1. Glenn
    March 3, 2017 @ 4:50 pm

    I don’t mean to be critical, but, rather, provide constructive criticism. I like what you guys are doing, but there are severe statistical problems with this analysis. I get it – the season starts, so people want content, like rankings and stats, but an analysis needs to have some meat behind it to mean something. You said it yourself – very few observations. Disparity in opponents is going to be a huge influence, as well. Specifically, about your scatter plot of Avg Contributors/game v Scoring Margin, Marquette is an obvious outlier (maybe NJIT, and ABS(Margin)?), which influences the trendline. My bet is that without Marquette, it’s pretty flat. Is it significant (p)? If it is, how strong of a relationship (r-sq)?
    To be a total armchair QB, lol, look at last year’s data (is it significant?), then compare to what teams are doing this year. Particularly, has there been a change of philosophy that has given different results? Navy underperforming – are they using fewer players than last year? PSU, BU over-achieving – are they using more players? Is this something that normally changes throughout the season?
    Right now, all you really have is this sentence:
    “At the very least, this type of analysis gives us a peek into the mindset of the coaching staffs around the country.”
    Yes, I’m bored today, lol. I think you guys are on the right track, but don’t cheapen your product with crappy analyses.


    • Lacrosse Reference
      March 3, 2017 @ 5:32 pm

      Fantastic points, I really appreciate the comments. (I saw your thoughts on Twitter last night too, but had to get to bed.) I 100% agree with your comments about statistical rigor and that we’ve been a bit lax in adhering to the rules of quality analysis to date. Guilty; my college professors would fail me.

      The plan is to update these analyses in two ways: 1) which you suggested, is to look at last year’s data and see how it looks for a full year and 2) update it for 2017 as more game data becomes available. I’m especially interested to see if teams that get contributions from more players hold up better over the course of a season. Does that effect change depending on the leverage of the situations when the role players are making their contributions? I’d love to have a way to compare coaches and really try to isolate their philosophies on this topic. But your main point still stands, the conclusions drawn from any analysis can be actively detrimental if the analysis is flawed or not rigorous enough.

      Now that we’ve introduced the analysis and the underlying metrics, I hope you will continue to call us out for lax statistical rigor as future iterations of the analysis are published.


Leave a Reply