Table 5.

Psychometric properties of revised learning model scales.

Learning Model: Refined Scales	Mean	Standard Deviation	Number of Items	αa	rb	Scale Biasc	Scale Sensitivityd
Preceptor Support	3.94	0.69	6	0.847	0.510	0.104	0.336
Learning Opportunities	4.00	0.62	6	0.837	0.510	0.214	0.304
Patient Consultations (estimated number)	10.67	6.66	1	0.475	0.580	-0.182	0.715
Skills Improvement	3.65	0.86	6	0.873	0.520	0.093	0.622
Attitude Enhancement	4.03	0.65	6	0.837	0.520	-0.157	0.393
Overall Study Model Scale (refinedf)	3.89	0.56	25	0.920	0.796	0.057	0.574
Overall Study Model Scale (original)f	3.87	0.47	70	0.959	0.816	0.121	0.410

α^a – Cronbach's alpha-reliability, and the index of convergent validity; how well do items converge on each scale's central concept.

r^b – Highest correlation with any other scale. Low correlations mean the scale does not duplicate information in other scales.

^c – Pearson r with preceptor vs. student. Non-significant r's indicate absence of bias.

^d – Pearson r (or eta) with study arm. Positive correlations indicate sensitivity to predicting enhanced clerkship.

^e – Since the patient consultation estimate was a single variable measure, its α-reliability estimate is its SMR with the other four indices.

^f – Both Refined and Original scales are listed to document that streamlining resulted in no appreciable loss of scale power and efficiency.