 We can concur that the 1993 data (retention.txt) does support the research findings in Druzdzel & Glymour 1994

 It can be deduced from the patterns that student retention is caused by average tests scores as inferred in the study

 We devised alternate temporal orderings, as inferred through descriptive statistics (later discussed in the report), we still found that student retention (apret) was steadily and directly caused by average tests scores (tstsc)

 Changing the significance levels does not affect the primary result structure of the pattern, thus we can still conclude that apret (average student retention rate) is causally and directly dependent on (tstsc) average tests scores

 By changing the significance levels, we get rid of other “latent common cause” connections, salar at p <= 0.51 and pacc at p < 0.248

 Data present in retention.txt can be almost said to be normally distributed and linearly dependent, because of a few skewed distributions (rejr, spend). The PC algorithm also assumes that the data is normally distributed, we verified the same by running descriptive statistics on it.