We will collect 200 participants from MTurk, who have U.S. IP addresses, have participates in less than 500 studies, are NOT Master workers, and have a complete rate of >= 95%.
Consistent with its use, we will sum all of the scores of the authoritarianism scale into one score. We will also construct a sum score of the vocab test.
We will run the following model first in the entire data using just manifest variables:
missing are all (9999);
Analysis:
estimator = MLR;
ITERATIONS= 10000;
Model:
ktd ON vocab;
!Add in rwa as a covariate
ktd ON rwa;
Output:
Samp stdYX Residual Cinterval Tech4 Tech1;
We will predict that more intelligent people will think that kids are getting dumber.
We will then systematically remove participants for failing the three data quality checks (Flynn, seriousness, looked up words0 in order of smallest to largest percent of the data. If the total N gets below 160, we will not use the following latent variable approach (only including rwa if it was a predictor in the above analysis):
CATEGORICAL ARE ktd vocab1-vocab8;
missing are all (9999);
Analysis:
estimator = WLSMV;
ITERATIONS= 10000;
Model:
Lvocab BY vocab1-vocab8;
ktd ON Lvocab rwa;
Output:
Samp stdYX Residual Cinterval Tech4 Tech1;
