Collecting and analyzing data requires categories, as does making predictions, exploring causes, and taking actions (see earlier post). This post: a) identifies a set of questions that might expose problems with our choice of categories; b) presents a set of considerations that underlie researchers’ choices; and c) introduces a framework for thinking about modifying the categories by modifying these and other considerations.
First a story: Once a doctor prescribed me three tablets of some medicine per day and I asked about the dosage, observing that I’m not a small person and I have a high metabolism. His response was “medicines don’t work that way.” I questioned him on how they do work, but when he fudged an answer, I dropped the matter and, not so long after, the doctor. Was I too hard on that doctor—what would it mean for the two of us to customize the dosage to match my body weight? He might have to specify slicing pills in two or three and I’d have to be a patient who would “comply,” that is, take the time and care to do that. Moreover, if there were research on metabolizing that medicine in relation to body weight, the doctor would have to know about it, be able to access it, and do the required calculations on the spot. And, if there were such research on some similar drug but not on this one, he’d have to make a judgement call in order to extrapolate from that research to this situation. Moreover, if other factors were involved, such as my metabolic rate, the doctor would have to do more tests to ascertain what that rate was for me. In short, there are lots of practical consequences if the doctor decided to customize my drug dosage rather than follow the guidelines given in his pharmacological handbooks, guidelines that presumably treat me as belonging to the category adult. Of course, there are sometimes clinical consequences in not customizing dosages. A quick google search exposed an item with this quote
Children and infants are particularly at risk of medication errors mainly due to incorrect dosage, because of the need to modify dosages based on age and weight. The dosage modification may be either overlooked or miscalculated. Various studies have shown high error rates in doctors and nurses in calculating weight-dependent dosages in infants and especially neonates.
a. A set of questions that might expose problems with our choice of categories:
- Have we omitted relevant categories?
- Have we mixed different phenomena under one label?
- What basis do we have for subdividing a continuum into categories?
- Have we divided the continuum at the right point?
- How do we ensure correct diagnosis and assignment to categories?
- What meaning do we intend to give to data collected in our categories?
- Is our category correctly named?
b. A set of considerations that underlie researchers’ choices (from Pickles & Angold’s  discussion of continuous dimensions versus categories in the definition of psychopathologies):
- the pragmatics of measurement
- the needs of clinical practice
- our ability to distinguish categories from dimensions empirically
- methods of analysis appropriate to each and how they relate
- the potential theoretical biases associated with each approach.
c. A framework for thinking about modifying the categories by modifying these and other considerations: Explore the practical consequences of using different categories (which requires exploring to find ideas about alternatives to established practices and the received wisdom). These consequences may be writ large, e.g., Galton only collected data about similarities among relatives so there was no way he could explore hypotheses about non-hereditary or environmental influences or illuminate action regarding those influences and so his research bolstered eugenic approaches to perceived social problems. Or they may be writ small and multiple and heterogeneous, e.g., in my story about customizing drug dosage, we could tease out additional issues about guidelines for insurance reimbursement, training of the pharmacists, constraints on amount of time doctors can spend in consultations with patients, malpractice law, the severity or otherwise of side-effects of misdosage for a healthy adult, and so on. Or the practical consequences maybe somewhere in between, as suggested by the ladder of steps from phenomena to action, with each step in the ladder providing a place where decisions have to be made (that is, they aren’t given by nature or the “real world”).
The writ-large approach leads to Big Critiques, e.g., dismiss Galton’s work as contributing to racism (via eugenics) (see Chase’s Legacy of Malthus). The writ-small approach identifies many different sites for engagement in change, each of which is do-able by some party given their skills and background, but none of which is guaranteed on its own to overcome the problem. Indeed, people doing the engagement and the sites of engagement need to be linked, and, even then, there is no guarantee that there won’t be unintended consequences that will warrant rethinking of categories and modification of engagement around practical consequences (see discussion of heterogeneous construction in this post and subsequent posts, as well as Taylor 2005).
An exercise for students or other readers: Identify an example of a problem related to each question in a. What can be done in practice to overcome the problem? In what ways does the problem and what can be done—or not—illustrate the list in b. or the framework in c?
Pickles, A. and A. Angold (2003). “Natural categories or fundamental dimensions: On carving nature at the joints and the rearticulation of psychopathology.” Development and Psychopathology 15: 529-551.
Taylor, P. (2005) Unruly Complexity: Ecology, Interpretation, Engagement (Chicago: U. Chicago Press)