In the concept mapping context, pattern matching is done at the level of clusters. Pattern matching allows us to combine any two measures aggregated at the cluster level to see to what degree the measures match or whether they disconnect. By examining such combinations of measures, we can address critical questions of interest to the group or organization.

Pattern matching is simple to implement and extremely powerful in its implications. Pattern matching always involves two patterns. The patterns are based on measurements taken at the statement level. Almost any kind of measure can be used, depending on the purpose.

Measures might include types of importance ratings we collect as part of concept mapping, ratings of how much different components (e.g., units, departments, training components) should cover or address in each statement, measures of how well each statement is addressed in the implementation of a program, costs of addressing or implementing each statement; or the amount of change or gain that occurs for each statement as a result of some program or treatment. The measures might be ratings, dollars, achievement scores, quality scores, or just about anything else one might measure for each statement.

All measures are, for convenience, classified into two types:

Theoretical: This type of measure reflects what some individual or group thinks will happen or what they would like to see happen (e.g., importance ratings).

Observed: This type of measure reflects something that has happened (e.g., an implementation, performance, or outcome measure).

In pattern matching, we indicate a Theoretical measure with a T and an Observed measure with an O.

Pattern Matches

A pattern match itself consists of two elements. First, there is the visual picture of the match. Second, every pattern match has a correlation coefficient associated with it.

The visual picture of the match is shown through a ladder graph which is essentially two vertical scales (one for each measure) joined by horizontal lines for each cluster, showing comparative performance on the two measures. If the match is a perfect one, the lines are all horizontal and the resulting graph looks like a type of ladder. Ladder graphs are especially useful for quickly spotting disconnects between two measures. For example, the following graph shows a consensus pattern match between managers and staff in a strategic planning project:

Obviously, in this example there is not much consensus between the managers and staff (except on the relatively low importance of Community Relations).

The correlation coefficient associated with each match describes the strength of the relationship or match between the two variables. The correlation ranges between -1 and +1. Values near 0 indicate the absence of a match; values close to either pole indicate stronger matches. Negative values imply an inverse relationship (when one measure is high, the other is low and vice versa). Positive values imply a synchronic relationship (high with high and low with low). Together, the ladder graph and correlation describe the relationship between the patterns of the two measures.

Three basic types of pattern matches are possible, depending on which types of measures are combined in the match:

Consensus Pattern Match (T-T). In this match, the theoretical (T) ratings of one group or individual are compared with those of another group or individual. Essentially, we are looking at the agreement or consensus that exists between the measures being compared.

Consistency Pattern Match (O-O). In this match, the observed performance (O) of a group or individual is compared with that of another. Essentially, we are asking how consistent these measures are with respect to each other.

Outcome Pattern Match (T-O). In this match, an observed outcome or performance measure (O) is compared with what was expected (T). Essentially, we are examining whether we achieved what we hoped to accomplish.

Following are three examples of pattern matching questions that make use of concept maps--one for each of the three types of pattern matches:

Consensus: Is there consensus across groups or individuals and how might we achieve it?

Consistency: Is the quality of a process maintained over time?

Outcome: Is a program having its desired outcomes?