Video Lecture

My lecture video for public about choosing statistical tests. This lecture is less about knowing how each test works and more about being able to use the tables and figures to determine which statistical test to use.
- Slides
- Video Lecture embedded questions and answers
- Enlarged Figure from the Slides
Other Optional Resources:
- UCLA Stats Consulting Finding the Correct Statistical Test
- Another video that covers some of the tests I covered

Choosing the Right Statistical Test Exercise

Activity Info

Objectives and What Students Should Be Able To Do After This Activity
- identify which statistical test should be run to test a hypothesis or answer a question
- this class currently covers the following situations
  - note, the goal is not that students understand how to conduct each test and interpret them all, but at least to understand that by knowing the number of IVs, the types of IVs and DVs, they can figure out which test to learn more about
  - 1-group tests (just a DV, no IV)
  - for situations with 1 IV and 1 DV
    - 4 types of IVs: 2 independent samples, 2 dependent samples, 3 or more groups (categorical), metric/interval/continuous
    - 4 types of DVs: binary, categorical, metric/interval/continuous (assuming normality), ordinal
  - for regression situations with 2 or more DVs
    - choose between linear, ordinal, logistic, or multinomial (for the four types of DVs) - we don't go into each type of regression but students should be able to pick out the type and know they exist
  - for 2 binary IVS, 2x2 ANOVA: between x between, between x within, within x within
Prerequisite Knowledge
- Be able to identify if each variable is binary, categorical, ordinal, or metric/interval/continuous
- Be able to read either of the two resources for selecting the appropriate statistical test
What Students Struggle with
- Identifying the type of each variable. For example, sometimes they confuse a categorical variable with three levels for three separate variables.
- The distinction between ordinal vs. metric/interval/continuous is never clear (as it causes challenges for researchers too). I usually say that if an ordinal variable has more than 5 levels then usually it gets treated as metrical/interval/continuous