Survey Research: Measurement, Reliability & Validity

Vocabulary flashcards covering measurement levels, scale types, reliability, and validity concepts from the lecture.

Nominal Scale

Qualitative level of measurement that categorizes data without any order (e.g., gender, ethnicity).

Ordinal Scale

Categorical level of measurement that ranks data in a specific order without equal intervals (e.g., education levels).

Interval Scale

Numerical measurement scale with equal intervals but no true zero (e.g., Likert-summed scores, temperature in °C).

Ratio Scale

Numerical measurement scale with equal intervals and an absolute zero, allowing meaningful ratios (e.g., income, age in years).

Absolute Zero

A true, non-arbitrary zero point found only in ratio scales, indicating the complete absence of the attribute.

Categorical Data

Data that place entities into distinct groups or categories (Nominal and Ordinal scales).

Likert Scale

An itemized interval scale typically ranging from strong agreement to strong disagreement, used to measure attitudes.

Graphic Rating Scale

Ordinal scale where respondents mark their rating on a continuous line between two endpoints.

Constant Sum Scale

Ordinal scale where respondents distribute a fixed number of points (e.g., 100) across items to indicate relative importance.

Itemized Rating Scale

Interval-type scale with numbered categories and verbal anchors measuring variables like satisfaction or likelihood.

Stapel Scale

Interval scale that uses a single adjective and numerical values from +5 to −5 to measure direction and intensity of attitudes.

Semantic Differential Scale

Interval scale with bipolar adjective pairs (e.g., good–bad) measuring connotative meaning of objects or concepts.

Numeric Rating Scale

Interval scale where respondents choose a number (e.g., 0–10) representing intensity, often used for pain assessment.

Dichotomous Scale

Nominal scale offering two mutually exclusive response options (e.g., yes/no, pass/fail).

Category Scale

Nominal scale with multiple response options (e.g., urban/suburban/rural).

Ranking Scale

Scale requiring respondents to order items relative to each other (Paired Comparison, Forced Choice, Comparative).

Paired Comparison

Ranking method where respondents choose a preferred option within each pair of items.

Forced Choice

Ranking method where respondents select the most and least characteristic items among sets, forcing discrimination.

Comparative Scale

Ranking scale providing a benchmark for assessing current attitudes or conditions relative to a standard.

Goodness of Measure

Overall quality of an instrument, encompassing reliability and validity.

Reliability

Extent to which an instrument yields consistent results over time, items, and conditions.

Validity

Extent to which an instrument measures what it is intended to measure.

Test–Retest Reliability

Stability measure obtained by correlating scores from the same respondents at two different times.

Carryover Effects

Improved test performance on a retest due to familiarity with the instrument, affecting test–retest reliability.

Split-Half Reliability

Internal consistency method that correlates scores from two halves of the same test (e.g., odd vs. even items).

Cronbach’s Alpha

Statistic measuring internal consistency; acceptable values generally ≥ .70 in social sciences.

Inter-Item Consistency

Degree to which items on a scale correlate with one another, often assessed by Cronbach’s alpha.

Content Validity

Degree to which an instrument fully represents the construct’s domain, judged by subject experts.

Face Validity

Subjective assessment of whether a test appears to measure the intended concept.

Construct Validity

Extent to which an instrument accurately measures the theoretical construct, assessed via factor analysis.

Criterion-Related Validity

Extent to which instrument scores correlate with an outcome criterion (concurrent or predictive).

Concurrent Validity

Correlation between the instrument and criterion measured at the same time.

Predictive Validity

Ability of an instrument to forecast future outcomes related to the construct.

Convergent Validity

High correlation between measures that theoretically should be related.

Discriminant Validity

Low correlation between measures of theoretically unrelated constructs.

Item Analysis

Statistical procedure (e.g., t-tests) to identify items most related to the construct by comparing high vs. low scorers.

Median

Measure of central tendency appropriate for ordinal data.

Mode

Measure of central tendency appropriate for nominal data.

Mean

Arithmetic average appropriate for interval and ratio data.

Standard Deviation

Measure of dispersion calculated for interval and ratio data.

Chi-Square Test (χ²)

Statistical test often used with nominal data to examine relationships between categories.

Coefficient of Variation

Relative dispersion measure (SD/mean) applicable to ratio data.

Ordinal Data Examples

Education level, job rank, BMI categories, attitude ratings.

Interval Data Examples

Summed Likert attitude score, temperature in Fahrenheit, customer satisfaction index.

Ratio Data Examples

Income, age in years, number of products sold, weight.

Nominal Data Examples

Gender, marital status, currency type, job type (professional/admin/technical).

Absolute Zero Example

Zero dollars of wealth or zero pregnant underage girls—indicates none of the attribute.

Scale Transformation

Same variable (e.g., age) can be captured as ratio, interval, or ordinal depending on format.

Adapted Instrument

Existing validated measure that is modified to suit a new context, preferred over creating new scales.