Search code examples
spss

SPSS Bootstrap with custom sample size

I have a very large sample of 11236 cases for each of my two variables (ms and gar). I now want to calculate Spearman's rho correlation with bootstrapping in SPSS.

I figured out the standard syntax for bootstrapping in SPSS with bias corrected and accelerated confidence intervals:

DATASET ACTIVATE DataSet1.
BOOTSTRAP
  /SAMPLING METHOD=SIMPLE
  /VARIABLES INPUT=ms gar 
  /CRITERIA CILEVEL=95 CITYPE=BCA  NSAMPLES=10000
  /MISSING USERMISSING=EXCLUDE.
NONPAR CORR
  /VARIABLES=ms gar
  /PRINT=SPEARMAN TWOTAIL NOSIG
  /MISSING=PAIRWISE.

But this syntax is resampling my 11236 cases 10000 times.

How can i achieve taking a random sample of 106 cases (√11236), calculate Spearman's rho and repeat 10000 times (with new random sample of 106 cases each bootstrap step)?

Solution

  • Use the sample selection procedures - Data > Select Cases. You can specify an approximate or exact random sample or select specific cases. Then run the BOOTSTRAP and NONPAR CORR commands.