Well, after reading Paul Krugman’s encouragingly negative piece on Giuliani’s statements about cancer survival rates, I trotted off to research my own damning indictment of Giuliani’s propaganda against Democratic calls for healthcare reform. I should have known better. What I found wasn’t so encouraging.
Of all people, I should have known better. I’m English; my father died of cancer, in England. I know first hand that the American health care system, in some ways — like shorter waiting periods and accessibility of newer treatments — tends to be better than the English healthcare system. (I also know first hand that in other ways — like community outreach — the English system seems better than the American system.) Giuliani’s propaganda should be condemned, perhaps, but not because he lied; as far as I can tell he didn’t lie he simply chose to use the most conveniently damningÂ statistics.
It’s unfortunate that people abuse statistics as they do. The world would be a much better place if one needed a license to wield statistics. We need a license to drive a bus or to inspect an elevator, why not a license to fire off numbers?
Statistics should be carefully handled, thoroughly understood. Very few statistics tell their stories in straightforward, unequivocal terms. How were they gathered? what’s been included, what’s been excluded? what was the sample size? were other influencing factors eliminated, if so, how? And even if we understand the statistics, how can we use them in such a way that we disclose everything we know about the statistics so as not to mislead people?
Here’s an example: If someone were to release statistics about the time I get on the subway to head into the city on weekday mornings, they could cite my average alighting time as 8:06am. But the standard deviation around this mean departure time is over 49 minutes! Â Pretty eractic, what a flake.
Now, only those who had access to and took the time to study the underlying data would see that on three days each week I leave at 7:30am (so that I can go to the pool for an early dip). And on the other two days I get on the train at 9am after dropping my son at pre-school. Â I keep a pretty regular schedule around those times, with a standard deviation on any particular day of the week of no more than a few minutes.
We like statistics because they feel definite and concrete. They feel as if they will support the weight of some action that we can take to alter them in some way. If my departure time appears erractic, I must strive to be more regular in my schedule! But unless the statistics really do tell the story we think they tell, then they will only support incorrect conclusions and unhelpful actions.
Giuliani clearly liked the statistics that seemed to show that the English healthcare system was vastly inferior to the American healthcare system. He didn’t go looking broadly for statistics that would bear out this conclusion in all facets of the healthcare system. He didn’t drill down to show how the Democratic health reform bill would specifically lead to problems of reduced quality care in the American system. He simply plucked appropriately scary numbers off the statistic vine and tossed them out to support his aversion to increased government healthcare spending. Giuliani selected statistics that on their face lead us to want to reject the Democratic healthcare plan, whereas they don’t necessarily support this rejection at all. We would need to understand a great deal more about the proposed plan and about the potential impact on the quality of healthcare before arriving at such a conclusion.
It could be that the healthcare plan will improve the overall quality of care in America. I would imagine that this is what its proponents intend. But instead of finding out theÂ factsÂ we’re stuck talking about whether Giuliani lied or not. So often, politics gets stuck in meta-discussion, and we all lose out.