DReMSS Webinar: Contact Letters and Envelope Design — Two Experiments on Survey Participation

Two presentations are on the programme :

-Effectiveness of personalised Contact Letters on Survey Participation 

Presenters : Tobias Ackermann (FORS) & Max Felder (FORS)

Abstract: Declining response rates strain the budgets of postal surveys. We test whether personalising invitation letters and reminders increases participation, using demographic information already in the sampling frame (sex, age, canton). In DDS21, a recurring national survey on Swiss direct democracy and voting behaviour (N = 10,786), we tested two dimensions of personalisation: a group-salience appeal, highlighting the respondent's own demographic group, and a tailored social norm. Group salience raised participation by about 3 percentage points, while the tailored social norm showed no reliable effect. To confirm this promising result, we also present the design of an upcoming replication in the next DDS21 wave.

-Effects of different logos on response and composition in a political probability-based web survey
Presenter : Anke Tresch (FORS &UNIL)

Abstract: Political surveys tend to underrepresent politically disengaged individuals and supporters of radical-right parties. These groups often have low institutional trust, which may reduce survey participation. In addition, survey features influence participation depending on their perceived relevance according to leverage-salience theory and the authority principle.

Using two experiments in the "Direct Democracy in Switzerland" project, we compare two logos on invitation envelopes, one signaling political content and the other the research institution, with a no-logo condition and with each other. The identical invitation letter displayed both logos, limiting the treatment contrast to receipt and opening. We examine survey starts, completion, conditional panel consent, and sample composition.
We found no evidence that either logo affected survey starts, completion, or conditional panel consent : only one out of twelve comparisons reached the 5% level before correction. Starting stage estimates were slightly negative, so a small deterrent effect cannot be ruled out. Overall, our findings show that logos on envelopes neither improve participation nor meaningfully harm it. Compositional differences were rare. The only robust difference was the topic-specific logo attracting a larger share of left-leaning respondents in one of the two surveys, which represented a shift in composition rather than an improvement in sample quality.