E-voting and applicability in Türkiye

In both traditional voting systems and e-voting systems, it is important not only for the election process to be conducted transparently and reliably, but also for voters to perceive this process as trustworthy. In Turkey, little is known about whether voters trust the e-voting system, and a valid and reliable scale that applies at the national level is not available in the literature. In this context, the aim of the current research is to develop a valid and reliable scale to measure the perceived trust in e-voting for voters in Turkey, to assess how trust is perceived based on variables such as electoral district, education level, gender, employment status, income level, and age group, and to provide predictions about the applicability of e-voting in Turkey. In summary, this study focuses on voter trust regarding a potential e-voting system in Turkey. In the current study, all stages of Likert-type scale development were carried out to develop a perceived trust scale for e-voting. In the first stage, a pool of 46 items was created based on literature. A data collection tool was prepared consisting of two sections: Likert-type scale items and demographic data. The data for the study were collected in three phases: a pilot study, the main study, and a test-retest analysis. The research involved three separate participant groups: a pilot study group of 118 participants, a main study group of 400 participants, and a test-retest analysis group of 18 participants. A stratified sampling method was used to ensure that subgroups within the population were represented in the sample. Thus, the overall picture of voters in Turkey was reflected in the research. All electoral districts in Turkey (81 provinces), voters' gender, and age distribution were identified as subgroups and represented in the sample. In addition, education level, employment status, and income level were collected as variables from the participants. Following the pilot study, revisions were made to the data collection form based on expert feedback. To assess the reliability of the scale, Cronbach's Alpha was calculated, and item analysis (including average, standard deviation, and Cronbach's alpha when items were deleted), item discrimination index (using the top-bottom 27% difference test), and exploratory factor analysis were applied to the data from the main study. After the analysis, 6 items were removed, resulting in a 40-item scale. Discriminant validity was tested by examining the variation of the scale's dimensions and sub-dimensions according to demographic characteristics using both parametric (Independent samples t-test) and non-parametric (Kruskal-Wallis H test) methods. After confirming the construct validity, confirmatory factor analysis was conducted to validate the model, using the maximum likelihood estimation method. For the confirmatory factor analysis, standardized and unstandardized regression coefficients, critical values, and p-values were obtained. The test-retest reliability was measured using the intraclass correlation coefficient (ICC). As a result, a five-dimensional, 40-item perceived trust scale for e-voting was developed with high reliability and validity, with a Cronbach's alpha of .958 and an ICC of .970. The developed scale is named the "Perceived Trust Scale for E-Voting". The current study found that variables such as electoral district, education level, employment status, income level, and age group did not create significant differences in Turkish voters' perceived trust in e-voting. However, the gender variable showed a difference in the perceived trust scale measurements, with women scoring lower than men on all dimensions. Based on the responses to the validated Perceived Trust in E-Voting Scale, Turkish voters perceived a high level of trust in the economic dimension (x̄=3.93), political dimension (x̄=3.47), social dimension (x̄=3.21), and legal dimension (x̄=3.17) of e-voting, while the technological dimension (x̄=2.97) was perceived with moderate trust. The results of the study are discussed in light of the existing literature.