Dear Professor Swanson,

I had always assumed that the main purpose of a population forecast was to produce the best possible estimate of what the future population would be.  Your work made me think about forecasting differently. I was particularly interested in your recent work on probabilistic population forecasts, because it shifts some of the attention from the projected number itself to the uncertainty surrounding it.  That made me wonder whether a forecast can sometimes become more useful by appearing less certain. For someone making a real decision about schools, housing, infrastructure, or services, knowing how wrong a forecast might reasonably be, could matter almost as much as knowing its central estimate.  I may be understanding this too simply, but may I ask one question?

I’m curious about your experience: when uncertainty is explained repeatedly to decision-makers, do they generally become better at thinking probabilistically, or do they still tend to look for a single number as the final answer?

Best,

Melissa Reed

Dear Ms. Reed,

What a great line of thought and question. It is generating ideas I had not thought directly about. Thanks for sending it.

My experience leads me to think that decision makers tend to underestimate the uncertainty in a forecast more than they underestimate the consequences of planning re an accepted forecast. Part of the reason I believe this is based on my article, “Deep Structure Learning and Statistical Literacy,” which is essentially the ability to think probabilistically, a  characteristic I believe many people, to include decision-makers, do not possess.

Based on  work by King and Kirchner (1994), Roberts (2002) and Schield (1999), the following two tables taken from the article outline this process. The first shows the three major stages and the manifestations of the issues that characterize them.  The second table shows six conceptualization domains that foster the development of critical thinking: (1) alternative explanations and solutions to problems; (2) manipulation of symbols;  (3) ambiguity and uncertainty; (4) causality as a multivariate process;  (5) probabilistic interpretation; and (6) internal and external dialogues needed to communicate analytical results.  In terms of the three stages of the deep structure learning process (Table 1), each of the six domains has a characteristic feature (Table 2).

Best regards,

David A. Swanson

Table 1.  The Three Major Stages of Deep Structure Learning and Their Issues 

TABLE 2.  Conceptual Domains and Deep Structure Learning Stages

References

King, P., and K. Kitchener. (1994). Developing Reflective Thinking: Understanding and Promoting Intellectual Growth and Critical Thinking in Adolescents and Adults. San Francisco, CA: Jossey-Bass.

Roberts, K. (2002). “Ironies of Effective Teaching: Deep Structure Learning and Constructions of the Classroom.” Teaching Sociology 30 (January): 1-25.

Schield, M. (1999). “Statistical Literacy: Thinking Critically About Statistics”  Of Significance 1(1): 15-21.

Swanson, D.A. (2005). 2005  “Deep Structure Learning and Statistical Literacy.” Delta Education Journal 3(1): 41-52.