PS295 Lecture 6: Estimating the Reliability of a Measure
Reliability:
-reliability refers to the consistency of a measure
-it also refers to the degree to which study results can be replicated using the same
measures and procedures
-a researcher's choice of measurement can affect the reliability of the data collected
-some scales are more reliable than others
-measurement error occurs when other factors distort a participant's true score (a
participant's score collected using a perfect measure)
-because no scale is perfect, all measures contain some degree of measurement error
-measurement error undermines the reliability of a measure
Observed score = true score + measurement error
Estimating the Reliability of a Measure:
-assessing a measure's reliability involves analyzing the variability in a set of scores
-researchers estimate the proportion of the total variance in a set of scores that is
systematic variance vs error variance
Systematic variance:
-variability that is systematically related to the variance being meausred
-expected variability
Document Summary
Reliability refers to the consistency of a measure. It also refers to the degree to which study results can be replicated using the same measures and procedures. A researcher"s choice of measurement can affect the reliability of the data collected. Measurement error occurs when other factors distort a participant"s true score (a participant"s score collected using a perfect measure) Because no scale is perfect, all measures contain some degree of measurement error. Measurement error undermines the reliability of a measure. Assessing a measure"s reliability involves analyzing the variability in a set of scores. Researchers estimate the proportion of the total variance in a set of scores that is systematic variance vs error variance. Variability that is systematically related to the variance being meausred. Observed score = true score + measurement error. Researchers are interested in how scores vary among participants. Total variance = systematic variance + error variance. A reliablity of . 70 is considered sufficiently reliable for research purposes.