+ Reply to Thread
Results 1 to 2 of 2

Thread: hausman test

  1. #1
    Points: 4, Level: 1
    Level completed: 7%, Points required for next Level: 46

    Thanked 0 Times in 0 Posts

    hausman test

    Dear community, I analyse purchase decisions on different products j over a duration of t days. I would like to include random or fixed effects. When I run a Hausman test and include only the time variant variables, then random effects are favored rejecting fixed effects by prob>chi2 =0.075. When I include my time invariant variables such as product category (which is constant over panel=product j), then prob>chi2= 0.000. However, some time invariant variables are dropped and others not in the fixed effect model. I am now wondering whether to use random or fixed effects. I thought it makes more sense to include only the time variant variables as usually all the invariant variables should be excluded by stata. I appreciate any recommendations. Best

  2. #2
    Points: 4,664, Level: 43
    Level completed: 57%, Points required for next Level: 86
    kiton's Avatar
    Corn field
    Thanked 51 Times in 46 Posts

    Re: hausman test


    Hausman test is surely a proper test to decide on the fixed or random effects. Yet, there is also theory part to it -- (a) what is the nature of the variables in the model (i.e., correlated with omitted or not), (b) are there any omitted variables that you are trying to correct for, (c) are you research questions related to estimation of time-invariant effects (if not, then inclusion of time-invariant variables in the fixed effects may be meaningless), etc.? These and some other notions are important on deciding on the effects, not hausman test alone. I suggest you find some articles on fixed vs. random effects by Paul Allison, he explains everything amazingly good. Good luck

+ Reply to Thread


Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts

Advertise on Talk Stats