Emotional Incident Methodology: Theoretical Underpinnings

Social psychologists who use inductive research techniques ask the respondents to tell them what they believe, how they feel, or what they did (e.g. to capture beliefs, attitudes, and behaviors) and then the psychologist clusters the responses into categories.  Social psychologists that use deductive research techniques create lists of attributes from research and ask respondents to cluster those findings into categories.  The inductive researcher creates the categories herself.  The deductive researcher relies on the wisdom of crowds to create the categories.  However, the inductive researcher relied on the wisdom of crowds to identify the original attributes to be clustered, whereas the deductive researcher relied on her own wisdom to identify and select what research to use.  There are clear pros and cons to each research methodology.
Emotional Incident Methodology is a hybrid of both social psychology techniques.  The original factors that are to be categorized are provided by the respondents.  A separate group of people then cluster those responses.  This is actually a fairly common event in social psychology research.
In classical deductive research, the psychologist starts with a theory and a hypothesis.  Observations are used to test the hypothesis.  
In classical inductive research, the psychologist starts with observations that lead her to a hypothesis.  The hypothesis is then used to create a general theory.
Pure practitioners of both research methodologies will criticize Emotional Incident Methodology.  The deductive researchers will say that the results are invalid because the project began without a base theory or hypothesis.  (This is the criticism often made about The Big Five personality attributes, but that is the only model that has any validity when trying to predict workplace performance.)  The inductive researchers will say that the results are invalid because the project ended with a group of lay persons determining the final hypothesis and theory.  They believe that only trained experts should be making the final determination.
At the end of the day, there is a solid reason why people often say, "There are lies, damn lies, and then statistics."  Research methodologies are all flawed when it comes to measuring the human condition - we are just too complex.  Causality is impossible to prove because no research into human beliefs, attitudes, and behaviors can replicate the same results 100% of the time.  We also live in a personal bubble of perception because we individually decide which facts are true or not.  In fact, if we behave in a way that is inconsistent with our beliefs and attitudes, we simply create new beliefs to justify our attitude and behavior in that moment.  And what researcher is completely free of her own biases?  That would have to start with a complete and accurate understanding of self that no one has.  That is why triangulation is used in Emotional Incident Methodology.  Collectively, the group is less fallible than any one individual, even a trained expert.
If enough objectivity permeates the results of Emotional Incident Methodology then you will see measurable results when you apply the lessons learned.  My own findings at Royal Caribbean Cruises and Silver Hill Financial led to statistically significant results.  But I cannot claim absolute causality because there were many other variables at play within the organizations, industries, and economy.  Again, the complexity of the situation prevents the type of certainty that classic researchers demand.
Emotional Incident Methodology is a post-positivism approach that rejects the relativist idea that we cannot reach agreement because our unique life filters and biology have resulted in us having different biases.  It is a constructivist approach that relies on triangulation to gain objectivity across multiple fallible perspectives.  After all, social psychology is the study of individuals within the context of a group and no two people in any group are exactly the same in every respect, not even identical twins.