We describe five phases of research desired to refine the psychometric
properties of the PAC, a measure of psychological adjustment in cance
r patients. PAC items were originally generated by modification of an
existing diabetes instrument, and item discrimination examined differe
nces between 91 cancer patients and 173 age-matched patients with diab
etes. Cancer was considerably more stressful than diabetes but patient
s reported a similar range of qualitative responses. Focus group techn
iques were used to further refine the item set. A series of analyses i
n different samples identified six relatively stable factors and a Cro
nbach's alpha of 0.74-0.86. The test-retest correlation for the total
PAC score was 0.92 and convergent validity was shown by a consistent p
attern of association with criterion measures. In a randomised control
led trial, total PAC scores were shown to be sensitive to experimental
manipulations of the medium, message, and context of cancer terminolo
gy. A factor replication study of the 53-item scale in a sample of 283
patients with melanoma resulted in a shorter 21-item, two-factor scal
e with superior psychometric properties.