Who Should be Interviewed? A Response from Cluster Analysis

Objective: This article presents an application of cluster analysis for social sciences researches especially those studies that have an interview as part of their data collection. This application is more suitable for sequential mixed method researchers who use quantitative data to frame subsequent qualitative subsamples for conducting interviews. Method: In more detail, the algorithm (i.e., single linkage) employed for cluster analysis in this article is suitable for identifying the potential candidates for conducting interviews when the researcher is interested in outliers. Outliers provide interesting contrasts and distinction with other observations in a data set and are an interest for qualitative data analysis strategies. Results: The authors believe that cluster analysis is a better option than the traditional procedures for finding outliers (e.g., using explore or boxplot in SPSS) because cluster analysis finds outliers while considering different variables whereas the traditional methods has limitations and find outliers in respect to one variable. To present this application, first, cluster analysis and the single linkage which can be used for finding outlier data is presented. Then, a data set related to the psychology of learning mathematics was used to illustrate how outliers can be identified with cluster analysis via IBM SPSS 22. Conclusion: Finally, the results obtained from cluster analysis was interpreted. This is happened in order to explore whether the chosen algorithm for cluster analysis is accurate for finding suitable candidates for interviews.