Advice on mode effects in surveys from Understanding Society team

We’ve brought together advice on mode effects to help researchers

The new guidance has been prepared by survey methodologists working on Understanding Society following questions from researchers on how to take mode effects into account when analysing survey data. The guidance has been published in our Main Survey User Guide to explain mode effects, how they may affect data, and how to take them into account in your analysis. 

Mode effects come about because people’s answers to survey questions may differ depending on the way the survey is carried out.

In face-to-face interviews, for example, social desirability bias can affect the answers. This is due to the often unconscious desire of the respondent to be seen positively by the interviewer, or to hide any socially ‘undesirable’ behaviour or opinions. An online survey may encourage respondents to be more honest in their opinions, but the absence of an interviewer to motivate and help respondents if they are having difficulty may reduce the quality of the data.

Studies also have to take into account the fact that the sample of people who answer an online survey may be different to those who answer a face-to-face survey.

The new guidance goes into more depth on the different types of mode effect, and how Understanding Society has dealt with them – as well as a reading list about the testing we have done to investigate mode effects in the Study.

The guidance also gives some answers on how the COVID-19 pandemic may affect our data. We have had to suspend face-to-face interviews and invite all our sample members to take part online or by telephone.

Jon Burton, Understanding Society Associate Director, Surveys; Peter Lynn, Associate Director Survey Methodology; and Michaela Benzeval, Director of Understanding Society, have written a paper for Survey Research Methods on how Understanding Society adapted to the pandemic.

Read the mode effects guidance (scroll to bottom of page)

Read the Survey Research Methods paper