What factors affects the speed of receiving accepted answers on Q&A websites?
Technical questions and answers (Q&A) websites accumulate a significant amount of knowledge from users. Developers are especially active on these Q&A websites, since developers are constantly facing new development challenges that require help from other experts. Over the years, Q&A website designers have derived several incentive systems (e.g., gamification) to encourage users to answer questions that are posted by others. However, the current incentive systems primarily focus on the quantity and quality of the answers instead of encouraging the rapid answering of questions. Improving the speed of getting an answer can significantly improve the user experience and increase user engagement on such Q&A websites
In this paper, we study the drivers for fast answers on such Q&A websites. Our goal is to explore how one may improve the current incentive systems to motivate fast answering of questions. We use a logistic regression model to analyze 48 factors along four dimensions (i.e., question, asker, answer, and answerer dimension) in order to understand the relationship between the studied factors and the needed time to get an accepted answer. We conduct our study on the four most popular (i.e., with the most questions) Q&A Stack Exchange websites: Stack Overflow, Mathematics, Ask Ubuntu, and Superuser. We find that i) factors in the answerer dimension have the strongest effect on the needed time to get an accepted answer, after controlling for other factors; ii) the current incentive system does not recognize non-frequent answerers who often answer questions which frequent answerers are not able to answer. Our findings suggest the Q&A website designers should improve their incentive systems to motivate non-frequent answerers to be more active and to answer questions fast, in order to shorten the waiting time to receive an answer (especially for questions that require specific knowledge that frequent answerers might not possess). The Q&A website designers should also consider improving their question delivery system to deliver the questions to non-frequent answerers sooner.
The factors that we study are listed as follows:
Results:
Using Regression Models to Study the Relationship Between the Studied Factors and the Speed of Getting an Accepted Answer
We build a random forest model on top 20% (fast) and bottom 20% (bottom) of the data and calculate the variable importance of each factors. The table below list the results.
There exists a strong relationship between the factors in the answerer dimension and the needed time to get an accepted answer. After controlling
for unchangeable factors, such as the length of the answer, the speed of how fast an answerer answers questions in the past is the most influential factor
in our model. Thus, in order to shorten the waiting time for an asker to get an accepted answer, Q&A website designers should make the incentive
system more sensitive to the speed of answering a question and spend more efforts on delivering the questions to the most suitable answerers as soon
as possible.
Understanding the Relationship Between the Answerer Community and the Speed of Getting an Accepted Answer
In general, non-frequent answerers answer questions slower than frequent answerers. However the questions that are answered by non-frequent answerers are as important as those that are answered by frequent answerers and such questions are usually more complex. 61.3%{86.9% of the questions that are answered by non-frequent answerers are slow-answered questions. Such slow-answered questions would have remained unanswered if they were not answered by the non-frequent answerers. The current incentive systems of the websites only motivate frequent answerers who usually tend to answer short questions, but not non-frequent answerers. Our findings suggest that Q&A website designers should improve the incentive system to attract the non-frequent answerers to be more active and answer questions fast (e.g., rewarding the non-frequent answerers more scores if they stay online for enough time) and improve the question answering incentive system to factor in the value and difficulty of the questions (e.g., providing additional rewards for harder questions or the questions that remain unanswered for long time).
To load the data, using the follow R code:
readRDS(datas, file ="4clearDataset")
readRDS(remainedXfeature, "4remainedFeaures")