Howdy, Stranger!

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

Supported by

[open] Priors and groups in contingency tables

Hello again,

Turns out there are still analyses for which I haven't inquired about priors.

So, contingency tables; is there an intuitive way to understand the prior concentration (in this case, Poisson sampling)? Unfortunately, Jamil et al. 2015 was not helpful to me in this regard.

Side-question: for directional hypotheses in a 2x2 table, apparently "group 1" is determined by columns (namely, as the left column)?

Thanks again for your patient and helpful replies,

Matti

Comments

  • EJEJ
    edited 3:16AM

    Hi Matti,

    About contingency tables: the interpretation of the prior concentration is the main suggestion for improvement in the review process (it has gotten "minor revision" at BRM). So we'll work on that. You can think of "a" as the vector of hypothetical prior observations in each of the categories. So if "a" is high, this means your H1 is relatively similar to H0. In the present version of JASP I believe the upper limit of "a" is only 2, but we are fixing that in the new release (out this week, I hope).

    With respect to the groups, I am not 100% sure but you should be able to tell what is what from the descriptives and the posterior plots.

    Cheers,
    E.J.

  • edited 3:16AM

    Hello,

    first I want to say how excited I am about JASP, never thought I could get enthusiastic about a statistics software but there you go ;)

    I am considering using bayesian contingency tables in my master's thesis and I'd also be very interested in the Jamil et al. publication (2015). I couldn't find it online, still doesn't seem to be published, so is there any way I could get my hands on it?

    Thank you so much in advance!

    Anna

  • EJEJ
    edited 3:16AM

    Hi Anna,

    You can find a preprint on my website:
    http://www.ejwagenmakers.com/inpress/JamilEtAlGunelDickeyinpress.pdf

    The paper is "in press" for Behavior Research Methods.

    Cheers,
    E.J.

  • edited 3:16AM

    Thank you very much!

Sign In or Register to comment.