Thesis Spotlights – Quality of Standardized Survey Research in Elderly with Cognitive Impairment
Patrick Kutschar, PhD, Austria
Email: patrick.kutschar [at] pmu.ac.at
Institution: Department of Sociology, Paris Lodron University Salzburg; Institute of Nursing Science and Practice, Paracelsus Medical University Salzburg
Time: 15 November 2015 – 24 June 2019
Supervisor: Ao. Univ. Prof. Dr. Martin Weichbold
Standardized surveys are regularly applied in sociological, gerontological, nursing and health sciences research examining elderly populations. Gathering high quality survey data among older persons with cognitive decline faces several methodological challenges. The consequences of age-associated cognitive, sensory and physical impairments have to be discussed critically against the background of fundamental assumptions of survey methodology. Information processing of persons with advanced age is affected substantially by deterioration in cognitive functions. Therefore, survey data quality is likely to be at risk.
As a part of this dissertation, survey errors among older respondents are systematised along the multidimensional Total Survey Error-Approach (TSE). Furthermore, population-specific response effects and bias are traced back to the cognitive stages of the survey response process. Survey error sources caused by specific characteristics of elderly respondents can be attributed to all TSE components (‘Respondent Selection Issues’, ‘Response Accuracy Issues’, ‘Survey Administration Issues’). Respondents’ decline in cognitive function and motivation interact with item difficulty, frequently leading to a less optimal information-processing of the cognitive stages whilst responding. If the question-answer-process is flawed in any way, measurement is erroneous and results in biased answers or no answers at all – in particular under the conditions that foster survey satisficing.
Item nonresponse, interviewer and context effects are analysed using data from two nursing homes (NH) studies (‘ABSM’, ‘PIASMA’) examining pain, depression and quality of life. More than 1,200 residents with up to moderate cognitive impairment were interviewed applying standardized face-to-face questionnaires. Respondent (cognitive function, dementia, age, gender, care level, pain presence) and item characteristics (question length, questionnaire length, filtering, answer option differentiation) affect item nonresponse rates significantly. Response distributions and item nonresponse vary considerably by interviewer gender and affiliation (NH-intern vs. NH-extern). The order of items within selected scales causes assimilation effects, while primacy as well as recency effects are observed due to the positioning of answer categories in dichotomous items. These errors are moderated substantially by the residents’ cognitive function. Effect directions as well as effect sizes vary systematically according to the grade of cognitive decline.
Findings show that elderly persons with up to moderate decline in cognitive function can be interviewed using standardized self-report instruments. However, the quality of the collected data is subject to limitations in several dimensions. This research clearly supports the need for a population-specific survey theory in order to enhance data quality and to include older persons with cognitive decline into survey research adequately.
Kutschar P. & Weichbold M. (2019). Interviewing elderly in nursing homes – Respondent and survey characteristics as predictors of item nonresponse. Survey Methods: Insights from the Field.
Kutschar P., Weichbold, M. & Osterbrink, J. (2019). Effects of age and cognitive function on data quality of standardized surveys in nursing home populations. BMC Geriatrics.
Comment on this article – log in with your ESA username and password: a comment field will appear.