Correlates of Response Latency on a Web Survey
Benjamin Phillips
Abt SRBI
Stanislav Kolenikov
Abt SRBI
Elaine Howard Ecklund
Rice University
Allison Ackermann
Abt SRBI
Ho Won Cheng
Abt SRBI
We examine factors associated with response latency in a web survey of scientists in biology and physics departments in Italian universities and research institutes. We found that respondents decrease their attention and start taking cognitive shortcuts with questions longer than about 100 words in question stem, as evidenced by changes in the slope of the latency vs. question length. We also found evidence for decreased attention shown by lower latency beyond the first 15 or so minutes of the survey, followed by respondents getting tired beyond 60 minutes of the survey. Opinion items had greater latency than factual items. Items requiring calculations, such as averaging, took longer than other items. Numeric and text entry items had greater latency than radio button items. Items in matrix (not measured separately) had higher latency than single items. Surveys taken in Italian (the native language of the survey population) had lower latency than those taken in English. These analyses provide important context for the perhaps simplistic interpretations of response latency: low latency being a desirable trait for items but undesirable for a respondent.