Toward causality and improving external validity

Smoking and healthcare expenditure reductions associated with the California Tobacco Control Program, 1989 to 2019: A predictive validation

BACKGROUND

Previous research used data through 2008 to estimate a model for the effect of the California Tobacco Control Program (CTCP) that used cumulative real per capita tobacco control expenditure to predict smoking behavior (current adult smoking prevalence and mean cigarette consumption per current smoker). Predicted changes in smoking behavior due to the CTCP were used to predict its effect on health care expenditure. This research updates the model using the most recently available data and estimates CTCP program effect through 2019.

METHODS

The data used in the previous research were updated, and the original model specification and a related predictive forecast model were re-estimated. The updated regression estimates were compared to those previously published and used to update estimates of CTCP program effect in 2019 dollars.

RESULTS

There was no evidence of structural change in the previously estimated model. The estimated effect of the CTCP program expenditures on adult current smoking prevalence and mean consumption per adult current smoker has remained stable over time. Over the life of the program, one additional dollar per capita of program expenditure was associated with a reduction of current adult smoking prevalence by about 0.05 percentage point and mean annual consumption per adult current smoker by about 2 packs. Using updated estimates, the program prevented 9.45 (SE 1.04) million person-years of smoking and cumulative consumption of 15.7 (SE 3.04) billion packs of cigarettes from 1989 to 2019. The program produced cumulative savings in real healthcare expenditure of $544 (SE $82) billion using the National Income and Product Accounts (NIPA), and $816 (SE $121) billion using the Center for Medicare and Medicaid Services (CMS) measure of medical costs. During this time, the CTCP expenditure was $3.5 billion.

CONCLUSION

A simple predictive model of the effectiveness of the CTCP program remained stable and retains its predictive performance out-of-sample. The updated estimates of program effect suggest that CTCP program has retained its effectiveness over its 31-year life and produced a return on investment of 231 to 1 in direct CMS medical expenditure.

Distributional regression modeling via generalized additive models for location, scale, and shape: An overview through a data set from learning analytics

The advent of technological developments is allowing to gather large amounts of data in several research fields. Learning analytics (LA)/educational data mining has access to big observational unstructured data captured from educational settings and relies mostly on unsupervised machine learning (ML) algorithms to make sense of such type of data. Generalized additive models for location, scale, and shape (GAMLSS) are a supervised statistical learning framework that allows modeling all the parameters of the distribution of the response variable with respect to the explanatory variables. This article overviews the power and flexibility of GAMLSS in relation to some ML techniques. Also, GAMLSS' capability to be tailored toward causality via causal regularization is briefly commented. This overview is illustrated via a data set from the field of LA. This article is categorized under:Application Areas > Education and LearningAlgorithmic Development > StatisticsTechnologies > Machine Learning.