Re: Gamma Correlation pro's & con's
- From: Konstanze Krueger <Konstanze-Krueger@xxxxxxx>
- Date: Sun, 27 Nov 2005 11:52:53 +0100
John Uebersax schrieb:
In our data we have got lot's of ties. The data is ordinal, either left screwed or right screwed.I don't know if this applies to your situation, but if variables are ordered-category (i.e., just a few levels), rather than ordinal (ranks), then: (1) both the gamma statistic and Spearman's rho are affected by the large number of tied cases that result; and (2) the polychoric correlation coefficient avoids this. -- John Uebersax PhD
We defined the values to be 0,1,2 and 3. Until now the data has been analysed with Spearman Rank Correlation, but because of all the ties we are looking for a test which fits better.
Konstanze
.
- Follow-Ups:
- Re: Gamma Correlation pro's & con's
- From: Richard Ulrich
- Re: Gamma Correlation pro's & con's
- From: Ray Koopman
- Re: Gamma Correlation pro's & con's
- References:
- Gamma Correlation pro's & con's
- From: Knut Krueger
- Re: Gamma Correlation pro's & con's
- From: John Uebersax
- Gamma Correlation pro's & con's
- Prev by Date: Re: Comparing different scales
- Next by Date: Re: Gamma Correlation pro's & con's
- Previous by thread: Re: Gamma Correlation pro's & con's
- Next by thread: Re: Gamma Correlation pro's & con's
- Index(es):
Relevant Pages
|