Chapter 6: Indexes, Scales, and Typologies, Validity & Reliability

Why do social scientists create composite scores (constructs)? 

• Combine multiple variables that reflect a concept to better capture its full range.

• Expands the range of possible scores compared to a single indicator.

• Serves as a data reduction tool, simplifying analysis while retaining information.

What is an index?

created by adding together scores from multiple variables (dichotomous, nominal, ordinal, interval, or ratio) to represent a concept more fully

What is a scale?

it assigns scores to patterns of attributes.

What are the types of scales?

1. Guttman Scale: Tests strict item response consistency (e.g., political activity).

2. Bogardus Social Distance Scale: Measures social distance; as distance declines, difficulty of agreeing positively increases.

3. Thurstone Scale: Items ranked by “judges” for difficulty or agreement; used in comparative research.

4. Likert Scale: Assigns scores based on intensity of agreement/disagreement; now broadly used as matrix-style ordinal response patterns.

What is a typology?

represents the intersection between two or more dimensions of a construct

What is reliability in research?

a term describing a measure that accurately reflects the concept it is intended to measure  

What is validity in research?

the quality of a measurement method that ensures it actually measures what it is intended to measure

How do we evaluate the reliability of indicators?

• Test-retest reliability: Measure the same concept at two points in time; results should be consistent.

• Inter-item consistency: Check if multiple indicators of the same concept give similar results (e.g., Cronbach’s alpha).

• Inter-observer reliability: Multiple observers should record the same outcomes under the same conditions.

What is face validity for composite scores and indicators?

asks whether the indicators make sense at first glance and align with common-sense notions of the concept

What is construct (internal) validity for indexes and scales?

Logical relationship among indicators.

• Index: Do indicators correlate with each other, and does the composite correlate more strongly with each indicator than indicators do with each other?

• Scale: Is there item-response consistency across the scale?

What is predictive (criterion-related) validity?

Does the new composite score (index or scale) predict a related external behavior, attitude, or characteristic in the manner expected?  

What is content validity?

• Does the composite score cover the full range of the concept or dimension?

• Does it predict related behaviors more strongly than any single indicator alone?