Causation | Does smoking cause cost? |
Economic research in general | Are the methods of economic research legitimate? Are the results of economic analyses credible? |
Analysis of cofactors and/or confounders | Does the model take confounding and cofactors into account appropriately? |
Issue framing | Is the issue a medical one or a statistical one? |
Death benefit | Does the model take into account the savings to health insurance, social security, and pension plans that result from premature death from smoking-related illness? This was a way of trying to cast the model in a framework that considers tobacco as contributing to an overall benefit rather than a cost |
• Lifetime v annual costs | Is the model cast in a lifetime cost framework thus considering savings that accrue from early death? |
• Incidence v annual costs | (Same argument as above, but with different terminology.) Is the model cast in an incidence-based cost framework, thus considering savings that accrue from early death? |
Sample selection | Did the models use the appropriate dataset? |
• CPS II | Any discussion of the merits of the CPS II data |
Scientific judgment | What role did subjective judgment play in development of the models? |
Use of statistical models | |
• Missing data | Was missing data handled appropriately? |
• Statistical reliability/ validity | Were the models used statistically reliable and valid? |
• Statistical science | Is the science of statistics a legitimate one? |
• Statistical significance | Were the variables of interest statistically significant? |
• Omitted variables | Were any variables that should have been included omitted from the models? |