methodology: Results of an Experiment

Abstract: Review of past E-mail surveys indicates that a method- ology to achieve consistently high response rates similar to those that can be obtained by traditional mail has not been developed. In addition, researchers have tended to use E-mail surveys only for populations with universal E-mail access. This study utilizes knowl- edge from past mail-survey research to develop an E-mail proce- dure. Further, an experiment is conducted to assess the potential for using a multimode strategy to obtain responses from individuals unreachable through E-mail. The multimode approach proved to be successful and techniques shown to be effective in standard mail surveys were also found to be appropriate for an E-mail survey.

Principles for Constructing Web Surveys

Abstract: Summary and Conclusions Use of the World Wide Web to conduct surveys provides enormous opportunities as well aschallenges. The cost advantages of collecting large amounts of data at very little cost means thatits use will escalate rapidly. At the same time, maintaining quality, whereby the results ofsurveys can be generalized to a large population is far from automatic.Respondent-friendly questionnaire design, found important to improving response to self-administered mail questionnaires, is also important for the development of web questionnaires.However, in a web context the term takes on a broader meaning. A respondent-friendly webquestionnaire is one that interfaces effectively with the wide variety of computers and browserspossessed by respondents. It also makes other aspects of the response task easy and interestingfor the respondent to complete. Respondent-friendly designed is aimed explicitly at reducingthree of the four types of error that typically prevent accurate surveys from being done, i.e.,nonresponse, measurement, and coverage of the survey population.Three important criteria for achieving respondent-friendly design have been stated. Theyinclude, first, compliancy with technology available to the respondent. If a browser cannotrecognize the programmed information being sent to it, the respondent cannot receive it.Questionnaires that use advanced programming that takes longer transmission times and requiresmore advanced browsers may result in unintended coverage and nonresponse error. A secondcriterion is the necessity of bridging between the logic by which respondents expect a question-naire to operate and the logic associated with operating a computer. It’s this connection thatprovides a questionnaire design challenge not previously faced by survey methodologists. Inaddition to these criteria, web designers must also consider the high likelihood that coverageproblems will require that most quality web surveys, for the foreseeable future, will require themixing of survey modes, whereby some respondents are surveyed by mail or telephone whileothers are surveyed via the web.