Testing methods to increase response to questionnaires in a longitudinal study

Michaela Goodwin, Tanya Walsh, William Whittaker, Michael Kelly, Eric Rooney, Richard Elmsley, Matt Sutton, Martin Tickle, Iain A. Pretty

Research output: Contribution to conferenceAbstract


Introduction: Non-response or non-consent can result in selection bias and affect the validity of results. Results are presented from the recruitment and first year of data collection from a longitudinal birth cohort study, examining different methods used to increase response rate.

Methods: Participants were recruited from two hospitals in England, UK. Baseline questionnaires were sent out to consented parents following the birth of their child, through email or by post. Following low response, methods to increase response rate included shortening the questionnaire and an alternative cover letter and free pen. Families who did not return the questionnaire were randomly allocated to receive a reminder by phone, postcard or an additional questionnaire.

Results: A 70% consent rate and overall 47% response rate has been achieved. Consent and response were negatively associated with deprivation (IMD) (consent M=25.1, 95%CI 24.5-25.7 and non-consent M=28.1, 95%CI 27.1-29.1, t(2843)=5.018, p=0.001) (responders M=22.3 95%CI 21.4-23.2 and non-responders M=28.4 95%CI 27.4-29.4 t(1701)=9.011, p<0.001).

Before any changes were implemented non-response was an issue with only 25% of questionnaires returned from the first attempt and 46% overall (including second consents).

Following the alternative cover letter and questionnaire, initial response rose to 38%. Resending the entire questionnaire resulted in the highest response (25%) compared to a postcard reminder (19%) and phone call (16%). The response rate (including second attempts) given the new methods was 49%. Following the changes there remained a significant difference in IMD between responders (m=22.0 95%CI 20.9-23.1) and non-responders overall (m=28.4 95%CI 27.1-29.6) t(1085)7.598,p=0.0001.

Discussion: Those who were more deprived (according to their IMD) were less likely to consent or complete questionnaires. Resending the entire questionnaire was the most effective method in increasing response but not reducing the difference in IMD. The most common reasons given for non-response were; misplaced questionnaire, chaotic lifestyle (new-born) or moving house.
Original languageEnglish
Publication statusPublished - 17 May 2016
EventSCT Annual Meeting: Thomas C Chalmers Student Scholarship Entry - Fairmont The Queen Elizabeth, Montreal, Canada
Duration: 15 May 201618 May 2016


ConferenceSCT Annual Meeting


Dive into the research topics of 'Testing methods to increase response to questionnaires in a longitudinal study'. Together they form a unique fingerprint.

Cite this