Last week, I kicked off 52 weeks of data pattern analysis. This is week 2. My goal for this series is to see if I can explain some new and unexpected findings related to my scientific research in criminal justice, criminology, and health. The past attempts that I have made to explain my results seem to have failed to generate any clear understanding. In my first post for Week 1, I mentioned the age crime curve and health cost curve puzzles as two examples of solutions I have found where I can’t seem to explain my results, but there are several other discoveries that I would like to be able to discuss as well.
I promised that I would try to keep the overall explanation simple enough for the general reader to understand. So for this week, I would like to discuss the concept of “paradigms.” A paradigm is a shared consensus about how science is supposed to be conducted.
https://en.wikipedia.org/wiki/Paradigm
Thomas Kuhn popularized the concept of paradigms in his book on the structure of scientific revolutions.
https://en.wikipedia.org/wiki/The_Structure_of_Scientific_Revolutions
Kuhn suggested that science does not progress smoothly, but advances in an uneven fashion, with slow and gradual advances for several years in a row, and then fast change. During the slow and steady advance phase, which Kuhn calls “normal science”, there is a set of “paradigms” or shared understandings of the generally accepted theories and research practices.
https://en.wikipedia.org/wiki/Normal_science
During the “normal science” phase the scientific paradigms are able to provide a set of relatively coherent solutions to the major scientific puzzles scientists are trying to solve. Then, over time, some new puzzles emerge which can’t be solved using the existing paradigms, and it becomes clear that the existing paradigms are flawed. When the flaws in the existing paradigms become obvious enough, a new approach is developed and a “paradigm shift” occurs.
https://en.wikipedia.org/wiki/Paradigm_shift
In my research, I discovered several instances where the prevailing scientific paradigms were flawed, and I had to make personal paradigm shifts in order to proceed. The process of finding flawed paradigms began with my master’s thesis.
I think that it is important to explain what happened with my thesis, even though most of you could care less about the subject I was studying, since the findings from my thesis were what caused the beginning of my 10 year, million dollar, odyssey into the analysis of change and the solution to the age crime and health cost curves. In my thesis, I found a flawed paradigm.
My thesis was based on a paradigm called “dynamic predictive validity” that was popular in the criminal justice literature at the time. The dynamic predictive validity paradigm states that if you measure something twice, and whatever you are measuring has changed between measurements, the second measurement will be a better predictor of outcome than the first measurement in the period after the second measurement. In my thesis, I was measuring the risk of reoffending for criminal offenders on parole, and testing to see if repeated assessments improved the prediction of reoffending rates.
The dynamic predictive validity paradigm had been tested many times and never failed to generate better predictions on the second assessment. In my thesis, I changed the test a little, adding tests for improved prediction between the 2nd and 3rd assessment, and the 3rd and 4th assessments. In these cases, prediction failed to improve. The paradigm seemed to be flawed somehow.
Why should prediction improve between the 1st and 2nd assessments and not between the next two pairs of assessments? Was the propensity for criminal offending really changing or not? Scores changed. Did risk change?
I would like to stop here and continue on the next phase of my research next week. In that post, I will discuss some research I did on the nature of change. In the years after I finished my thesis, I became obsessed to finding the answer to the puzzles it generated. This was where I truly began to question the accuracy of our paradigms about the nature of traits like the risk of criminal offending.
Recent Comments