* Encoding: UTF-8.
comment This program computes the sample size necessary to achieve target levels of power
comment for correlation and partial correlation coefficients such as are reported in Tables 4 and 5 in
comment Algina and Olejnik (2003).
comment The population correlation can be changed by changing .05 in the code rho = .05.
comment The number of control variables can be changed by changing 0 in the code q=0.
comment The Type I error rate can be changed by changing .05 in the code alpha = .05.
comment The number of tails can be changed by changing 2 to 1 in the code ntails=2.
comment The target power can be changed by changing .50 in the code tpower=.50.
NEW file.
INPUT PROGRAM.
compute rho=.05.
compute q=0.
compute alpha=.05.
compute tails = 2.
compute tpower =.50.
loop #nn = q + 3 to 10000.
compute #prob=1-alpha.
if (tails=2) #prob=1-alpha/2.
compute #df=#nn-q-2.
compute #cval=IDF.T(#prob,#nn-2).
compute #g=sqrt(2*(1-rho**2)/(2-rho**2)).
compute #deltasq=((2*(#nn-q)-3)*rho**2)/(2-rho**2).
compute #delta=sqrt(#deltasq).
compute #p=1-NCDF.T(#g*#cval,#df,#delta).
END LOOP if (#p>=tpower).
compute power=#p.
compute n=#nn.
END CASE.
END FILE.
END INPUT PROGRAM.
EXECUTE.