Love it as always, Alex.
One suggestion and one question.
Suggestion: Don't use significant (you know better, stats man!). Use substantial. Also, what was the cut off for the number of faceoffs?
Question: In mapping the trends, did you identify anyone who has a more prominent slope? In other words, who became a better faceoff centerman faster? Alternatively, who has had a slower decline?
Much obliged.
Yeh, you are right, substantial would have been more apt!
The cut off was the hardest part. I decided on 5 per game, as when the faceoffs per game are averaged this is about the most guys predominately playing wing ever have... Benn, Steen and Brouwer were pretty much the only 'wingers' over 5 per game last season... but none maintained that over career, and none predominantly centers, so none included in the sample. I would have preferred higher... but when I started I felt it could be unworkable as so few players come into the league taking a large number of draws, and many players play parts of seasons at wing... the sample would have been far smaller and simply a list of elite faceoff guys with high scoring centers as well. (and I wanted to try an include guys who were not so good on the dot) One or two guys seasons were listed though with say ~4.8-4.9 per game... (not perfect I know.) SPSS seemed to be ok with the diagnostics at that point... any lower and it started shouting at me!
Though if I were submitting it for a University assignment I must admit I would feel slightly uncomfortable with some of the individual player stats (one or two of the guys towards the lower end of faceoffs taken who had years while injured and missing half of the games, their yearly stats looked fine and all fitted with what the model would predict (within standard deviation etc)... but after I finished I thought if I were to do it again I would probably knock it down to ~60 or so players and increase the threshold.)... though happy and confident to a high level with the overall year to year stats.
As for the 2nd question... I shall get back to that tomorrow! My laptop I did the work on is in a different city!
Hey Apples, I was just curious, but how much time on average would you say you spend on writing an article (including background research, not just the act of writing it up)?
It depends. This one and the last one on defensemen and their maturation the statistics were kind of semi-long term hobbies. (which annoys me everytime... as by the time they are finished I have thought of numerous ways to make them better. My 'draft pick value' stuff I posted here a few years ago was the worst for that... by the time I was half way through I wanted to start over, but it had taken weeks of a few hours a night a few times a week compiling the data, so I kind of stuck with something that I knew could be better. That study is something I am certainly going to re-do when I have more free time... hopefully this summer.) Just did bits on them for a few months before I really thought of putting them into article form, just to satisfy my own curiosity. As for the writing part, usually one evening... I am the kind of person (even with Academic work) who cannot put something down and go away from it, come back and continue easily... so for example the Lindblom article was written up in a big ~3 hour long session... but I knew what I wanted to write before I started. (the Couturier and Mason ones were similar I suppose, though took slightly less time to write up.)