Kognitive Prozesse und Antwortverhalten in einer Internet-Befragung

Abstract

So far, certain problems of web surveys have received great attention: Coverage problems, sampling issues and response rates as well as screen design and usability issues. With some of the disadvantages mentioned diminishing, web surveys are increasingly adopted by survey researchers. In terms of the validity of web surveys early studies suggested that — controlling for other sources of survey error — little differences are to be expected in terms of response distributions. However, little insight is available regarding the cognitive processes at work in a respondent’s mind when answering a survey question on the internet. We assume that the cognitive capacity available when answering a survey question is impaired by the respondent’s navigational efforts when working through the questionnaire. Comparing the navigational tasks in a paper and pencil survey and a web survey, the later is considered to be more demanding compared with a paper questionnaire. Reading and flipping through pages is a task that can be done with little distraction from the question answer process. By contrast, browsing through an online form using mouse and buttons while reading is more demanding and distracting for the respondent. Thus, working through an online survey reduces the respondent’s attention for the question answer process. In this paper we present results from a large scale field experiment (n = 5,000) on measurement error in mixed mode survey (p&p and online) among the German youth conducted in spring of 2001. In this survey, 5 experiments on measurement error were incorporated: 1 experiment on question order, 2 experiments on response categories, and 1 experiment on response order in long lists and 1 experiment on social desirability. The results suggest that measurement effects are generally smaller in the web portion of the survey. In the conclusions the paper discusses the question whether or not we need to worry about the respondent’s redued atteniton to the survey questions under web conditions.