“Lies, damned lies, and statistics” - Part 1

I just heard Hillary Clinton utter it again last night, “47 million people without health insurance”. All too often statistics get thrown around, and at some point take on a life of their own; this would be one of those cases. As Disraeli lamented about statistics, all too often they are misused to bolster inaccurate arguments.

This blog posting will analyze the ‘47 million uninsured’ number, not for the sake of discussing the health care issues confronting the US, rather to provide an approach to analyzing statistics presented in articles.

Step 1)

The most important thing to do, is to understand what the specific criterion for any statistic is; in this case, what does it take to be counted as one of the 47 Million people who ‘don’t have health care insurance’?

To qualify as uninsured, a survey respondent must have claimed that in the previous year they did not have any insurance coverage. There is no reduction made for people who qualified for programs but did not enroll, estimated in 2004 to be 14 million. No reductions are made for illegal immigrants who would not technically qualify for any universal program, estimated to be 10 million. Lastly, no reductions or 'qualifications' are made to reflect whether the individual 'could have' afforded health care insurance, it's estimated that 18 million of the 47 million had household incomes of over $50K/year.

So, is the real problem the 5 million who don’t fall into one of these three buckets? Note, there is likely little overlap between the three groups outlined, as $50K disqualifies you from most government programs, and more illegals don't make over $50K.

Step 2)

Always be weary of absolute numbers, more times than not, they are too large to comprehend, and are often used because they don’t give the reader a perspective of relevance. As such, the first thing you should do when presented with an absolute number is to convert it into a percent.

In this case:

• The % of the US population without health care insurance, 47M out of 301M = 15.6%

Sometimes inverting the number helps also:

• The % of US population with health care insurance, 100% - 15.6% = 84.4%

Now, re-evaluate both statements:

A. 47,000,000 US citizens do not have health care insurance
B. 84.4% of US citizens have health care insurance

Does one seem more dramatic than the other? When was the last time you were concerned about something that ‘only’ 84.4% of people had? Why do people use the 47,000,000 statistic instead of the seemingly smaller 15.6%?

Step 3)

Reevaluate the policy prescriptions in light of the new ‘re-framed’ statistics.

Ask yourself:

• Why do advocates of a certain position keep using numbers like ‘47 million uninsured’ to make their point, even though the number is neither an accurate representation of the current situation; given that the majority of uninsured could either afford to buy it, or be insured through state plans?
• Why aren't people focused on trying to get more people to a) enroll in programs they already qualify for or b) to prioritize spending in health care above other expenditures? It seems that the 47 Million masks three (or more) problems.

In Summary:

This post is not disputing the ‘47 million’ number, rather it’s simply pointing out that readers need to be willing to dig deeper into statistics that are presented to them to understand what they really mean.

When thinking about statistics, a better adage than Disraeli’s may be … “Statistics don’t lie, people do”
_____________

UPDATE

• Beyond Those Health Care Numbers
• Paul Krugman, Your Doctor Is Calling
• Beyond Those Health Care Numbers: US Looks Good
• Crisis of the Uninsured 2008

My take …

So where do I stand? After a lot of reading on this topic, I’m most shocked by how little is actually known about this problem. Compared to things like unemployment numbers, where you can track details like short-term unemployed versus long-term unemployed, changes in pay and benefits as people take new jobs; voluntary vs. involuntary changes; health care statistics very rudimentary.

Why does this matter? Without the data, it’s difficult to really know what the problem is. If 18 million of the 47 million are without health care insurance could afford some sort of health care; then passing a law that requires you to buy health care might seem like a good idea ... except, almost every state 'requires' people to get auto-insurance, yet in CA it's estimated that over 25% of drivers don't carry auto insurance; I'm not sure why a similar law requiring health care would have any different results ... in fact, the % of people without health insurance is already less than 25% without auto insurance, even though auto insurance is legally required.

Similarly, if 14 million people are eligible for health care plans (Medicare, Medicaid or SCHIP) today, but don’t enroll, making it a law won’t seem to address this problem either, unless you think these people will be motivated by a law more than their health. Generally, it's a lack of knowledge about the programs, that causes the low enrollment; perhaps funds would be better spent on education than enforcement.

There are undoubtedly millions of Americans who go without health care, that don’t qualify for government plans, and can’t afford health care insurance. It seems that these uninsured people are so, because of high health care costs; meaning that any solution would need to address these costs. Passing a law that requires someone to buy health care, doesn’t address this problem, especially if they can’t afford it in the first place. Worst of all, we don’t exactly know how many people are in this situation.

Without a more accurate holistic picture of the situation based on accurate data, you can’t begin to understand the problem, let alone start proposing solutions.