127-129
Validity in Measurement
Definition of Validity
Validity refers to how well an empirical measure reflects the intended concept.
Measurement validity addresses whether a measurement truly captures what it claims to measure.
Operational Definitions
Operational definitions specify how concepts are measured.
A valid measure accurately reflects the concept in question.
Example of Validity
A laser speed detector accurately measures vehicle speed.
A wind velocity indicator does not validly measure vehicle speed due to different displacement mechanics.
Assessing Validity
Face Validity
Face validity assesses whether a measure intuitively appears to be valid based on common understanding and logical reasoning.
Example: Counting citizen complaints may or may not adequately measure satisfaction with police services.
Illustration: A study by Ditte Johansen (2006) where electronic records of food purchases indicated healthiness based on drink choices highlighted face validity.
Criterion-Related Validity
Involves comparing a measure to an established external criterion to assess its validity.
Convergent Validity: Shows a measure's ability to predict cases on other, accepted measures.
Example: College Board exams' effectiveness in predicting college success.
Example: Voas et al. (2009) validated a measure of alcohol-related auto accidents by comparing it to blood alcohol test results from drivers.
Discriminant Validity
This demonstrates that a measure can differentiate between similar but distinct concepts.
Example: Research by Jacinta Gau (2011) distinguishes between obligations to obey legal authorities and trust in police.
Construct Validity
Definition
Assessing construct validity involves examining the logical relationships among different variables.
Example: The measure of fear of crime should relate logically to other related measures in expected ways.
Content Validity
Definition
Content validity evaluates whether a measure captures the full range of meanings within a concept.
Example: A question on safety after dark captures some aspects of fear but lacks comprehensiveness for the broader concept of fear of crime (which includes burglary, auto theft, etc.).
Multiple Measures
Approach
Comparing multiple measures within a given concept can support the validation of individual measures.
Example: Examining the relationship between fear of crime and the tendency to limit outings at night.
Validity is strengthened if theoretical expectations about relationships hold true, as shown in studies by Ditton and Farrall (2007).
Reliability and Validity Relationship
Understanding Reliability vs. Validity
Illustrative analogy:
A reliable measure consistently hits a 'target.'
A valid measure accurately hits the target (bull's-eye).
Implications
Validity failures indicate systematic error, while reliability failures result in random error.
Utility of a measure diminishes if it lacks both reliability and validity.
Composite Measures
Utility of Combining Measures
Composite measures, such as indices, often provide more reliable and valid indicators than individual measures.
Example: A property crime index could combine burglary, larceny, and vehicle theft; a violent crime index would include murder, robbery, etc.
Reasons for Composite Measures
Sometimes, concepts are complex, requiring more than one indicator for valid measurement (e.g., fear of crime).
Ordinal measures might need a refined range of categories, achievable through indexes or scales.
Increased efficiency in data analysis through multiple data items leads to more comprehensive evaluations.