What does “95% likely” mean?

The latest IPCC (Intergovernmental Panel on Climate Change) report says that it is “95% likely” that the main cause of global warming in the last 50 years is human activity. That sounds like it’s a probability, just as when you read that you have a 1/175,000,000 chance of winning the Powerball. The problem is, it’s not.  

That 95% is using a number to talk about something that is not numerical.  There’s nothing wrong with doing that – sometimes it’s a powerful way to express an idea.   It’s a metaphor, just like, “My love is deep as the ocean.”  But like every metaphor, it’s a very bad idea to take it literally. Since terms such as “likely” and “improbable” are used to talk about probabilistic events, such as winning a lottery or filling an inside straight, the natural tendency is to think that they are talking about the same kind of thing, some uncertain event whose “likelihood” can be measured by a ratio.  Having that ratio, though, is the crucial part of the picture. With lotteries, gambling games, actuarial calculations, public health information, even the famous “odds of being struck by lightning,” the numerical probability is a ratio, the ratio of the number of cases of interest to the number of cases overall.  When there’s no ratio, numerical language is a metaphor.

In the case of the IPCC estimate, there is obviously no set of cases.  They are, of course, not saying, “In a set of planetary climates similar to that of the earth, 95% of the cases of global warming were due to human activity.”  What then are they saying?

The IPCCis using a numerical metaphor to talk about a situation that is uncertain but not probabilistic, that is, not involving a ratio of numbers of cases.  The mistake of taking the metaphor literally leads to the mistake of thinking that there is a single thing, “probability,” that comes in two forms, “objective” and “subjective.”  It’s a mistake that has been made in Western culture for 350 years.  P. G. Ossorio once observed that a fictional asset is not a special kind of asset.  He was right – even though “fictional” is an adjective modifying the noun “asset.”  “Subjective probability” is the same kind of error.

Uncertainty is commonplace.  Lots of things are uncertain.  Whether I will win the Powerball is one.  Whether Iran will obtain a nuclear weapon, whether we will have another recession within 5 years, whether a Democrat will win the 2016 Presidential Election, whether I, personally, will get pneumonia within the next year, etc. etc. etc. are also uncertainties – but not the same kind of uncertainty as the Powerball uncertainty.  The pneumonia example illustrates a large class of very important uncertainties: single case uncertainties.  “Will I get pneumonia” cannot be a probability, because there is only one case, namely myself.  It is not the same question as, “What are the chances someone like me will get pneumonia?” Every time an investor asks, “What are the chances this stock will go up?” they are asking about a non-probabilistic, single-case, uncertainty.  Perhaps the most poignant is the patient’s question,   “What are my chances?”  As anyone who ever had a serious illness knows, that’s not a question about a group of people.

So what is the IPCC doing with their numerical metaphor?  They are expressing  a particular kind of judgment of the relationship between global warming and fossil fuel use, the kind of judgment called an appraisal.  An appraisal is description of a situation that automatically gives reason to act in certain ways; when you express an appraisal, you are saying you have reason to do some things, and not do others. If you hear me say, “This tastes great!”  you are not surprised when you see me eating it eagerly and ordering more.  It’s the same with appraisals of danger, which call for escaping  the situation. When the doctor tells the patient, “You have a 5% chance of survival,” they are giving the patient an appraisal related to end-of-life actions and the relative priorities those actions should have. If an investor decides, “It’s a 75% chance this stock will go up by 10% over the next 6 months,” they are expressing an appraisal of the act of investing in that stock, and the relative priority those actions have for them. Just as with the doctor and investor, the IPCC is giving a kind of appraisal, an uncertainty appraisal: a description of the relationship between global warming and fossil fuel use that automatically gives reasons to do certain things and not do others, and the relative importance of those things.  More specifically, they are saying, “If global warming is to be addressed effectively, addressing human fossil fuel use is a very, very high priority.   “Nine-five percent likely,” and all the other uses of numerical language to talk about non-probabilistic uncertainty, are uncertainty appraisals.

But let’s be careful: “95% likely” and “75% chance” are not decisions.  The decision is a choice of what to do, based on all the relevant facts and their appraisals.  That’s one of the things that upsets some people about the IPCC: to them, it appears that the IPCC is claiming there is only one thing to do, only one, overwhelmingly important, priority.  And that reveals one of the problems with “subjective probability” language: it leaves the actions, and their relative priorities, implicit.  And, thereby, unarticulated, unexplained, and subject to massive misinterpretation.  Taking metaphors literally is perilous.

So what difference does all this make?  A big one.  A probability is refined or improved by better case-gathering and counting.  When you have a “subjective appraisal,” you can’t make it better by adding more decimal places to the percentage. Uncertainty appraisals are improved by better specification of actions potentially called for, reasons for or against each, and relative priorities of those reasons.  That calls for a procedure of eliciting possible actions, identifying the various perspectives from which the situation needs to be viewed, and assigning sets of priorities to the possible actions in accordance with each perspective.  How to do that effectively in a group in which many perspectives are present is the topic of a paper in preparation.

This entry was posted in Behavioral Economics, Probability, Subjective Probability. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s