time: time should be clearly specified (for example ‘attending this exercise class three
Ajzen (1988) proposed that these four factors should be assessed at the same level of generality/specificity. Of course, very high correspondence, such as predicting attendance at the fitness club on Tuesdays from items that refer only to exercising at the club on Tuesdays, will have limited generalisability and, indeed, may appear trivial. While vague attitude measures about ‘exercise’ may be less than precise and be weak predictors, we also want to have the ability to generalise, thus some compromise between the two may sometimes be necessary, depending on what it is we wish to predict. Kenyon’s (1968) ‘attitude toward physical activity’ (ATPA) scale, for example, assesses only the target (physical activity) and not the action, context or time. Perhaps this has led to some researchers being confused about the role of attitudes in physical activity. For example, Dishman and Sallis (1994) concluded, in their review of studies of supervised and unsupervised settings, that ‘attitudes’ are largely unrelated to physical activity. This is a surprising conclusion given the clear evidence showing
that both intentions and attitudes are related to physical activity when appropriate theoretica