TY - JOUR
T1 - Response heaping in interviewer-administered surveys
T2 - Is it really a form of satisficing?
AU - Holbrook, Allyson L.
AU - Anand, Sowmya
AU - Johnson, Timothy P.
AU - Cho, Young Ik
AU - Shavitt, Sharon
AU - Chavez, Noel
AU - Weiner, Saul
N1 - Publisher Copyright:
© 2014 The Author.
PY - 2014/9/1
Y1 - 2014/9/1
N2 - Response heaping (also referred to as rounding or digit preference) occurs when respondents show a preference for rounded numbers (often those divisible by five or 10). Conventional wisdom is that this is the result of taking cognitive shortcuts to make question answering easier, and as such, that it may be a form of survey satisficing. In four studies, we test this conventional wisdom for the first time by exploring whether response heaping occurs for five types of survey questions (behavioral frequency questions, questions that ask about an individual's personal characteristics, questions that ask about an individual's age at the time of an event, questions that ask the respondent to report a percentage, and feeling-thermometer attitude reports) under the conditions thought to foster survey satisficing (e.g., among respondents lower in ability and motivation, when the task of question-answering is difficult, and later in a long questionnaire) and whether heaped responses show effects of survey satisficing (e.g., shorter response latencies, less accuracy, and lowered predictive validity). We also examine the prevalence of response heaping and the extent to which heaping is associated across questions. Heaping above chance levels was found for most types of questions (although the prevalence of heaping varied systematically across different types of questions), but we found little evidence that heaping for most types of questions is more common under conditions thought to foster satisficing. In fact, heaping for some questions may actually reflect more thoughtful processes and result in higher data quality.
AB - Response heaping (also referred to as rounding or digit preference) occurs when respondents show a preference for rounded numbers (often those divisible by five or 10). Conventional wisdom is that this is the result of taking cognitive shortcuts to make question answering easier, and as such, that it may be a form of survey satisficing. In four studies, we test this conventional wisdom for the first time by exploring whether response heaping occurs for five types of survey questions (behavioral frequency questions, questions that ask about an individual's personal characteristics, questions that ask about an individual's age at the time of an event, questions that ask the respondent to report a percentage, and feeling-thermometer attitude reports) under the conditions thought to foster survey satisficing (e.g., among respondents lower in ability and motivation, when the task of question-answering is difficult, and later in a long questionnaire) and whether heaped responses show effects of survey satisficing (e.g., shorter response latencies, less accuracy, and lowered predictive validity). We also examine the prevalence of response heaping and the extent to which heaping is associated across questions. Heaping above chance levels was found for most types of questions (although the prevalence of heaping varied systematically across different types of questions), but we found little evidence that heaping for most types of questions is more common under conditions thought to foster satisficing. In fact, heaping for some questions may actually reflect more thoughtful processes and result in higher data quality.
UR - http://www.scopus.com/inward/record.url?scp=84928820244&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84928820244&partnerID=8YFLogxK
U2 - 10.1093/poq/nfu017
DO - 10.1093/poq/nfu017
M3 - Article
AN - SCOPUS:84928820244
SN - 0033-362X
VL - 78
SP - 591
EP - 633
JO - Public Opinion Quarterly
JF - Public Opinion Quarterly
IS - 3
ER -