+ Reply to Thread
Results 1 to 5 of 5

Thread: Stata: Test difference in average RATE of two groups

  1. #1
    Points: 2,651, Level: 31
    Level completed: 34%, Points required for next Level: 99

    Location
    Europe
    Posts
    6
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Stata: Test difference in average RATE of two groups




    Question: I need to test the difference in average (!) rates of two groups. Just to illustrate:
    - Does the average rate of unemployed women in European countries differ from the average rate of unemployed women in US states
    (so there is within-group (between countries and between states) variation and between group variation (between the US and Europe) - I'm interested only in the last.

    "chitesti" by Cox seems to be a bit in the direction, but not what I need. Does anyone have a suggestion? I would be very grateful!!

  2. #2
    Points: 2,651, Level: 31
    Level completed: 34%, Points required for next Level: 99

    Location
    Europe
    Posts
    6
    Thanks
    0
    Thanked 0 Times in 0 Posts

    2-way ANOVA might serve?

    Some use a Two-Way ANOVA:
    http://jorthod.maneyjournals.org/cgi...t/full/29/1/31
    But this seems not to be entirely appropriate as it doesn't deal with rates, etc:
    http://people.richland.edu/james/lec.../ch13-2wy.html

  3. #3
    Points: 2,603, Level: 31
    Level completed: 2%, Points required for next Level: 147

    Location
    Philadelphia
    Posts
    23
    Thanks
    0
    Thanked 0 Times in 0 Posts
    It sounds like you have two independent sets of data, one of the unemployment rate for women in each European country [France, Germany, ...], and one for each US state [Alabama, Arkansas, ...]. You want to know if you can be confident that there is a difference between the means of each set. My text (Weiers) recommends an unequal-variances t-test, if you're comfortable with the assumption that the unemployment rates are normally distributed within each set. Excel can do this if you have the data analysis package installed, or probably any statistics program.

  4. #4
    Points: 2,651, Level: 31
    Level completed: 34%, Points required for next Level: 99

    Location
    Europe
    Posts
    6
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Thank you!!

    Thank you very much!! After reading some texts myself, I came to this conclusion as well, using the "ttest Y, by(X) unequal level(95)" Stata command (and the Welch option - giving basically the same result).
    Just checked your reply and was happy to have it confirmed
    Indeed, equal variance assumption rejected.
    Still thinking about if it is possible to do this with more categories (X-es) and whether things change with me using basically an entire (small, N=26) population.

  5. #5
    Points: 2,651, Level: 31
    Level completed: 34%, Points required for next Level: 99

    Location
    Europe
    Posts
    6
    Thanks
    0
    Thanked 0 Times in 0 Posts

    (m)anova


    Just for future readers of this post with similar problems:

    "anova Y X1 X2 X1*X2, regress" might well be an appropriate solution i you have more than 1 independent variables.

    If you have, in addition, more than one dependent variable:
    "manova Y1 Y2 = X1 X2 X1*X2"
    ttest is basically the same as twoway anova, I think at least.

    Corrections and comments are very welcome!!

+ Reply to Thread

           




Similar Threads

  1. difference in correlated groups
    By Deep in forum SAS
    Replies: 6
    Last Post: 07-16-2010, 09:59 AM
  2. Replies: 0
    Last Post: 03-16-2010, 03:36 PM
  3. Test for significant difference between groups
    By gianmarco in forum Statistics
    Replies: 0
    Last Post: 01-07-2010, 04:35 PM
  4. Replies: 2
    Last Post: 09-15-2009, 06:24 PM
  5. Replies: 3
    Last Post: 03-22-2009, 06:13 PM

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