The boxplot in figure 3 shows the difference in the distribution between the sample groups, where pass rider have higher median trip duration compared to membership riders median. The position of the box for the pass group is higher than the position of the box for the membership group, which suggest that the distribution of trip duration for pass rider is higher than distribution of trip duration for membership rider. This then supports the T-Test result in which the null is rejected in favor for the alternative hypothesis.
While the result of the T-Test works well and shows a significant different in the distribution of the two user type, it still have weakness. One main weakness is that it assumes a normal distribution of the samples and that the sample is collected randomly, which may not work well if the population is larger than the the one in this analysis and has non-gaussian distribution. Therefore, more statistical test is needed to have a more insightful understanding of the difference in the interaction between the two group.