Measures, Reliability, Validity
Scales of Measurement
Data are defined through a process known as scaling, which outlines the level of information represented by numeric codes used for data recording.
It is crucial to know the type of data before starting research.
All research data falls into one of four measurement scales: Nominal, Ordinal, Interval, and Ratio.
Nominal Data
Represents categories or labels that have no quantitative properties (i.e., no inherent value).
Involves grouping objects into classes, ensuring all members share equivalent attributes.
Allows for the fewest statistical tests due to its categorical nature.
Ordinal Data
Assigns numbers to indicate rank order among items, suggesting a qualitative distinction.
Items can be ranked from high to low, revealing differing amounts of a characteristic without defining the magnitude of differences.
No specific value is attributed to intervals between numbers in the scale; the focus is solely on ranking.
Interval Data
Uses numbers to define actual scores that describe the magnitude and differences among items.
Maintains order and equal magnitude of differences between consecutive scale items.
Features equal intervals, but lacks an absolute zero; zero does not signify the total absence of the measured property.
Can include negative values.
Ratio Data
The most precise scale of measurement with scores related through a single dimension.
Separated by equal intervals, it includes an absolute zero indicating the total absence of the variable measured.
Measurement Scale Identification Exercises
Fever (body temperature): Interval
Telephone number: Nominal
Number of rolls of toilet paper: Ratio
Length of a class (in minutes): Ratio
Anxiety score (measured on a scale): Interval
Religious persuasion: Nominal
Distance a student commutes: Ratio
Top ten best dressed people: Ordinal
Rank on a self-esteem scale: Ordinal
Number of dates last month: Ratio
Reliability
Refers to the consistency of a measurement over time.
Components of Reliability
True Score: The actual value of a variable, representing the “real” score.
Measurement Error: Variability in scores indicating how much the measurement deviates from the true score.
Types of Reliability
Test-Retest Reliability: Measurement taken multiple times to check consistency.
Internal Consistency: Responses measured at one point in time.
Split-Half Reliability: Correlates total scores on one half of the measure against the other half.
Cronbach’s Alpha: Correlates each item with every other item on the measure to assess consistency.
Inter-Rater Reliability: Agreement among different observers regarding events.
Validity
Indicates the capacity of a measurement to accurately reflect the variable in study.
This assesses how well an operational definition aligns with the theoretical meaning of the variable.
Types of Validity
Construct Validity: Measures the adequacy of the operational definition to represent the variables.
Face Validity: Assessment based on surface appearances.
Predictive Validity: Ability to forecast future behavior.
Concurrent Validity: Comparison of test results with behavior.
Reactivity
Refers to how the awareness of being measured can alter an individual’s behavior.