TABLE OF CONTENTS The description of the path analysis above focuses on the actions of web- activity-dependent firms. While the sample studied here includes a small number of observations for business models in which activity is not ex ante expected to be a substantial source of long-term revenues (Kozberg, 2001), these firms are likely to prove exceptions to the rule. If firms are attempting to maximize revenue streams from multiple sources, primary or not, then website activity should translate into increased revenues for these companies as well.
Due to the use of partial regression coefficients in the path analysis, it would first be helpful to examine the overall correlations among the variables tested. [12] The correlations in Table 4 are sorted from left-to-right (top-to-bottom) based upon the particular variables' position in Figure 2. From Table 4, it can be seen that a number of pairs of variables are highly correlated, such as pageviews and advertisements shown (0.74). This result would seem to support the need for a mechanism to control for possible endogeneity problems suggested by high multicollinearity in the data. From the organization of the data, it can be seen that these high correlations among the variables tends to fall as the number of hypothesized steps between them increases. These correlations are, therefore, consistent with the predicted effects in the last section (e.g., pageviews influences advertisements shown which in turn has some, albeit smaller, effect on click-throughs as a result of this intermediate step).
