The pros and cons of collecting data through self-report questionnaires

22218 views
Johanne Telnes Instanes
About the Author

Johanne Telnes Instanes, MD, PhD, is working 50% position at the University of Bergen, Norway. She is investigating the possible effect of mother´s diet during pregnancy on compulsive or impulsive behavioural traits in the child. The research is based on data from the Norwegian Mother, Father and Child Cohort Study (MoBa).


Data used in scientific research take on many different forms and are collected in multiple ways. One method I use in my research is collecting data from self-report questionnaires. This helps me to study whether a mother’s diet during pregnancy can be linked to impulsive or compulsive behavioural traits in her children. In this blog, I will explain to you the benefits of using such questionnaires, but also some of the limitations such data have.

In medical research, self-report questionnaires are frequently used to collect data. They have several advantages. In general, they are inexpensive and simple to administer, making it possible to collect a broad amount of data in a short time. Today, the possibility for online surveys has made data collection even easier. In addition, the results can be automatically collected, reducing the risk for errors occurring with manual registration processes. Further, the results are not dependent on an interviewer´s interpretation of behaviour, which may influence the results from a clinical interview.

Self-report questionnaires can be designed in multiple ways. There are many considerations to be taken when choosing the design and interpreting the results. Commonly, a vast majority of the questions in the questionnaire are closed, such as “yes” /”no” questions or Likert scale responses; typically “strongly disagree”, “disagree”, “neither”, “agree” and “strongly agree”. Such a design makes it easy to rate the answers and standardize the results. On the other hand, the answers are restricted. The participants may experience that none of the answers fit their opinions or situation. Using the Likert scale, some people tend to respond either to the extreme ends of the scale, others to the middle of the scale, and the researchers need to take this into consideration when interpreting the results. Another concern is when respondents are prone to give answers which they consider to be the most socially acceptable ones, so-called “social desirability bias”.

With respect to symptoms and functioning, the participant may lack insight into their own situation. Thus, information relying on the participant only may bias the results. Further, some respondents may get tired or lose interest after answering the first set of questions. Some may rush through the final questions, or not answer at all, thus decreasing the response rate.  Consequently, it is important to consider the length of the questionnaire – the results may be more valid if the questionnaire is short and to the point.

Another important consideration is the relevance of the questions for the specific participants of the survey. If the participant finds the topic interesting and relevant, they are more motivated to respond and complete all the questions.

In summary, it is feasible to collect data from self-report questionnaires, but researchers need to be aware of the limitations of using such data.  Ideally, the best solution is to combine information from several sources and methods, securing a broad base of information. For instance, information from population-based registries and clinical interviews can be combined with information from self-report questionnaires. Comparing or summarising the results based on different sources increases the generalisability of the results.

In the Eat2beNICE research project, we use data from the large population-based longitudinal Norwegian Mother, Father, and Child Cohort Study (MoBa). In this study, mothers reported on their diet during pregnancy using a special type of self-report questionnaire called “Food Frequency Questionnaire.” The mothers also filled out questionnaires about their child’s behaviour at several time points. The data from the questionnaires are linked with the Norwegian Medical Birth Registry. Some information, for instance on smoking, is registered both in MoBA and the birth registry itself. Thus, we can compare information on for instance smoking from two different sources, improving data quality in our analyses.

Visit our blogs for more information on research methods and data collection:
Is maternal excess weight or obesity prior to pregnancy a risk factor for ADHD? By Lin Li and Henrik Larsson.  
(About randomized controlled trials and quasi-experimental family-based designs)
https://newbrainnutrition.com/is-maternal-excess-weight-or-obesity-prior-to-pregnancy-a-risk-factor-for-adhd/

Food Frequency Questionnaire – what is it? By Berit Skretting Solberg.
(About food-frequency questionnaire)
https://newbrainnutrition.com/food-frequency-questionnaire-what-is-it/

How ambulatory assessment can help to monitor what someone eats throughout the day, by Elena D. Koch
(About ambulatory assessment – using smartphones to retrieve information on someone’s diet & activity throughout the day)
https://newbrainnutrition.com/how-ambulatory-assessment-can-help-to-monitor-what-someone-eats-throughout-the-day/

References:

Krosnick JA, Presser S. Chapter 9: Question and Questionnaire Design. In: Marsden P, Wright  J eds.  Handbook of Survey Research (second edition) Bingley, Emerald Group Publishing Limited; 2010: 263-313. https://web.stanford.edu/dept/communication/faculty/krosnick/docs/2009/2009_handbook_krosnick.pdf

McDonald JD. Measuring personality constructs: The advantages and disadvantages of self-reports, informant reports and behavioural assessments. Enquire. 2008;1(1):75-94. https://www.nottingham.ac.uk/sociology/documents/enquire/volume-1-issue-1-dodorico-mcdonald.pdf

Rolstad S, Adler J, Ryden A. Response burden and questionnaire length: is shorter better? A review and meta-analysis. Value Health. 2011;14(8):1101-1108. https://pubmed.ncbi.nlm.nih.gov/22152180/