Reliability
Reliability is the stability, consistency, and precision of a given measure
Reliability is an empirical concept derived from Classical Test Theory (CTT). This theory postulated that an individual's true score on a given psychological instrument was a function of their observed score (O) on the instrument plus (+) the inherent systematic and unsystematic error that exists when measuring human psychological constructs (E). In essence, the true score is the individual's "real" score that we cannot obtain due to measurement error.
Measurement error plays a central role in understanding reliability. A fundamental assumption of reliability is that measurement error is dispersed randomly within a given population. This is assumption is fundamental because it allows researchers to surmise that increases or decreases in observed scores are caused by measurement error that is completely independent of the individual's true score or ability. As scores fluctuate across homogeneous and heterogeneous populations, the fluctuation of scores occurs at random and is not fully indicative of the individual's true ability. Thus, as measurement error increases, our ability to discern a precise measure of the individual's true ability is decreased.
There are statistical measures that assess the amount of measurement error in a survey, the reliability coefficient and the standard error of measurement (SEM). The SEM is like a "bell-curve" for an individual person's true ability that is based on the amount of error associated with the measure. As reliability increases, the SEM will constrict and we can become more confident that an individual's observed score on the survey is indicative of their true score or ability. The inverse holds true as well. As reliability decreases, the SEM will widen and we become less confident that an individual's observed score is indicative of their true score.
There are statistical measures that assess the amount of measurement error in a survey, the reliability coefficient and the standard error of measurement (SEM). The SEM is like a "bell-curve" for an individual person's true ability that is based on the amount of error associated with the measure. As reliability increases, the SEM will constrict and we can become more confident that an individual's observed score on the survey is indicative of their true score or ability. The inverse holds true as well. As reliability decreases, the SEM will widen and we become less confident that an individual's observed score is indicative of their true score.
Types of reliability
There are four primary types of reliability: Internal consistency, test-retest, alternate/parallel forms, and inter-rater. Click on a button below to learn more about that type of reliability. The methods for conducting and interpreting each type of reliability in SPSS will be presented.
Survey items are written to cover a content area in the literature related to one construct and the items should be correlated at a theoretical or conceptual level.
Survey items are written to assess the stability of a trait, characteristic, behavior, emotional state, or construct across time..
Researchers are writing two different survey instruments that cover the same content area.
A panel of experts are rating the performance of a behavior(s), task(s), or construct in either a simulated or natural environment.
Hire A Statistician
DO YOU NEED TO HIRE A STATISTICIAN?
Eric Heidel, Ph.D., PStat will provide you with statistical consultation services for your research project at $100/hour. Secure checkout is available with Stripe, Venmo, Zelle, or PayPal.
- Statistical Analysis on any kind of project
- Dissertation and Thesis Projects
- DNP Capstone Projects
- Clinical Trials
- Analysis of Survey Data
