Psychology Chapter 5: Good Measurement

Ways to Measure Variables

• Three primary measurement methods: self-report, observational, and physiological.

Self-Report Measure

• Gathers data by having individuals answer questions in surveys or interviews.
• Commonly measures personal attributes like gender, stress, religiosity, motivation, and personality.

Observational Measure (Behavioral Measure)

• Captures data by noting observable actions or traces from behaviors.
• Examples include assessing intelligence through puzzle-solving or measuring happiness by counting smiles.

Physiological Measure

• Involves recording biological indicators such as brain activity, hormone levels, or heart rate.
• Techniques include fMRI for brain scanning and facial EMG to detect muscle movements indicating emotions.

Scales of Measurement

• Two main categories: categorical and quantitative variables.

Categorical Variables

• Comprise discrete categories with no numerical meaning, such as gender or species.
• Coded arbitrarily, e.g., 1 for Male and 2 for Female.

Quantitative Variables

• Represent measurable quantities with meaningful numerical values, such as height or test scores.

Types of Quantitative Variables

• Ordinal scale: ranks order without indicating difference degree.
• Interval scale: equal intervals with no true zero, e.g., IQ scores or temperature.
• Ratio scale: possesses equal intervals and a true zero, e.g., height or time measurements.

Reliability

• Indicates the consistency of measurement results.

Validity

• Evaluates whether the measurement accurately reflects the intended variable.

Types of Reliability

• Test-retest reliability: Consistency across multiple uses of the same measure.
• Interrater reliability: Consistency regardless of who conducts the measurement.
• Internal reliability: Consistency of responses among related items within a measure.

Evaluating Reliability with R

• Positive correlation (r > 0.5) indicates strong test-retest reliability.
• Interrater reliability evaluated with an ideal positive correlation (r > 0.7).

Measuring Internal Reliability

• Assessed by the association among items meant to evaluate the same construct, using Cronbach's alpha.

Cronbach's Alpha

• A correlation metric used to examine the internal reliability of measures.

Construct Validity

• Integrates both measurement reliability and validity.

Measurement Validity

• Reflects how effectively a measure represents the intended construct, e.g., assessing self-reported religiosity accuracy.

Reliability vs. Validity

• Reliability focuses on consistency, while validity emphasizes accuracy; a measure can be reliable yet not valid.

Valid Measures of Abstract Constructs

• Psychologists may operationalize concepts like happiness through varied methods: questionnaires, observation, or biological indicators.

Face Validity

• Subjective assessment indicating whether a measure seems plausible; e.g., head circumference as an indicator of hat size appears valid but not for IQ.

Content Validity

• Ensures that the measure encompasses all facets of the construct being evaluated.

Criterion Validity

• Validity assessed through empirical evidence showing correlation with key outcomes, exemplified by the relationship between SAT scores and college GPA.

Known Groups Paradigm

• Evaluates if a measure differentiates between groups with already understood behavior.

Convergent Validity

• Measures of the same construct should correlate strongly with each other.

Discriminant Validity

• A measure should exhibit weaker correlations with different but related constructs.

Reliability and Validity Relationship

• A precise measure can be reliable without being valid; however, validity cannot exceed reliability levels.