The document summarizes the creation of a prototype scorecard to measure and compare various dimensions of health across cities. It discusses abstracting health data from cities, statistically analyzing the data, and converting results into dollar values. It then evaluates the scorecard's truth value based on internal consistency (coherence), correspondence to existing regional classifications, practical implications (pragmatism), and ability to generate new insights (modus tollens test and "so what" question). The scorecard shows relationships between health measures and economic activity, suggesting improving low-cost public assets like parks could boost a city's tax base more than costly alternatives.
1. The Institute for Healthy Air, Water, and Soil asked me to create a
prototype scorecard for the harmony circle so I have three goals to
accomplish over the next few minutes. First, I’ll dash through how to
abstract the from the harmony circle relevant dimensions of data,
summarize those dimensions into numbers, and convert the results
into dollars. Second, along the way I’ll use five different ways to
assess the truth value of the scorecard. Finally, I’ll try to draw broader
lessons from the exercise.
2. I started by taking a sample of aspects of health, of similar cities, and of hard
endpoint outcome data. Soft data answers the question: Did you enjoy your
hospital stay? Hard data answers: Did you survive your hospital data. I chose
outcome data, because I think it is the only thing we actually care about. The
number of schools we have matters only to the degree that it brings about
student learning. Our hospitals are inputs; health is the outcome we seek. I
don’t really care how hard you worked. I care about the results you achieved.
I then passed the data through a series of statistical filters to get rid of
measures that didn’t work. For example, we can measure mathematics skill
with problems involving addition, subtraction, multiplication, and division, but
it turns out that word problems are poor measures of mathematics skill,
because they actually measure reading ability.
We pause now for a quick truth-check. One of the ways to assess the truth
value of a proposition is with coherence. Coherence looks for internal
consistency. The number to watch is the alpha. Anything above 0.7 is
considered good.
3. Another way to assess a claim’s truth-value is correspondence, which asks whether
the proposition corresponds to something else we know to be true. One long-
standing way to section the United States is along the Mississippi River and the
Mason-Dixon Line. Indeed, we see here that cities in the same region—such as St.
Louis and Kansas city—have a similar health profile (relatively high physical and
psychological health and relatively low cultural and economic health) even though
one of them rates highest and one lowest. Similarly, Louisville and Nashville look
similar to each other and Columbus stands alone. And finally, is anyone shocked
that the greater metropolitan area that contains Ferguson is the one that scores the
lowest?
Coherence and correspondence are the traditional philosophical ways of assessing
truth, but America’s sole contribution to philosophy suggests another way.
Pragmatism tells us to measure a proposition by what you can do with it. This
scorecard does suggest a path toward action. Louisville does relatively well with
environmental health and poorly with physical health. The obvious prescription is
that you need to get out more.
A clear thinker, however, does not take Yes for an answer. I started by saying we
need to look at not just inputs—such as schools, work, and hospitals—but also
outputs—such as learning, results, and health. Then I assessed the scorecard largely
relative to only what I started with, but if everything that has been said this
morning is true, then something different, something valuable, something—as
William James would say—with cash value should follow.
4. I compared the various measures of health with per capita retail sales in each
city. Individual aspects of health—even economic health—are weak and
unstable predictors of sales. Taken together, however, we see evidence for the
idea that this way to measure cities makes sense, that health really does
equal wealth, and that various aspects of health are interconnected. In turn,
this means that doing something fairly inexpensive—like cleaning up a park—
might add more to the tax base than a more costly alternative.
5. So where are we? We have a preliminary prototype with a small sample of
cities that doesn’t cover all aspects of health and there are better ways to
analyze the data and I’m sure more and better outcome data can be found.
Even more important is that I haven’t had time to ask: What would dis-prove
this approach? So it fails or at least hasn’t had the opportunity to pass the
modus tollens test.
The final way that I like to assess the truth value of a proposition is by what
other insights it can lead to so I always ask the question: So what?
Even with all of the shortcomings I mentioned, it can still be valuable to ask:
What might we learning from this exercise? I get three lessons. First, it seems
possible to measure cities in monetary terms on various aspects of health,
which—as we’ve seen—might be useful. Second, we will succeed only if we
look at the big picture of how the different aspects of health work in concert
with each other. Third, it seems that the only thing we control are inputs
(building schools and hospitals or working hard), but the only thing we care
about are results (education, health, and so on.) Fortunately, we will learn
though the rest of the day how to bridge inputs and outputs, our efforts and
our desired results.