Sufficient condition Data analysis

sufficient condition is tested in a theory-testing case study. This rela- tion can then be formulated as follows: Proposition 1: Value X A of A is a sufficient condition for value X B of B. In this proposition, X A is the value of A by which the subgroup is defined in which this relation was discovered e.g. a minimum level of management commitment and X B is the value of B observed in that subgroup e.g. success of a project.

9.1.7.2 Necessary condition

Next assess whether there is evidence for a necessary condition. A nec- essary condition exists if a specific value of concept B only exists if there is a specific value of concept A. The existence of a necessary condition can be assessed by ordering the data matrix in such a way that cases with the same value of concept B are grouped together. If the value of concept A is constant in a subgroup of cases with the same value of B, then this can be taken as evidence that the specific value of A is a nec- essary condition for the value of B. This relation can then be formu- lated as follows: Proposition 2: Value X A of A is a necessary condition for value X B of B. In this proposition, X B is the value of B by which the subgroup is defined in which this relation was discovered e.g. success of a project and X A is the value of A observed in that subgroup e.g. minimum level of management commitment.

9.1.7.3 Deterministic relation

Next assess whether there is evidence for a deterministic relation, mean- ing that an increase or decrease in the value of concept A consistently results in a change in a consistent direction in the value of concept B. The existence of a deterministic relation can be assessed by rank ordering the cases in the data matrix in accordance with the value of concept A. If, in the resulting rank order, the value of concept B con- sistently increases or decreases as well, then this can be taken as evi- dence that A and B have a deterministic relation. This relation can then be formulated as follows: Proposition 3: Concept A has a deterministic relation with concept B.

9.1.7.4 Probabilistic relation

Finally, assess whether there is evidence for a probabilistic relation, mean- ing that an increase or decrease in the value of concept A results in a higher chance of an increase or decrease in the value of concept B. The existence of a probabilistic relation can, again, be assessed by rank ordering the cases in the data matrix in accordance with the value of concept A. If, in the resulting rank order, the value of concept B seems also to increase or decrease, though not consistently, then this can be taken as evidence that A and B have a probabilistic relation. The same criteria for assessing whether the probabilistic relation actually exists between A and B in this data set apply as discussed in Chapter 7 for the testing of a probabilistic relation. This relation can then be formulated as followed: Proposition 4: Concept A has a probabilistic relation with concept B. If correctly derived from the data and, thus, proven to be true in the selected cases, the proposition is an appropriate result of the theory- building case study.

9.1.8 An example of data analysis

The following invented example of a data matrix generated in a theory-building study of factors that determine the success of innovation projects, shows ten very diverse cases, five with success and five without success Table 9.1. The table has ten rows, one row for each case, four independent concepts or “success” factors and the dependent concept absence or presence of success.

9.1.8.1 Sufficient condition

A sufficient condition exists if a specific value of concept A always results in a specific value of concept B. In this data matrix, we have four potential success factors and each value of each of these factors could be a sufficient condition for a specific value Yes or No of success. If we look at all four cases with value high on management com- mitment, we see that they all have been successful, whereas the two cases with low levels of management commitment have been unsuccessful. A high level of management commitment, thus, seems to be a sufficient