A nonsignificant χ 2 statistic is preferable because it reflects “exact fit” between the hypothesized model and the data; however, we rely primarily on the RMSEA because the χ 2 is highly affected by sample size: larger samples produce larger χ 2 results that are more likely to be significant (type I error).
← all excerpts
Relationships Between Theoretically Derived Short-Term Outcomes and Support for Policy Among the Public and Decision-Makers.
1
—
—