Some comparisons, inferences, methods, and measures trigger special review.
|Survey||Check response rate, representativeness of the sample, and analyze the questions.|
|Causality||Making causal inferences from correlational data requires additional rigor. See Hill’s criteria.|
|Missing||Missing data always biases the analysis. Explore the underlying reasons for missing data. Verify that the option used for plugging missing data is acceptable.|
|Time-series||Common problems: arbitrarily picking start and end points, splitting the data at the high point.|
|Transformations||Verify the math and applicability of each transformation. Simple checks such as marginal totals often reveal problems.|