syntax for testing significant difference b/w two alphas

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

syntax for testing significant difference b/w two alphas

Jims More
I got a syntax (pasted below) for testing a significant difference between two independent cronbach alphas.  However, it doesnot work.  I dont know the format of the input data.  Your suggestion is highly solicited. Jims
 
/* This program computes a test of the significance of the differences */.

/* between k independent coefficient alphas using a procedure described by */.

/* Hakstian & Whalen (1976).  The syntax is HWALPHA alpha n j, where alpha */.

/* is a variable in an SPSS file containing the k values of alpha, n is the */.

/* variable in the file containing the sample sizes, and j is the variable */.

/* containing the number of items in the measure */.

/* Written by Andrew F. Hayes */.

/* The Ohio State University */.

/* School of Journalism and Communication */.

/* [hidden email] */.

 

 

DEFINE HWALPHA (!positional !token(3)).

MATRIX.

get mat/variables = !1/missing = omit.

compute alpha = mat(:,1).

compute n=mat(:,2).

compute j=mat(:,3).

compute df = nrow(mat)-1.

compute a = ((j-1)&*((9*n-11)&**2))/((18*j)&*(n-1)).

compute top = a&*((1-alpha)&**(-1/3)).

compute bot = a&*((1-alpha)&**(-2/3)).

compute m = csum(a)-((csum(top)**2)/csum(bot)).

print/title = "** SIGNIFICANCE TEST OF THE DIFFERENCE BETWEEN INDEPENDENT ALPHA COEFFICIENTS **".

print/title = "From Hakstian & Whalen (1976), Psychometrika, 41, 219-231.".

print mat/clabels = "ALPHA" "N" "ITEMS"/title = " "/format F10.4.

compute p = 1-chicdf(m,df).

compute res = {m, df, p}.

print res/clabels = "CHISQ" "DF" "P-VALUE"/title = " "/format F10.4.

END MATRIX.

!END DEFINE.

 

 


Stay connected to the people that matter most with a smarter inbox. Take a look.
Reply | Threaded
Open this post in threaded view
|

Re: syntax for testing significant difference b/w two alphas

Mario Giesel
Hello, Jims,
 
The last command should be "!ENDDEFINE." - without a blank?!
 
Good luck,
Mario


Von: Jims More <[hidden email]>
An: [hidden email]
Gesendet: Samstag, den 21. Februar 2009, 23:12:08 Uhr
Betreff: syntax for testing significant difference b/w two alphas

I got a syntax (pasted below) for testing a significant difference between two independent cronbach alphas.  However, it doesnot work.  I dont know the format of the input data.  Your suggestion is highly solicited. Jims
 
/* This program computes a test of the significance of the differences */.

/* between k independent coefficient alphas using a procedure described by */.

/* Hakstian & Whalen (1976).  The syntax is HWALPHA alpha n j, where alpha */.

/* is a variable in an SPSS file containing the k values of alpha, n is the */.

/* variable in the file containing the sample sizes, and j is the variable */.

/* containing the number of items in the measure */.

/* Written by Andrew F. Hayes */.

/* The Ohio State University */.

/* School of Journalism and Communication */.

/* [hidden email] */.

 

 

DEFINE HWALPHA (!positional !token(3)).

MATRIX.

get mat/variables = !1/missing = omit.

compute alpha = mat(:,1).

compute n=mat(:,2).

compute j=mat(:,3).

compute df = nrow(mat)-1.

compute a = ((j-1)&*((9*n-11)&**2))/((18*j)&*(n-1)).

compute top = a&*((1-alpha)&**(-1/3)).

compute bot = a&*((1-alpha)&**(-2/3)).

compute m = csum(a)-((csum(top)**2)/csum(bot)).

print/title = "** SIGNIFICANCE TEST OF THE DIFFERENCE BETWEEN INDEPENDENT ALPHA COEFFICIENTS **".

print/title = "From Hakstian & Whalen (1976), Psychometrika, 41, 219-231.".

print mat/clabels = "ALPHA" "N" "ITEMS"/title = " "/format F10.4.

compute p = 1-chicdf(m,df).

compute res = {m, df, p}.

print res/clabels = "CHISQ" "DF" "P-VALUE"/title = " "/format F10.4.

END MATRIX.

!END DEFINE.

 

 


Stay connected to the people that matter most with a smarter inbox. Take a look.

Mario Giesel
Munich, Germany
Reply | Threaded
Open this post in threaded view
|

Re: syntax for testing significant difference b/w two alphas

Art Kendall
In reply to this post by Jims More
try this. replace the last line with this syntax.

if the original article has an example you might want to try it to verify the it works as you expect..

