Re: GLM usage.

[John Darrington]

On Thu, Feb 05, 2015 at 05:39:46PM -0600, Shulin Yang wrote:
     
     I found that PSPP can conduct two-way anova by search, however nowhere to
     get how it does. I tried the Univariate analysis under Analyze menu, the
     result for my example dataset does not look right, the type iii ssq is 0
     and results in an infinity mean square.  Here is my data.
     
     weight    strain    dose
     106    1    1
     116    1    2
     145    1    3
     42    2    1
     68    2    2
     115    2    3
     70    3    1
     111    3    2
     133    3    3
     42    4    1
     63    4    2
     87    4    3

I am not an expert in this field, but I think the result is probably correct.  
The SSQ
is a measure of variance.  In your example, the weight data is distributed very 
uniformly over
all the factor variables.  Therefore, when all interactions are included, the 
variance is
very close to zero.
     
     I did manage to get correct result by running it without interaction term
     with command,
     GLM weight by strain dose
         /DESIGN strain dose
     
     Can anyone tell me the usage of GLM? Any other sub-commands I can use?
     Thank you very much.

It looks as if the subcommands currently available are 

/MISSING={INCLUDE,EXCLUDE}
/INTERCEPT={INCLUDE,EXCLUDE}
/CRITERIA=ALPHA(n)
/METHOD=SSTYPE(n)
/DESIGN
/SHOWCODES

(The last one is intended for debugging and not for general use)
     
     

-- 
PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See http://sks-keyservers.net or any PGP keyserver for public key.

signature.asc
Description: Digital signature