above were sub-categorized as games with simple rules, games with rules and symbolic games (Barnes, 2004; Pilten & Pilten, 2013). The drawings of the students were categorized accordingly and are shown in the tables. The demographic data are presented as percentages and frequencies. In a qualitative study, it is important in terms of achieving validity in the study to report the data in detail, included direct quotes from individuals and explain the results (Yildirim & Simsek 2011). An example of a drawing for each category was presented in tables based on mutual decision. The factors that affect students’ game preferences are determined by CHAID analysis, a decision tree method. The difference of CHAID analysis from other comparative analyses (such as t-test, ANOVA) is that it clusters the independent variable based on a certain dependent variable and starts a new clustering operation on the derived sub-clusters based on other independent variables. This way, while it analyzes the dependent variable in terms of the independent variables, it produces a result by evaluating various independent variable (Kilmen & Kosterelioglu, 2017). In general terms, as depicted by Kass (1980), decision trees are nonlinear methods that incrementally divide independent variables into smaller groups (Ture et al., 2005). CHAID analysis is an effective decision tree that repeatedly splits subsets of the space into two or more nodes, beginning with the entire data set (Michael & Gordon, 2004). To identify the best split at any node, any allowable pair of categories of the predictor variables is merged. The splitting continues until there is no statistically significant difference within the pair with respect to the target variable (Kass, 1980). CHAID is useful for analyzing a large number of predictor variables, and unlike similar statistical analyses, the CHAID algorithm does not require the data to be normally distributed or need assumptions such as homogeneity of the variants (Horner, Fireman & Wang, 2010). The only required assumption for CHAID analysis is to identify scale types of the predicted and predictor variables (SPSS, 2012). CHAID was preferred in this study because it also enables simultaneous analysis of nominal, interval and ratio scale data and demonstrates the relationships between the predicted and predicting variables in detail including all possible hierarchy (Yildiz, 2006; Horner, Fireman & Wang, 2010).