Test Reliability and Validity
Most of the studies have been done with the paper-pencil version of the CK. Internal consistency coefficients (Kuder-Richardson 20) for male and female college students combined in one study for the six scales ranged from .69 to .92 with a mean of .74. Test-retest correlations over a three-week interval ranged from .74 to .88, with a mean of .82. In a study of high school students (Jones & Ward, 2002), the test-retest reliabilities for mostly ninth graders over a four-week interval were in the 80's, although the Conventional scale's reliability coefficient (.63) was lower than the others. These findings were similar to two comparable studies reported in the manual for the Self-Directed Search (Holland, Fritzsche, & Powell (1997).
Investigations of the validity of CK scales have been positive. In one study (Jones, Gorman, & Schroeder, 1989) comparing the CK with Holland's Self-Directed Search found no difference between SDS and CK in their matches with Holland's Vocational Preference Inventory (VPI; Holland, 1985c). Another study (Jones, 1990) found that the three-letter code of the CK, when compared to the three-letter code of the VPI for each student, was what Iachan (1984) called a "reasonably close match." Jones & Ward (2002) did the same type of analysis. The findings were quite similar to the 1990 study; the three-letter CK and VPI codes were reasonably close matches.
The intercorrelations among the CK six scales were found (Jones, 1990) to match the theoretical expectations for 44 of the 48 relationships compared to 41 of the 48 for Holland's VPI. In the same study the criterion-related validity among college students, comparing their first letter code with the first-letter code of their major found 40% were hits which, according to Holland, Magoon, and Spokane (1981) is in the range of hits found with most inventories examining criterion-related validity. When the hits for the CK were compared with those for the VPI, using weighted kappa, the CK did significantly better (.37 vs. .20).