Inspite of the quality of my retention study, I give up on getting anything to come of it. I provide the paper here for anyone who's interested.  But I have failed to get it published or even to get my own school to act on it.  The first journal I sent it to refused to even submit to referees.  The editor said it was too "narrow."  I guess the material in that journal bears on more general topics that are more important than helping more students make it through college.  The next journal that I tried yielded on the first review one referee that liked the paper, but one who hated it.  The editor was sympathetic and sent it out again.  Two bad reviews this time.  I tried another journal where once again the editor refused to even have it reviewed.  He said that the correlations that I find must come from some omitted variable.  No suggestion of what that variable might be.  I have admit that even my closest colleagues are not impressed by what I've found.  They don't think that some students might be so weak in their study habits that class on Friday would make a difference.  They don't think that a five day class week would increase student's contact with advisors and peers in a meaningful way.  After I had the rejections described above, one of our deans told me that she liked my ideas and I invited her to coauthor the paper;  to strengthen references to the education literature.  She didn't want to do that, but she did provide me with good info about the literature.  I used this to put the paper into a the more scholarly form which is linked above.  And I asked the second journal to try it again.  The editor agreed but the readers didn't like it this time either.  So there I am.  The fact is that my idea about retention is unwelcome to the majority of folks in higher education.  Inspite of strong evidendce that I provide, they can find reasons to reject it.  It's true that my idea that a five day class week improves retention has only indirect support in the education literature.  And even a study such as mine with huge t stats and Rsquares can not prove that omitted variables aren't a factor.  (If this standard were applied generally there would be no regressions published.)  So I see no point in enlarging the sample.  Pointless work.  I am going on to other things.