Linear Regression Results For The Effect Of Expected Consequences Of Growth On Growth Willingness For The Full Sample.
|Full sample n=1158||Full sample withdummies n=1142|
|Survival of crises||.09**||.08*|
The most important reason for the modest explanatory power is instead, arguably, that regression coefficients in cross-section data represent average effects. The true effect for each individual differs from this average. For some managers quality is a great concern, whereas for others their own workload is the top-of-the-mind issue. This fact, i.e., that the coefficients represent average effects is a general problem and probably the major reason why explanatory power rarely reaches much higher values than ours even when seemingly all relevant variables have been included. The second major reason for this is that measurement error is always substantial in this kind of data, because the questions do not have exactly the same meaning to each respondent and because people differ in their response styles. We regard our modest explanatory power as a result of these general problems, rather than the omission of important explanatory variables.
Admittedly, however, there is one particular weakness
about our measures. We have, in the language of attitude theory,
regarded expected consequences as cognitive responses. This seems
unproblematic. Our classifying the over-all growth willingness
measure as a proxy for behavioural response is more debatable. We
would argue, however, that this is a problem in relation to
psychological theory rather than a problem of questionable logic.
That is, we find it
reasonable to regard expected consequences as causes of over-all growth willingness as measured. The alternative argument would hold that over-all growth willingness has an affective rather than cognitive basis, and that it is this over-all attitude that determines the evaluations of specific dimensions such as workload, work tasks, etc. This implies reversed causality and thus questions our use of the regression model. We would recommend researchers who would like to pursue the same theoretical perspective to use as dependent variable a measure that more clearly reflects goals, plans, or intentions to expand the firm, realising that finding or creating such a measure is no easy task
© 1997 Babson College All Rights Reserved
Last Updated 06/01/98