!ENDDEFINE.

data list list /test(f1) myalpha(f4.3) samplesize(f5) n_items (f5).
begin data
1 .981 101 13
2 .892  50 13
3 .855  25 13
4 .722  20 13
end data.

hwalpha myalpha samplesize n_items.

Art Kendall
Social Research Consultants

Jims More wrote:
I got a syntax (pasted below) for testing a significant difference between two independent cronbach alphas.  However, it doesnot work.  I dont know the format of the input data.  Your suggestion is highly solicited. Jims
 
/* This program computes a test of the significance of the differences */.

/* between k independent coefficient alphas using a procedure described by */.

/* Hakstian & Whalen (1976).  The syntax is HWALPHA alpha n j, where alpha */.

/* is a variable in an SPSS file containing the k values of alpha, n is the */.

/* variable in the file containing the sample sizes, and j is the variable */.

/* containing the number of items in the measure */.

/* Written by Andrew F. Hayes */.

/* The Ohio State University */.

/* School of Journalism and Communication */.

/* [hidden email] */.





DEFINE HWALPHA (!positional !token(3)).

MATRIX.

get mat/variables = !1/missing = omit.

compute alpha = mat(:,1).

compute n=mat(:,2).

compute j=mat(:,3).

compute df = nrow(mat)-1.

compute a = ((j-1)&*((9*n-11)&**2))/((18*j)&*(n-1)).

compute top = a&*((1-alpha)&**(-1/3)).

compute bot = a&*((1-alpha)&**(-2/3)).

compute m = csum(a)-((csum(top)**2)/csum(bot)).

print/title = "** SIGNIFICANCE TEST OF THE DIFFERENCE BETWEEN INDEPENDENT ALPHA COEFFICIENTS **".

print/title = "From Hakstian & Whalen (1976), Psychometrika, 41, 219-231.".

print mat/clabels = "ALPHA" "N" "ITEMS"/title = " "/format F10.4.

compute p = 1-chicdf(m,df).

compute res = {m, df, p}.

print res/clabels = "CHISQ" "DF" "P-VALUE"/title = " "/format F10.4.

END MATRIX.

!END DEFINE.




        


Stay connected to the people that matter most with a smarter inbox. Take a look.
Art Kendall
Social Research Consultants
Reply | Threaded
Open this post in threaded view
|

Re: syntax for testing significant difference b/w two alphas

mphilramesh
In reply to this post by Jims More
Hi,


Please send the SPSS syntax of the covariate meta-analysis...
Jims More wrote
I got a syntax (pasted below) for testing a significant difference between two independent cronbach alphas.  However, it doesnot work.  I dont know the format of the input data.  Your suggestion is highly solicited. JimsĀ /* This program computes a test of the significance of the differences */.

/* between k independent coefficient alphas using a procedure described by */.

/* Hakstian & Whalen (1976).  The syntax is HWALPHA alpha n j, where alpha */.

/* is a variable in an SPSS file containing the k values of alpha, n is the */.

/* variable in the file containing the sample sizes, and j is the variable */.

/* containing the number of items in the measure */.

/* Written by Andrew F. Hayes */.

/* The Ohio State University */.

/* School of Journalism and Communication */.

/* [hidden email] */.

 

 

DEFINE HWALPHA (!positional !token(3)).

MATRIX.

get mat/variables = !1/missing = omit.

compute alpha = mat(:,1).

compute n=mat(:,2).

compute j=mat(:,3).

compute df = nrow(mat)-1.

compute a = ((j-1)&*((9*n-11)&**2))/((18*j)&*(n-1)).

compute top = a&*((1-alpha)&**(-1/3)).

compute bot = a&*((1-alpha)&**(-2/3)).

compute m = csum(a)-((csum(top)**2)/csum(bot)).

print/title = "** SIGNIFICANCE TEST OF THE DIFFERENCE BETWEEN INDEPENDENT ALPHA COEFFICIENTS **".

print/title = "From Hakstian & Whalen (1976), Psychometrika, 41, 219-231.".

print mat/clabels = "ALPHA" "N" "ITEMS"/title = " "/format F10.4.

compute p = 1-chicdf(m,df).

compute res = {m, df, p}.

print res/clabels = "CHISQ" "DF" "P-VALUE"/title = " "/format F10.4.

END MATRIX.

!END DEFINE.

 

 



      Stay connected to the people that matter most with a smarter inbox. Take a look http://au.docs.yahoo.com/mail/smarterinbox