Experimental Design Details
This field experiment consists of randomization of youth voters into a treatment group, which receives SMS (Short Message Service) reminders and into a control group. The target population of this study is eligible voters in Finnish 2023 parliamentary elections aged from 18 to 30 years old living in municipalities with an electronic voting registry, meaning that the outcome variable is available for these individuals. After combining information from the election registry and phone number availability from the ProFinder database, we are left with 49,866 individuals with a cellular phone number. We randomize these individuals into the treatment group (60% of the total sample) and into the control group (40% of the total sample). Below are the text messages which will be sent out for individuals in the treatment group both in Finnish and Swedish (translated to English):
1.Before the advance voting period
“Hi, a reminder for you that the parliamentary elections are held on the 2nd of April. The domestic advance voting period is from 22nd of March until 28th of March. Read more vaalit.fi. Best Regards, the Ministry of Justice”
2.Before the election day
“Hi, a reminder for you that the parliamentary elections are held on the 2nd of April. Read more vaalit.fi. Best Regards, the Ministry of Justice”
After the election the outcome variable from the electronic voting registry is merged into treatment status of the county and parliamentary election experiments and covariate data, and pseudonymised by Statistics of Finland. We will use the Linear Probability Model as our main model for the ease of the interpretation of the coefficients but will also conduct analysis using logit as a robustness check.
In order to estimate the dynamic effects, four randomized groups will be formed; individuals who got a SMS reminder in both elections, individuals who got a reminder in the previous county elections but not in upcoming parliamentary elections, individuals who did not get a reminder in the county elections but got a reminder in the parliamentary elections and individuals who did not get a SMS reminder in either election.
We will estimate potential heterogeneous effects by predicted voting propensity, predicted left-right voting, voting area party support, previous voting participation, geographical area (municipality type) and socioeconomic status (education and profession). In addition to that we will measure the primary outcome for individuals outside of the sample in order to explore spill over effects within the household, close by neighbourhood (by residential building code) and workplace.
We will predict a propensity to vote for every individual using the available administrative data utilising logit and elastic net models using control group individuals as the sample and voting in 2023 parliamentary elections as the dependent variable. Independent variables, measured before the treatment, for the prediction model are individuals’ gender, age, immigration background, logarithm of taxable income, educational background, SES background, eligibility to vote for the first time and municipality fixed effects. We will group the voting propensities by 25th, 25-75th, and top 25th percentiles in order to test possible heterogenous treatment effects between different voting propensity groups.
Left-right voting prediction is estimated based on surveys on voting in the parliamentary elections for youth voters and the survey sharing covariates which are observable in the registry data in order to get a left-right voting prediction for our sample individuals. In addition to that, we will employ honest causal forest machine learning algorithm in order to further explore possible heterogenous treatment effects.