There is a growing interest in the study of elite behavior in political science (Kertzer and Renshon, Reference Kertzer and Renshon2022). Scholars of political representation, party politics, and legislative politics increasingly rely on surveys of elected politicians (e.g., Pereira, Reference Pereira2021; Sheffer et al., Reference Sheffer, John Loewen, Soroka, Walgrave and Sheafer2018) and bureaucrats (e.g., Bækgaard et al., Reference Bækgaard, Christensen, Mondrup Dahlmann, Mathiasen and Bjørn Grund Petersen2019; Brierley, Reference Brierley2020; Heinzel et al., Reference Heinzel, Weaver and Briggs2025) to test underlying assumptions of existing theories or to gather new insights into the motivations, attitudes, and behavior of policymakers. Elite surveys have become more common with the blurring lines between the study of political behavior and institutions in political science, the growing interest in subnational politics, and the normalization of survey experimental designs (Butler and Pereira, Reference Butler and Pereira2025; Walgrave and Joly, Reference Walgrave and Joly2018).

A key challenge of conducting surveys of political elites, as well as other hard-to-reach populations, is recruitment (Bailer, Reference Bailer2014; López, Reference López2023; Maestas, Neeley, and Richardson, Reference Maestas, Neeley and Richardson2003). Quantitative surveys with political elites increasingly rely on non-incentivized online surveys including survey experiments (e.g., Butler et al., Reference Butler, Volden, Dynes and Shor2017; Dynes, Hassell, and Miles, Reference Dynes, Hassell and Miles2023; Mayne and Peters, Reference Mayne and Peters2023; Lee, Reference Lee2022; Lucas et al., Reference Lucas, Sheffer and John Loewen2024; Pereira et al., Reference Pereira, Coroado, Sousa and Magalhães2025). Response rates in these online surveys are typically low and depend on the population targeted, the geographical scope, and the timing of fieldwork (Miller, Reference Miller2022; Krause et al., Reference Krause, Fatemi, Nguyen Long, Arnold and Hofmeyer2024; Vis and Stolwijk, Reference Vis and Stolwijk2021; Walgrave and Joly, Reference Walgrave and Joly2018). Low response rates can result in biased samples and underpowered designs, threatening the validity of descriptive and experimental scholarship. Some recent scholarship looks at how incentives in political elite surveys can impact sample sizes and survey responses (Butler and Pereira, Reference Butler and Pereira2018; Heinzel et al., Reference Heinzel, Weaver and Briggs2025). Sending formal invitation letters has a long history in elite studies and is still an established practice when soliciting face-to-face interviews with national politicians (Walgrave and Joly, Reference Walgrave and Joly2018). Formal letters add to the credibility of the research and this seems particularly needed in times when unsolicited emails are increasingly received with skepticism and either end up in the spam folder or Respondents don’t dare click on the included link because of cybersecurity concerns (Krause et al., Reference Krause, Fatemi, Nguyen Long, Arnold and Hofmeyer2024). Finally, earlier research suggests that for mass citizen surveys formal invitation letters—together with email reminders—have proven effective in boosting response rates but the effects are modest (Porter and Whitcomb, Reference Porter and Whitcomb2007; Kaplowitz et al., Reference Kaplowitz, Lupi, Couper and Thorp2012).

In this research note, we assess whether recruiting participants to an online survey via a postal letter increases response rates among local elected officials. We expect formal invitation letters to increase response rates through a number of complementary mechanisms. Letters signal the credibility of the research endeavor and overcome spam and security concerns specific to emails. Finally, given older generations are overrepresented in local offices, letter invitations may be more in line with the daily routine of the population studied.

Results

Figure 1 presents the effects of different recruitment modes on response rates to the online survey. We find that invitation letters produced a significant boost in response rates in both countries.Footnote ³ Response rates increased by 4.6 percentage points in Germany (s.e = 0.01) and by 2.1 points in the United Kingdom (s.e. = 0.004). Letters were significantly more effective at increasing response rates in Germany (difference in differences 0.025; ${p}$ -value = 0.01), despite the lower baseline response rate in the UK. This suggests important heterogeneity in the effectiveness of this recruitment strategy. When pooling results across countries, response rates improved by 2.8 percentage points (s.e. = 0.004).

Figure 1. Response rates in local politicians’ survey by recruitment model. Note: Bars describe response rates by randomly assigned recruitment modes in Germany, United Kingdom, and in both countries combined. Difference-in-means estimates from linear models (with country FEs in the pooled model) were reported over the bars for each group **( ${p}$ <0.001), *( ${p}$ <0.01).

But did the different recruitment modes impact the type of officials recruited, i.e., the characteristics of those taking the survey? Table 1 shows descriptive differences between subjects who completed the survey (post-treatment) and were recruited either by email or mail. We find that representatives recruited via letters tend to come from slightly smaller municipalities and are more likely to be men without a university education. There are no significant differences for the remaining observables. We also find no differences in response quality. Table 2 shows the effects of sending letter invitations on the quality of responses among those who participated in the survey. We find no relationship between recruitment mode and survey progress (column 1), whether officials finalized the survey (column 2), and time spent in the survey (column 3).

Table 1. Descriptives of survey respondents by recruitment mode

Table 2. The effects of recruitment mode on response quality

Note: *p ${\lt}$ 0.01; **p ${\lt}$ 0.001.

Cost-effectiveness of letter invitations

We have shown that invitation letters increase response rates of local politicians in Germany and the United Kingdom. However, this recruitment mode is costly. We paid 83 Euro Cent per letter (printing and stamp).Footnote ⁴ Table 3 shows the number of letters sent and responses obtained by country. The cost per response in Germany was around 16 Euros and increased to 37 Euros in the United Kingdom. Using within-country postal services or targeting specific subpopulations of politicians that are more responsive to letter invitations could improve the cost-effectiveness of this recruitment mode.

Table 3. Costs per completed survey by mail