**Manipulation Check**
We will run an ANOVA on how difficult participants found it to list things about themselves that had changed with easy/difficult as one factor and highschool/4-8 years as the second factor. This will be done with the following code:
BOOTSTRAP
/SAMPLING METHOD=SIMPLE
/VARIABLES TARGET=manipcheck INPUT=hs difficult
/CRITERIA CILEVEL=95 CITYPE=PERCENTILE NSAMPLES=10000
/MISSING USERMISSING=EXCLUDE.
UNIANOVA manipcheck BY hs difficult
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/EMMEANS=TABLES(OVERALL)
/EMMEANS=TABLES(hs) COMPARE ADJ(LSD)
/EMMEANS=TABLES(difficult) COMPARE ADJ(LSD)
/EMMEANS=TABLES(hs*difficult)
/PRINT=DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=hs difficult hs*difficult.
To replicate the findings we require that the manipulation check work in either of the groups. We will only pursue folow-up testing if this is the case on only for the groups where the manipulation was successful.
**Primary Dependent Measures**
Given that the manipulation works, we will replicate the original investigation by combining the four dependent variables into a summary score.
We will then run separate-variance t-tests on whichever the HS or 4-8 years experiments successfully manipulated difficulty.
We will forego the mediation model as the model uses the manipulation check as the mediator which leads to overly large manipulation to mediator (a) paths which can upwardly bias indirect (ab) paths. Furthermore, the original investigation did not find a statistically-significant mediation (reported as marginally significant) and thus we do not pursue it here.