Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

Supported by

Calculation of odds ratio

edited May 2016 in JASP & BayesFactor

Hi,
I used JASP to analyze a 2x2 cross table. However, I am wondering which formula is used to calculate the logOR. For the following numbers, logOR = 2.598 is given. If I put the numbers in a calculator like this one: https://www.medcalc.org/calc/odds_ratio.php and log(x) the result, I get 2.911225, which makes a great difference if one uses exp(x). Can you explain, where the formulas differ?

image

Comments

  • EJEJ
    edited 1:19PM

    Hi Duplex,

    The Bayesian analysis produces a posterior distribution for the log odds ratio. This distribution is induced by the posteriors for the model parameters. You are comparing the result to the outcome of a classical analysis (the classical analysis in JASP should produce the numbers you mention).

    Cheers,
    E.J.

  • edited 1:19PM

    Thanks, that makes sense. What do you suggest to report?
    One more question: Where can I find the script that calculates the 95% credible intervals for the plots of the Bayesian t-test?

  • EJEJ
    edited 1:19PM

    Hi Duplex,

    I would report whatever is consistent with your general mode of inference (classical vs Bayesian). Or report both. All code for JASP is available on GitHub. If you want help to locate the code more quickly I recommend you Email Tahira Jamil (http://www.uva.nl/en/about-the-uva/organisation/staff-members/content/j/a/t.jamil/t.jamil.html)

    Cheers,
    E.J.

  • edited 1:19PM

    Thank you very much. If you need help for an eventual German translation, I am glad to help.

Sign In or Register to comment.