Beyond the p-value (Wintersemester 2020/2021)

Lecturer: Prof. Dr. Bernhard Renard (Data Analytics and Computational Statistics) , Dr. Sven Giese (Data Analytics and Computational Statistics)
Course Website: https://hpi.de/friedrich/moodle/course/view.php?id=128

General Information

• Weekly Hours: 2
• Credits: 3
• Graded: yes
• Enrolment Deadline: 01.10.2020 -20.11.2020
• Teaching Form: Seminar
• Enrolment Type: Compulsory Module
• Course Language: English
• Maximum number of participants: 10

Programs, Module Groups & Modules

Description

P-values are a key statistical tool to evaluate evidence. Believing and aiming for results below a magical 0.05 threshold has become an almost religious activity in some fields of science. The mathematical definition itself in hypothesis testing does lack some of this magic charm: It is the probability of obtaining results at least as extreme as the results observed under the assumption that the null hypothesis is correct.

Within the seminar we will look at the problems resulting from regarding p-values for more that they actually are. Beyond the description of why arguably most published research results are wrong, we will look at various ways to fix the problem – from mathematical approaches to more process-oriented codes of conduct. 

Learning objectives

• You learn to analyze strengths and weaknesses of approaches to evaluate statistical results
• You learn to identify open challenges in hypothesis testing
• You can present a scientific manuscript in this field and lead a discussion

Requirements

You should have some mathematical background (at least Mathe 1+2 of the ITSE bachelor or comparable) as well as have taken at least one class in statistics. Good knowledge of English to understand and discuss current literature.

Literature

• Why most published research is wrong (https://journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0020124)
• Significance in the Bayesian Framework (https://www.frontiersin.org/articles/10.3389/fpsyg.2019.02767/full)
• The salmon study  (https://www.semanticscholar.org/paper/of-Serendipitous-and-Unexpected-Results-Neural-of-%3A-Bennett-Baird/3ace3864cc5ada47b31a74d8fea91edb48bc019d)
• FDR as a concept (https://projecteuclid.org/euclid.aos/1074290335)
• The dirty dozen p-value flaws (https://doi.org/10.1053/j.seminhematol.2008.04.003)
• Effect sizes (http://www.leeds.ac.uk/educol/documents/00002182.htm)

Learning

• Seminar for master students 
• Language of instruction: English
• Maximum number of participants: 10

Topics will be presented in the first Session (November 3, 2020). For topic assignments, participants will have to write an E-Mail by November 10^th, 2020 in which they can give preferences for up to 3 of the presented topics. Then, the topics will be assigned by us. In case of too many applicants, we will decide randomly.
The seminar will be conducted virtually via zoom (we can provide a room for those on campus).
Please let us know by email (office-renard@hpi.de) if you plan to attend the first session and we will share the dial-in details

Examination

In the seminar, each participant will give a presentation about a predefined topic within the research area of truth discovery and write a short report. The final grade consists of the following three parts:

• Presentation (45%)
• Written report (35%)
• Discussion in the seminar sessions (20%)

Zurück