Chapter Listing | Return to 1997 Topical Index

Strategy variables

Acquisition of Firms and / or Technology Joint Ventures/Licensing
Cost Leadership Market Development
Differentiation Market Focus
Diversification Market Penetration
Divestiture of Products and / or Divisions Product Development
Formal Planning Quality Commitment or TQM
High Price Uniqueness
Innovation Use of New Technologies
Internal Growth  


Within the context of the dynamics of strategy and the need to study only firms within a narrow industry, the following hypotheses were tested:

1. Firms in different industries would adopt different strategies. It was assumed that firms would have different strategic typologies in different industries.

2. Over time, firms will have stable strategies. It was assumed that a firm's strategyshould be long term in nature, not tactical and changing every year. If this were true, then "strategies" would remain in place for a period of time, certainly more than one or two years.

3. Higher growth firms have distinctly different strategies than firms with slower growth. The literature, as previously discussed, had concluded that firms which were distinctive by their growth rates exhibited different strategy topologies among high and low growth firms.


Content Analysis

Over the years the authors have developed, pre-tested, tested, collected, coded, and analyzed many surveys, with response rates of 10 to 25 % being normal. As an alternative to primary data collection, content analysis of secondary data appeared to have some efficacy. Marino (1989) suggested that content analysis would be useful for studies of IPOs. Following Marino's suggestion, the authors content analyzed the President's Letter and the Management Report for each firm and for each year of the analysis. There are legitimate concerns about this technique relating to the validity and reliability (Weber, 1990) of results. Nevertheless, studies utilizing official corporate statements made by a firms' officers seem an appropriate methodology.

Dummy variables

The technique of utilizing dummy variables to represent an event or an occurrence (reporting of strategy/tactics) and relating that event through regression analysis in estimating financial results has been reported in the literature (Karafiath, 1988). The technique has been used in real estate (Gilson, 1992), law, healthcare (Vita, 1990), publishing (van Vliet, 1988) and in other areas. The technique has even been reported to be superior than conventional survey techniques (McCrohan and Harvey, 1989).

Non-response error

There is no non-response error in this study because the selected firms were not surveyed. The data set of firms is a subset of the universe of firms. Thus, results may not be representative of the universe. Secondly, the size of the data set and observations are enlarging, but still small, and results could change as the data set continues to be enlarged. Thirdly, as only one of the authors performed the content analysis, there is no interrater reliability issue to discuss, but the content analyzed data could be somewhat different, if a different person analyzed the content of the documents. Thus, the reliability and validity of the data could be questioned. The test of the utility of the techniques used in this or similar studies would be the results of the data analyses. It is left to the reader to decide as to whether or not the analysis is valid.


Hypothesis One: Firms in different industries would adopt different strategies

The first hypothesis was tested by simple contingency table analysis. The number of times each strategy was identified by the firms in each of the three firm types was recorded. A contingency table analysis resulted in a chi-square value of 44.307 with 32 degrees of freedom. The critical value of chi-square with 32 degrees of freedom at the 0.05 confidence level is 42.5658. Thus firms in different industries do adopt different strategies.

Hypothesis Two: Over time, firms will have stable strategies

The second hypothesis was tested using cluster analysis. Since differences in recorded strategies were found among the data sets when the sample of observations were split into the three SIC groups a cluster analysis using Ward's method was run for each SIC group. The set of seventeen dichotomous strategy variables served as the cluster variables. The cases were the annual observations of firm data taken from Compact Disclosure. If a firm was in the data base, for example, for seven years, 1989 through 1995, there would have been seven observations, one for every year.

Table IV displays the number of firms and their corresponding number of years that have been included in the data base aggregated by the two digit SIC codes.


If a firm followed a common strategy, then it would be expected that multiple observations of a single firm would cluster together during the initial cluster formation. If a firm changed strategies, there would be a break in the cluster formations and the observations of a firm would be included in two clusters. Continuing this logic, a firm would be included in as many clusters as the number of times it had changed its strategy. The clustering of the observations appeared to have little or no organization based upon the firm. Table V displays the five cluster result.

Top of page | Chapter Listing | Return to 1997 Topical Index

1997 Babson College All Rights Reserved
Last Updated 03/03/98