.-
help for ^sampicc^
.-

 ^Sample Size and Power Estimation for Single ICC^
---------------------------------------------------

        .^sampicc <p1> <p0> <reps>^ [, ^A^lpha(#) ^P^ower(#) ^S^ample(#) ^W^idt
> h(#) ^CI^ ]


^Description^
^-----------^

^sampicc^ computes the sample size or, alternatively, the power required to tes
> t an
ICC (p1) against a null value (p0).  In addition, the program will compute the 
> sample
size for a specified width of a confidence interval for the ICC.  There are two
>  forms
of the command:

^FORM 1^: To compute the sample size for a test of a hypothesized value against
>  a null 
^------^  value, three inputs are required:

            ^p1^  --- the hypothesized ICC
            ^p0^  --- the null value for ICC
           ^reps^ --- the number of replicates taken on each subject (must be a
> n integer)


         ^Options for Form 1^
         ^------------------^
            ^alpha(#)^   The Type 1 error value (default is 5%)
            ^power(#)^   The power for the test (default is 80%)
           ^sample(#)^   The number of subjects to be tested.  If specified, th
> e program computes
                            power for the given sample size

         ^Syntax for Form 1^
         ^-----------------^

          .^sampicc <p1> <p0> <reps>^, <options>


^NOTE to Form 1^: To be conservative, all sample size estimates are rounded to 
> the next highest
                integer value, if the solution involves a decimal (e.g., 15.1 r
> esolves to 16)



^FORM 2^: To compute the sample size for a specified width or, alternatively, t
> o compute the width
^------^   of a confidence interval for a given sample size, the ^ci^ option mu
> st be specified.  
        
          ^p1^  --- the hypothesized ICC
         ^reps^ --- the number of replicates taken on each subject (must be an 
> integer)        


         ^Options for Form 2^
         ^------------------^
                ^ci^      ci ^is required^ for estimations relating to confiden
> ce intervals
            ^alpha(#)^    In this form, alpha(#) refers to 100(1-alpha)% CI lev
> el. The default
                             alpha is .05 (i.e., 95% CI)
            ^sample(#)^   The number of subjects to be tested.
             ^width(#)^   The desired width of the confidence interval. This va
> lue must be between
                             0 and 1


         ^Syntax for Form 2^
         ^-----------------^

             . ^sampicc p1 reps, w(#) ci^

             . ^sampicc p1 reps, s(#) ci^



^Examples^
^--------^

To compute sample size for an ICC against a null value, assuming a critical lev
> el of
   5% and power of 80% (Form 1)

        .^sampicc .4 .1 3^


To compute power, for an ICC against a null, with the sample size given (Form 1
> )

        . ^sampicc .6 .1 3, s(15)^


To compute sample size for a specified width of the confidence interval (Form 2
> )

        . ^sampicc .5 2, w(.15) ci^


To estimate width of a 90% CI for a given sample size (Form 2)

        . ^sampicc .5 2, s(100) alpha(.10) ci^

_______________________________________________________________________________
> _______________


^References^
^----------^
Shoukri MM, Asyali MH, Donner A. Sample size requirements for the design of rel
> iability study:
   review and new results. Statistical Methods in Medical Research, 2004; 13: 2
> 51-271.

Walter SD, Eliasziw M, & Donner A. Sample size and optimal designs for reliabil
> ity studies,
   Statistics in Medicine, 1998; 17: 101-110.


Author
------
Paul F. Visintainer, PhD
Springfield, MA 01089
visint46@gmail.com


Also see
--------
Manual or on-line help for: @sampsi@, @sxd3_1@ for sskdlg on SSC.