ICC Subcommand (RELIABILITY command)

ICC displays intraclass correlation coefficients for single measure and average measure. Single measure applies to single measurements—for example, the rating of judges, individual item scores, or the body weights of individuals. Average measure, however, applies to average measurements, for example, the average rating of k judges, or the average score for a k-item test.

MODEL. Model. You can specify the model for the computation of ICC. There are three keywords for this option. ONEWAY is the one-way random effects model (people effects are random). RANDOM is the two-way random effect model (people effects and the item effects are random). MIXED is the two-way mixed (people effects are random and the item effects are fixed). MIXED is the default. Only one model can be specified.

TYPE. Type of definition. When the model is RANDOM or MIXED, one of the two TYPE keywords may be given. CONSISTENCY is the consistency definition and ABSOLUTE is the absolute agreement definition. For the consistency coefficient, the between measures variance is excluded from the denominator variance, and for absolute agreement, it is not.

CIN. The value of the percent for confidence interval and significance level of the hypothesis testing.

TESTVAL. The value with which an estimate of ICC is compared. The value should be between 0 and 1.