ASA Connect

 View Only

  • 1.  Stratified regression analysis

    Posted 02-12-2018 18:15
    Hi All,
    I am trying to respond to a reviewer comment on a paper we are working on. The reviewer states "Mentioning strategies for stratifying analysis would be helpful"​. In trying to respond to this comment, this raised a bunch of questions for me, which I am hoping someone can help me answer.​


    If I am building a multivariable regression model. How do I know when to conduct stratified regression models. Should this always be stated a priori?  If I find a significant interaction? Does that suggest a stratified analysis may be appropriate? Is there some other way to test when a stratified analysis is best? If one decided to build stratified regression model are there some rules or best practices for building these models?
    ​Any advice is appreciated.
    Thanks,
    MG​


    ------------------------------
    Melody Goodman
    Associate Professor
    New York University
    ------------------------------


  • 2.  RE: Stratified regression analysis

    Posted 02-13-2018 11:08
    Hi, Melody. You sound real close to wanting to try a hierarchical (or multilevel) model, such as you might do with Stan. See http://www.mc-stan.org/ for insights in the extensive manual as well as extensive examples.

    Bill
    --
    Bill Harris
    Data & Analytics Consultant
    Snohomish County PUD
    Everett WA
    (425) 783-1790


    Original Message


Related Content