Variable Types in an Experiment Activities

Assignment 1:Fire and Fynbos, Winter Weather and Car Tires, Makeup
- Assignment or see TopHat
- Objective: practice identifying different sorts of variables in informal experiments posted on youtube
Assignment 2: Participant Variables and Stimuli Variables
- See TopHat
- excel sheet needed for these activities
- Objective: practice identifying different sorts of variables from hypothetical experiments based on the description of the study and looking at the data
Class Recording with answers and explanations
Activity Info
- Objectives
  - understand how experiments rule out reverse causality and confounds (internal validity)
  - understand the necessity of control groups in experiments
  - understand the relation between sample size and the chance of having large differences between the groups just by chance in an experiment with randomization
- What Students Should Be Able To Do After This Activity
  - identify different sorts of stimuli variables and subject variables (e.g., critical difference - what is being manipulated, design confound, noise variables, constants) in experiments, both from reading a textual description of an experiment and from data
- Prerequisite Knowledge
  - how to read a causal diagram
  - the excel activity requires remembering what a t-test does
- What Students Struggle with
  - Identifying variable types. In particular, students often get confused with the following:
  - Constants (there is no variance within or between groups) vs. noise variables (the average for each group is the same, but there is variance within a group)
  - When identifying variable types, only consider the people or stimuli who were included in the final analyses.
  - If a certain group of people were excluded from a study, that is not a confound - it is a concern about external validity / generalizability. Confounds have to do with internal validity.
  - Selection effect (random assignment fails and the two groups are different) is not the same thing as self-selection into groups (there was not random assignment) and volunteer bias (self-selection into the study, not a truly random sample)