Analysis of Factors Influencing User Contribution and Predicting Involvement of Users on Stack Overflow
Abstract: Active involvement and participation of community members are essential for Q & A platform such as Stack Overflow, to make the platform richer and more vibrant. However, more than half of its users contribute to the community only once and then tend to disappear. This decreases the diversity of viewpoints and experience on the platform. This paper aims to identify the dominant factors that influence the withdrawal of new users from active participation after their first posts. We collected the responses received by the new users in the form of answers, comments, upvotes, downvotes and overall tone of the feedback (negative or positive) against the questions they posted on the platform and analyzed their impact on the users' ongoing participation state. The users have registered to Stack Overflow for two years and they were classified into three different groups based on the number of posts (low, medium and high number of posts) they created. Our analysis shows that, getting no or minimal response, receiving downvotes and getting condescending or demeaning comments from their peers discourage the new users from contributing in the future. The differences in responses between the three groups have been validated by performing one-way ANOVA and Pearson’s chi-square test, based on their participation rate as a between-subject factor. Utilizing the conclusive results from our study of the users’ responses, we trained a machine learning model using SVM classifier to predict whether a user is likely to post or not. The model performed well with an accuracy of 88.69 %. This exploratory and predictive approach will contribute in identifying and quantifying the underlying factors behind the decline in the active participation of new users on Stack Overflow.