Exam Chapter 4 and 5

Reliability determined by correlating two forms of a test. also referred to as parallel or equivalent form reliability. Basic design is test retest.

Representation of the relationship between two variables on a Cartesian (X, Y) coordinate system; also called a scatterplot.

A measure of the internal consistency of items on a test, as the number of items increases, reliability increases. as the average interitem correlation increases, reliability increases. the formula shows that the alpha depends on the average correlation among items. number of items is important it relates to unreliability due to changes in content sampling and not due to changes in test administration, personal conditions, or scoring.

A band placed around a test score based on the standard error of measurement.

An error that makes scores consistently higher or lower thank warranted for an individual or group of individuals due to factors unrelated to the test's purpose.

The numerical expression, ranging from -1.00 to +1.00, of the relationship between two variables.

The hypothetical difference between a person's obtained score and true score.

A method for studying reliability that allows for examining several sources of unreliable variance simultaneously. using a single study to determine reliability for several occasions, several forms, and several scores. ANOVA provides the framework.

Excessive differences among individuals, especially greater differences than normal. yields an inflated correlation.

Different degrees of scatter at various points along a best-fitting line. the standard error is not equal throughout the range of variables although it is calculated as if it were equal throughout.

Yields a deflated correlation. less difference among individuals than is normal.

Equal degree of scatter at various points along a best fitting line.

Items that for the most part, are measuring the same trait or characteristic as indicated by the intercorrelations among the items.

The degree of agreement about individuals' performance among different scorers of a test.

A type of correlation expressing the degree of agreement among more than two raters or judges.