When Statistical Data Misleads (in this case medical)

  • Posted by a hidden member.
    Log in to view his profile

    Feb 22, 2010 12:55 PM GMT
    I was reading an online health article today, that included a map of the healthiest and least healthy counties in the US. The map is reproduced below. And I was a little surprised by the distribution and extent of the unhealthy counties.

    I immediately recognized that some of the unhealthy counties coincide with Native American (Indian) reservations, many of which I've visited, areas of poverty and poor medical care. But the other patterns didn't make sense, like Maine and Nevada, with vast unhealthy areas.

    Then I read again what the criteria was: EACH state showed its 5 most healthy, and 5 least healthy counties, WITHIN THAT STATE. It wasn't a relative national comparison to a single US standard, but an internal state comparison, shown on a national map.

    And every state HAD to show 10 counties, divided into healthy & unhealthy, even if the worst counties for health in California were all better than the worst counties in Mississippi. Nor was there a portrayal of ALL US counties by a single national health standard, to show which states tended to better or worse health. States had to have 5 worst and 5 best, even if all 10 were relatively close in their rankings, and regardless of how many counties a state has. Texas has 254 counties, so 10 barely show. New Hampshire only has 10 total, so that's the entire state!

    Finally, I located the source foundation that produced this data, and saw a better resolution map with county lines. And it confirmed another suspicion I began to form: that some states have very large counties in geographical size, and the map colors the entire county healthy or unhealthy. So have large counties, and should your mandatory 5 least healthy include them, and your state looks terrible. In this way Mississippi looks to the eye to be healthier than Utah or Maine.

    I thought the lesson here was to always think twice about these charts & graphs, and what they're really giving us. The purpose here was to show patterns of health distribution, but the impression is of health comparison among the states. A poor chart, and poorly explained. But likewise can many forms of data be unscrupulously manipulated to produce deliberately misleading results.

    The chart as it appears in the article:

    AP-COUNTY-HEALTH_640.gif

    A much bigger original view of the chart:

    CHR%202010%20Top%20and%20Bottom%205%20co
  • Posted by a hidden member.
    Log in to view his profile

    Feb 22, 2010 4:14 PM GMT
    It's fun to play around with statistics and graphs. In fact, I spent a lot of my juinor high math classes learning how easy it is to manipulate any kind of data to show exactly what you want it to show.

    It's sad that I learned this when I was 13ish, but most of the country doesn't take the time to read into the graphs, charts, and raw data constantly thrown their way.
  • Posted by a hidden member.
    Log in to view his profile

    Feb 22, 2010 7:56 PM GMT
    GwgTrunks saidIt's fun to play around with statistics and graphs. In fact, I spent a lot of my juinor high math classes learning how easy it is to manipulate any kind of data to show exactly what you want it to show.

    It's sad that I learned this when I was 13ish, but most of the country doesn't take the time to read into the graphs, charts, and raw data constantly thrown their way.

    Well, it's not like I'm a math geek or statistical guru or something, far from it. I HATE this stuff. I am, after all, an artiste. icon_rolleyes.gif

    But I'm also skeptical and quick to second-guess, rarely willing to blindly accept what's dropped in front of me. Critical review is largely responsible for what made my Army career, always alert to calling into question all data and assumptions of others.

    To passively accept is to purposely abdicate. I reserve the authority of executive review of everything that crosses my path. I've had the wool pulled over my eyes too many times in my life to want to be tricked anymore.

    In my youth the cliche was the country rube who gets fleeced by the city slickers. That still happens today, but in different ways. We may not be country rubes, but there are still things out there that will fleece us if we don't watch out.
  • Posted by a hidden member.
    Log in to view his profile

    Feb 22, 2010 8:26 PM GMT
    "least healthiest" makes me suspicious right off. The map seems meaningless. It smacks of something put together to prove some dubious point, but it is not clear what the point would be.
  • Posted by a hidden member.
    Log in to view his profile

    Feb 22, 2010 8:41 PM GMT
    SeaInTn said"least healthiest" makes me suspicious right off. The map seems meaningless. It smacks of something put together to prove some dubious point, but it is not clear what the point would be.

    Well, I think their intended point is that rural areas with a low population tend to have poorer health care than urban and suburban areas. But the Achilles heel of this conclusion is that it fails to account for the practical problems, and the human reluctance, of traveling 100 miles to see a doctor on a regular basis.

    I lived in a US state where, following the 2000 census, some of its counties were declared "frontier" which is the US Census Bureau term for a place where the population density is less than 2 persons per square mile. This is a statement of the obvious, for some parts of that map.

    If you live in a desolate place like that, how the hell do you get health care? You might as well be living on the Moon. Is there a solution, other than moving into town? Would establishing health clinics every 20 miles be the answer, to serve but a handful of people?

    A study and map like this would have better value if it considered population density as well. What are the counties with viable populations, where health care is nevertheless inadequate, for reasons like low income, and low government assistance? And perhaps also poor education and few public health care programs? That's a map I would like to see.