Home > Publications . Search All . Browse All . Country . Browse PSC Pubs . PSC Report Series

PSC In The News

RSS Feed icon

Owen-Smith says universities must demonstrate value of higher education

Armstrong says USC's removal of questions from a required Title IX training module may reflect student-administration relations

Fomby finds living with step- or half-siblings linked to higher aggression among 5 year olds

Highlights

PRB training program in policy communication for pre-docs. Application deadline, 2.28.2016

Call for proposals: PSID small grants for research on life course impacts on later life wellbeing

PSC News, fall 2015 now available

Barbara Anderson appointed chair of Census Scientific Advisory Committee

Next Brown Bag

Monday, Feb 1 at noon, 6050 ISR-Thompson
Sarah Miller

Question Difficulty and Respondents' Cognitive Ability: The Impact on Data Quality

Archived Abstract of Former PSC Researcher

Download PDF versionKnauper, Barbel A., Robert F. Belli, Daniel H. Hill, and A. Regula Herzog. "Question Difficulty and Respondents' Cognitive Ability: The Impact on Data Quality." AHEAD/HRS Report No. 96-038. July 1996.

There is increasing evidence that reductions in cognitive functioning can be negatively related to the quality of survey data. Research also indicates that the accuracy and completeness of reports decrease with increasing question difficulty. In the present paper the interaction of question difficulty with respondents' cognitive ability is investigated. It was expected that respondents with relatively low information processing ability are less able than those with higher ability to provide complete and accurate answers when responding to particularly difficul questions. The number of "don't know"-responses in a survey conducted with people over 70 years of age was used as an indicator of reduced data quality. The data were analyzed as a function of question difficulty and respondents' cognitive ability. The findings demonstrate the expected interaction: Respondents lower in cognitive ability were particularly likely to answer "don't know" to difficult questions. Respondents higher in cognitive ability were less affected by variations in question difficulty. The selective loss of data due to failures to respond may bias surveys findings because it leads to an underrepresentation of responses of lower cognitive ability respondents in the data. Implications for conducting surveys with lower cognitive ability respondents are discussed.

Dataset(s): AHEAD: U.S., 1993-94 (wave 1).

Browse | Search : All Pubs | Next