Cup of Coffee

January 11th, 2013: Jackpot

21 Comments

Today I start a three-day weekend after about three months of never having two days off in a row. I should probably write, but I'll be fighting the urge to do film and videogame marathons all the way.

21 thoughts on “January 11th, 2013: Jackpot”

  1. Commenting today on yesterday's discussion of that Deadspin critique of a Grantland "Mortality Risk 'Study'":
    Grantland guy should try taking some actuarial classes before he jumps into those waters. Holy crap was that an awful way to structure things. That's worse use of statistics than Pitcher wins and RBIs (or whatever that one ESPN "Awesome Index" or whatever was 5 years ago). It also turns out that baseball has gotten safer over time. Many more players who played with Charles Radbourn have died than players who played with Tom Glavine.

    I don't actually work with life contingencies, but a typical study would be an actual/expected. And rather than measure surviving to today, measure total life-years. Get a typical contingency table (better yet, one that varies by year to reflect continuing mortality improvements, I can't tell you which would be a good one though). Measure the life-years of each player from [fifth year being in the majors/NFL] until present day. Then judge how many years these same players would have been expected to live from that point until today. Total those up and compare them. Maybe football players would have 1.3 times more life years than expected from a sample of average men of those ages, while baseball players have 1.25 more life-years. That would suggest that football players have better mortality than baseballers. Then apply statistical tests that I assume these writers seemed to know how to do properly.

    I'd also break it down into segments by 5-year bands of birth and 5-year bands of debut to see if anything is changing or skewing results.

    I don't understand why having played 5 years in the majors/NFL is the cutoff. Anyone making it to that point should have played that many years in the minors, college, high school, etc. Maybe go with a minimum of a season's worth of games. (But why even? Just do all.)

    There's probably some even better normalization that should be done, but this would be a good start. The Deadspin writer understood many of the problems in the Grantland "study", but didn't really know how to best measure it.

    One thing I thought of immediately wasn't expansion skewing the sample, but just that at any given time MLB players (who typically spend several years in the minors) on average are older than NFL players (who do not).

    1. What if it turns out that NFL players live just as long as the general population but less than other athletes? Do we then shrug off all the negatives and say football pluses and minuses even out?

      In this type of analysis, should they adjust for other circumstances such as race or socioeconomics? I think both affect life expectancy and the NFL doesn't mirror the averages across society.

      1. Right, that's what I meant about better normalization. If I worked in this area, I'd probably know which features would be best to pick from.
        My guess would be that pro athletes live longer than the general population, due to better health from being fit. Maybe a "preferred" life contingency table would be better for this. (Again, goes to my lack of experience.)

        Insurance industry tables won't give you racial breakdown due to non-discrimination laws. Scholarly articles may, though they might be lacking in "preferred" and other risk categories.

        Might also be good to pick apart cause of death: self-inflicted injury (including one's own drunk driving), injury inflicted by others, accident (plane crashes), cancer, heart, etc. There should be general-population measures for these. (Accidental Death tables would give some.) Recent suicides may make us think that Athletes have higher risk from that, but that may not be true).

        Overall, I think that while drawing comparisons between athletes and the general population might be tricky due to all of the confounding factors, I think that comparing different sets of athletes should be more more straightforward, and that I think a good way to do that would be to look at actual/expected life-years where "expected" is the general population (with age/sex/year reflected).

  2. figured i might as well dig the old hockey league out of the closet...

    ZOMG! fantasy hockey!!1
    League ID: 56045
    Custom League URL: http://hockey.fantasysports.yahoo.com/league/stickandpuckguys
    Password: wgom.org

    i sent email invites to everyone from last year, and anyone else is welcome to join. currently the draft is set for friday, 18 january 2012 (the season starts saturday) at 8:30pm CT, i believe. any requested change to this should be made sooner than later.

  3. Heard on regional news this morning:
    "And now we go to (reporter's name), who has found a man that has a car that frosted over!"
    "Sir, what can you tell us that happened?"
    "I got frost on my car."
    "Wow! Look at that! (Reporter starts furiously scratching frost off the roof of the car) That's amazing!!"

  5. the Wolves should start signing zombies.

    Joan Niesen ‏@JoanNiesen
    JJ Barea still isn't with the team. Earliest he will rejoin it is in San Antonio. And Lazar Hayward is sick. That makes 9 healthy players.

Comments are closed.