Study Notes on Internal Validity, Confounds, and Statistical Tests
Internal Validity and Confounds
- Internal Validity:
- Definition: The extent to which we can be confident that changes in the independent variable (IV) caused changes in the dependent variable (DV).
- Confounds:
- Definition: A factor that covaries with the independent variable.
- Explanation of Covarying: “Covaries with the independent variable” means that different values of the confound are perfectly matched with different levels of the IV.
- Example:
- Toothpaste A is always used with a red toothbrush, while Toothpaste B is always used with a blue toothbrush.
- Implications of Confounds:
- They provide possible alternative explanations for the effects on the DV, indicating other potential causes apart from the IV.
Statistics Review: Testing the Significance of Group Differences
Statistical Tests for Two Samples
- Independent-Groups t-test:
- Definition: A statistical test where subjects in the groups are independent of each other.
- Related-Samples/Repeated-Measures t-test:
- Definition: A statistical test where the same subjects are used in all conditions, or related pairs of subjects are analyzed.
- Common Requirements for t-tests:
- The dependent variable (DV) must be a numerical variable (quantity or amount of something) measured for each subject.
- A mean value of the DV can be computed for each group.
Statistical Tests for Three or More Samples
- One-Way Independent-Groups ANOVA:
- Definition: A statistical test used with three or more levels of one independent variable, where subjects in the groups are independent of each other.
- One-Way Repeated-Measures ANOVA:
- Definition: A statistical test with three or more levels of one independent variable, where the same subjects perform in all conditions.
- Common Requirements for ANOVAs:
- The dependent variable (DV) must be a numerical variable (quantity or amount of something) measured for each subject.
- A mean value of the DV can be computed for each group.
Example Research Questions
1. Impact of Music on Sorting Mail
- Research Question: Does the presence of music affect how quickly postal workers sort mail?
- Independent-Groups t-test Design:
- IV: Presence or absence of music (1) music, (2) no music.
- DV: Time taken to sort 100 pieces of mail.
- Method: One group of postal workers sorts mail with music, another in silence. Measure sorting time per group.
- Repeated-Measures t-test Design:
- Method: Same group of postal workers sorts mail first with music, then in silence. Measure sorting time in both conditions.
2. Impact of TEMPO on Sorting Mail
- Research Question: Does the TEMPO of music affect how quickly postal workers sort mail?
- Rationale: Exploring different tempos with uncertainty on which might influence sorting efficiency.
- Independent-Groups ANOVA Design:
- IV: Tempo of music (1) 60 beats per minute (BPM), (2) 120 BPM, (3) 180 BPM.
- DV: Time taken to sort 100 pieces of mail (or how many pieces sorted in 5 minutes).
- Method: Separate groups of postal workers sort mail at each specified BPM. Measure sorting time for each group.
- Repeated-Measures ANOVA Design:
- Method: Same group of postal workers sorts mail at each tempo (60 BPM, then 120 BPM, and finally 180 BPM). Measure sorting time per condition.