Stats 201: Binary logistic regression (or why your team didn't make the playoffs)

I suppose that it’s the mark of a true nerd that I actually have a couple of “favorite” statistical techniques.


10 Responses to Stats 201: Binary logistic regression (or why your team didn't make the playoffs)

  1. […] Did I mention numbers-based geekery already? Statistically Speaking’s infamous Pizza Cutter sends more your way, letting you know why your team didn’t make the playoffs using a post that’ll make you wish you stayed awake in calculus. […]

  2. John Beamer says:

    PC — what is the best software package to do this in? Keep up the good work?

  3. Pizza Cutter says:

    John, I use SPSS personally, although that’s a specialty program used in the social sciences for research (psychology, sociology). I actually don’t know much about other stats programs.

  4. Ryne says:

    For those OK with a non-GUI interface, R ( is a free statistical program useful for all kinds of models. For those interested, I believe binary logit regression can be accessed with the ‘glm’ command with the addition of “family=binomial()” after the model statement, as follows:
    glm(playoff~RS RA, family=binomial())
    SPSS, SAS and the like are great if you have access, usually through a university. Great work, PC.

  5. “I suppose that it

  6. Sorry, that should be:
    (1 / (1 + e^-x))
    Good work, by the way.

  7. Jim A says:

    I’ve found R to be a good free package for someone dabbling in statistical analysis without access to commercial software. It may take a while for a novice to figure out how to do something in it, but there are lots of docs and help forums. The book Baseball Hacks has a bunch of good introductory examples using R, also.

  8. Kyle J says:

    For someone who had some basic training in statistics but has forgotten a great deal of it, these articles are great–both for baseball purposes and professional purposes. Keep up the good work!

  9. Bhagwan says:

    I am a student of Lincoln University New Zealand. I am doing analysis of my research data. However, I am getting problem about the R2. Because in Binary logistic Regression Cox and Snell R 2 is very low. How can I increase R2? could you please suggest as soon as possible. Thanks

  10. Jason says:

    Actually, binary dependent variables do not follow the Poisson distribution. They follow the probit and/or logit distributions. And, the poisson distribution approximates normality as a function of the value of lambda, the mean rate of occurrence of the phenomenon. Since in terms of logical form lambda is equivalent to the logit, to hold that the distribution of a binary dependent variable can be modeled with Poisson regression would be to impose unacceptable constraints, since logistic regression does not have a homogeneity of variance assumption.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: