Navarro et al. (2020), Learning Statistics with JASP - Chapter 2 Flashcards
Example:
* Age: “33 years.”
* Preference: “I dislike spicy food.”
* Gender: Chromosomal (male) vs. self-identified (male)
numerical values or labels
Example:
* Eye colors (blue, brown, green).
* Transportation modes (car, bus, train).
Nominal Scale (Categorical Variables):
- Example:
- Race positions: 1st > 2nd > 3rd.
- Climate change survey:
1. Believes it is human-caused
2. Believes it has an unknown cause
3. Believes it is not human-caused
Ordinal Scale:
Reliability: * Test-retest Reliability:
◦ Stability over time.
Reliability: * Inter-rater Reliability:
◦ Consistency between evaluators .
Experimental Research:
Key features:
* Researcher controls all aspects of the study, including participant experiences.
* Manipulates predictor variables (IVs) while allowing the outcome variable (DV) to vary
naturally.
* Employs randomization to assign participants to groups and minimize systematic
differences
Non-Experimental Research
Characteristics:
* Limited researcher control; variables are not manipulated.
* Suitable for cases where manipulation is unethical or impractical.
* Includes quasi-experimental designs (e.g., comparing smokers and non-smokers) and
case studies
Assessing the Validity of a Study
Validity addresses whether the study results can be trusted
Five types of validity
- Internal Validity
- External Validity
- Construct Validity
- Face Validity
- Ecological Validity
Internal Validity
The degree to which causal conclusions about the relationship between
variables can be made.
- Example: Comparing first-year and third-year students’ writing skills may reflect age or
experience, not just education
=> Focuses on causal relationships, while external validity assesses the generalisability of findings to other contexts
External Validity
The generalizability of findings beyond the study setting.
- Threats:
* Overreliance on narrow populations, like psychology students, which may not represent the
general population .
* Mismatched study structures (e.g., lab experiments vs. real-world scenarios)
Construct Validity
Whether the study measures the theoretical concept it claims to
measure.
- Example: Counting self-reported cheaters in a classroom measures willingness to admit
cheating, not actual cheating
Face Validity
Whether a measure appears to do what it claims.
- Importance:
* Useful for laypeople (e.g., policymakers) but less critical for scientific rigor
Ecological Validity
Whether the study setup resembles real-world conditions.
- Example: Lab-based eyewitness identification studies often lack ecological validity due to
artificial settings .
Confounders
- Unmeasured variables related to both predictors and outcomes
- Threaten internal validity by obscuring causal relationships
Artifacts
- Results specific to the study context that don’t generalize.
- Threaten external validity
Specific Threats:
- History Effects:
- Events during the study (e.g., natural disasters) influence outcomes
- Reactivity and Demand Effects:
- Participants change behavior because they know they’re being studied
- Example: The Hawthorne effect (e.g., improved productivity due to observation)
- Placebo Effects:
- Improvement due to belief in treatment rather than the treatment itself.
- Strongest in self-reported outcomes like pain
- Non-Response Bias:
- Systematic differences between responders and non-responders can skew results
- Regression to the Mean:
- Extreme initial measurements tend to move closer to the average upon subsequent
measurements - Situation, Measurement, and Sub-Population Effects:
- Study context, participant demographics, and tools can limit generalizability
- Fraud, Deception, and Self-Deception:
- Researchers may unintentionally or intentionally skew results. Issues include:
Data fabrication: Making up results
Hoaxes: Intentional false studies
