Skip to main content icon/video/no-internet
icon backReturn to Entries

Pattern Matching

Pattern matching is the comparison of two patterns to determine whether they match (i.e., that they are the same) or do not match (i.e., that they differ). Pattern matching is the core procedure of theory-testing with cases. Testing consists of matching an observed pattern (a pattern of measured values) with an expected pattern (a hypothesis) and deciding whether these patterns match (resulting in a confirmation of the hypothesis) or do not match (resulting in a disconfirmation). Essential to pattern matching (as opposed to pattern recognition, a procedure by which theory is built) is that the expected pattern is precisely specified before the matching takes place.

Conceptual Overview and Discussion

The Concept of Pattern Matching

A pattern is any arrangement of objects or entities. The term arrangement indicates that a pattern is nonrandom. Theories predict some pattern of values of variables. Such predictions are usually called hypotheses. The term expected pattern will be used here for specifications of the hypothesis that allow for a rigorous comparison with an observed pattern of values of variables in a test.

Donald T. Campbell coined the term pattern identification as a characteristic of qualitative analysis that he defined as holistic (i.e., analyzing the pattern) rather than atomistic (i.e., analyzing its constituents). He argued that the single case study design could provide for a strong test of a theory if an entire set of expectations deduced from that theory (which together would constitute an expected pattern) could be shown to be true in that case. Campbell also called this a configurational approach. He insisted that qualitative analysis in this design tends to disconfirm rather than confirm a prior belief because of the requirement that, in the test, each separate element of a pattern or configuration that is observed is exactly as expected. As noted by Thomas D. Cook and Donald T. Campbell, the strength of this nonequivalent, dependent variables design is precisely that the variables that constitute the pattern or configuration are nonequivalent, that is, not substitutable.

Yin's Approach to Pattern Matching

Robert Yin discussed pattern matching as the most desirable analytic strategy in case study research. He identified two main types of pattern matching in theory-testing: (1) the pattern in a nonequivalent dependent variables design (in which the initially predicted value must be found for each element of a pattern of dependent variables) and (2) the pattern in a nonequivalent independent variables design. An example of the latter is a pattern derived from a typological or configurational theory in management. Yin stated that pattern matching in the dependent variables design should be rigorous, such that the hypothesis is disconfirmed even if only one variable of the pattern does not behave as predicted. For the independent variables design, however, he recommended a different approach. Yin stated that one should formulate different expected patterns of independent variables, each based on a different and mutually exclusive (rival) theory, and that the concern of the case study would be to determine which of the rival patterns has the largest overlap with the observed one. An additional complication in this approach is that Yin presented some examples in which the rival pattern represents not a real (theoretical) explanation but rather a version of a null hypothesis.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Read next

More like this

Sage Recommends

We found other relevant content for you on other Sage platforms.

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution