We will run two analyses.
The first will be a straight replication of the original study. Scores on both 'how conservative' someone is and 'how likely theya re to vote for a conservative' will be averaged together and a *t*-test will be performed on the data.
The second analysis will run a MANOVA on the two dependent variables with 10,000 bootstraps to acocunt for posisble non-normalities in the data.
Code for this analysis will be:
BOOTSTRAP
/SAMPLING METHOD=SIMPLE
/VARIABLES TARGET=conservative votecon INPUT=implausible
/CRITERIA CILEVEL=95 CITYPE=PERCENTILE NSAMPLES=10000
/MISSING USERMISSING=EXCLUDE.
GLM conservative votecon BY implausible
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/EMMEANS=TABLES(implausible)
/PRINT=DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN= implausible.
We do not predict a successful replication in either analysis.