Beyond p values workshop -- orientation

USCOTS 2019 Theme: Evaluating evidence

This conference theme will embrace all aspects of evaluating evidence, including but not limited to helping students to:

  • Understand the reasoning process of statistical inference,
    • Not so much the process, but the purpose
  • Recognize appropriate interpretations and limitations of statistical inference,
  • Design studies to facilitate evaluating evidence,
    • Not really
  • Conduct research in a reproducible manner,
  • Consider alternatives to traditional methods for conducting inference,
    • Yes, Some will be mainstream and current in practice, e.g. bootstrapping, cross-validation. Others will be proposals from the statistics community. Still others will be Danny’s ideosyncratic proposals 1, 2, & 3.

and

  • Reflect on the role of inference in the context of big data and data science.
    • This workshop should be directly relevant

March 2019 & statistical inference

The American Statistician special issue on statistical inference.

ASA editorial:

"The ASA Statement on P-Values and Statistical Significance stopped just short of recommending that declarations of ‘statistical significance’ be abandoned. We take that step here. We conclude, based on our review of the articles in this special issue and the broader literature, that it is time to stop using the term ‘statistically significant’ entirely. Nor should variants such as ‘significantly different,’ ‘p < 0.05,’ and ‘nonsignificant’ survive, whether expressed in words, by asterisks in a table, or in some other way.

“In sum, ‘statistically significant’ — don’t say it and don’t use it.”

Nature call to “retire statistical significance.”

Agenda

Tuesday pm

  • examination of the current controversy about p-values
  • exploring the current curriculum to identify other shortcomings
  • introduction of a bit of writing/collaboration technology to support exploring ideas about what and how to teach and to help us bring ideas together

Wednesday am

  • How (or whether) data science is distinct from statistics.
  • How to manage the increased technological emphasis that would come with a greater emphasis on data science.
  • Explore and evaluate my proposal for connecting stats more appropriately with data science: Stats for Data Science book and related tutorials, apps, exercises, etc.
    • discussion of general principles for addressing the problems/opportunities of the call to abandon classical inference. Other general principles for choosing content, notation, and interaction.
    • as time allows, get started on the Wednesday pm agenda

Wednesday pm

  • Chapter by chapter overview of Stats for Data Science

Thursday am

  • Non-data-science proposals for “fixing” inference within today’s Stat 101.
  • Some topics that maybe need to be added to statistics:
    • false discovery
    • communicating about risk
    • adjustment
  • Consolidation of discussions into a “roadmap” for future development.
    • Develop a feasible course curriculum that does justice to the ASA p-value statement.
    • Develop ideas for bringing data science into your statistics courses
    • Help me review the basic structure for Stats for Data Science and identify the many weaknesses that I haven’t already identified.