Optimal Contact Strategy in a Mail and Web Mixed Mode Survey Anders Holmberg, Boris Lorenc and Peter Werner Statistics Sweden NTTS Conference Brussels, Belgium February 2008
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Outline Introduction to the problem The design of the experiment Results Summary and conclusions
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Introduction to the problem Swedish individual and household surveys with Self- Administered Questionnaires (SAQ) The problem: low use of the web alternative –belief in rationality of the web mode (editing, other quality aspects, cost, environment) –cannot publicly afford the Web mode only –possibly, a mode switch (more of which later) –goal: investigating if changed contact strategies can increase the proportion of web respondents and a cost reduction, without harming the response rate
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Background PAP (mail Q) as the standard mode for SAQ –cover letter, questionnaire, return envelope, in some cases separate instruction, incentives. –response rates – as elsewhere – fall gradually Web as an alternative mode for SAQ –introduced in the 2000s for ind/hh surveys in Sweden –with the expectations of; easier input editing, higher response rate, cost reduction, and greater acceptance –carried out by including a paper sheet in the above, with an URL and login data Reminders
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Experiment groups TimeActionA1SA2A3A4 Day 1Mailout 1PP+WW* W Day 10Mailout 2p*p+WP+W W* Day 23Mailout 3P+W p+WP+W Day 38Mailout 4p+W P+W
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Results TrialA1SA2A3A4 # responses Sample size Response rate74,875,771,472,073,3 Web response rate 2,614,545,044,664,7
Response rate and web response rate for S, A2 and A4 proportion No of Days
European Conference on Quality in Official Statistics Rome, Italy July 11, of 17 Summary and Conclusions Web intensive contact strategies considerably increases the web response rate With a sample size of a cost reduction between 12% and 20% compared to the cost of the standard strategy can be obtained The effect on the response rate is very small Statistics Sweden are developing routines and will change standard to a more web intensive contact strategy.