Exploring Bayesian Methods for Social Policy Research and Evaluation

Bayesian methods are emerging as the primary alternative to the conventional frequentist approach to statistical inference. Bayes' theorem is a model for learning from data. Using Bayes’ theorem, a researcher weights their prior beliefs about the size of an intervention’s effect by the data observed through experimentation.

Bayesian analysis results in a point estimate of the intervention’s effect and an interval for the credible value of the effect. With these in hand, the research can answer such questions as:

  • What is the probability that the effect of the intervention is greater than zero?
  • What is the probability that the effect of the intervention is between values X and Y? (Where X and Y might be values of policy relevance.)

The frequentist perspective cannot answer these types of questions.

OPRE’s 2017 Methods Meeting, Bayesian Methods for Social Policy Research and Evaluation Visit disclaimer page , informed many of our resources on Bayesian inference. This meeting was one in a series of annual meetings in our Methods Inquiries project that bring together expertise from varying disciplines across academia, government, and the private sector. Participants explore innovations in research design, analytic techniques, and data measurement with the potential to advance OPREsupported research so that it represents the most scientifically advanced approaches to determining effectiveness and efficiency of ACF programs.

Learn more about Bayesian methods and explore the agenda and slides from our 2017 Methods Meeting at https://opremethodsmeeting.org/ Visit disclaimer page

Point(s) of contact: Kriti Jain

Related Resources

For nearly 100 years, the null hypothesis significance testing (NHST) framework has been used to determine which findings are meaningful (Fisher 1925; Neyman and Pearson 1933). Under this framework, findings deemed meaningful are called “statistically significant.”  But the meaning of statistical significance is often...

Bayesian methods are emerging as the primary alternative to the conventional frequentist approach to statistical inference. This brief provides an overview of the Bayesian perspective and highlights potential advantages of Bayesian inference over frequentist inference. In light of the increasing value and viability of Bayesian methods to contemporary...

Bayesian methods allow researchers to describe findings in probabilistic terms. They also allow for incorporating prior knowledge and considering uncertainty in parameters.

 

Bayesian methods may also help with transparency...

There is a growing understanding that there are some inherent limitations in using p-values to guide decisions about programs and policies. Bayesian methods are emerging as the primary alternative to p-values and offer a number of advantages...

Probability (p) values are widely used in social science research and evaluation to guide decisions on program and policy changes. However, they have some inherent limitations, sometimes leading to misuse, misinterpretation, or misinformed decisions. Bayesian methods...

Social policy evaluations usually use classical statistical methods, which may, for example, compare outcomes…