Yesterday I discussed the use of statistics to predict how likely a person in recovery is to maintain sobriety. These sorts of statistics can be discouraging, as they often make it seem like the addict is unlikely to ever obtain lasting sobriety.

I made the case, however, that statistics have their time and place, but cannot be applied in *every* situation. They are useful for representing the level of uncertainty in the observer, but not for describing the actual state of the observed. Thus, a coin that is covered only exists in one state (heads *or* tails), but the uncertainty of the observer is divided between two (heads *and* tails).

Today I want to consider another fact of statistics which makes it unhelpful in determining what your personal future entails. That fact is that statistics are a tool for measuring groups, but not individuals.

### The Sad Case of Sally Clark)

There is a tragic real-life example of how the misapplication of a broad statistic to a single individual is both in appropriate and dangerous. Sally Clark was an English woman who lost one son to SIDS (sudden infant death syndrome) in 1996, and then lost a second son as well in 1998. This was terrible enough, but then it was made worse when Clark was tried for murder. One SIDS death in the family was believable, but two?

As it turned out, there was no concrete evidence to show that Sally Clark had killed either of her two children, but the prosecution brought in Professor Sir Roy Meadow to describe the statistical probability that this mother would have lost two sons to the same rare cause.

Professor Meadow argued that the odds of such an occurrence would be 1 in 73 million, an event so improbable that it could be rejected as a virtual impossibility. Mathematically, he said, Sally Clark *had* to be a murderer. The jury was convinced, and Sally Clark was sentenced to life in prison.

Four years later, though, it was discovered that the lab reports on the deceased children had omitted clear evidence of the two sons’ deaths being due to natural causes. Clark’s conviction was overturned, and she was released. At this point, though, she had already slipped into a terrible depression, and died shortly thereafter of alcohol poisoning.

Professor Meadow’s application of statistics was torn apart by other mathematicians and statisticians. The numbers he arrived at were simplistic and faulty and he never considered calculating what the likelihood for a mother committing double infanticide is, about 4.5-to-9 times more unlikely than double-SIDS as it turns out! Perhaps most importantly, though, he had made the critical mistake of applying *group statistics to an individual*.

Consider this, if the odds of a double-SIDS family were genuinely 1 in 73 million, then that would mean for every 73 million mothers you would expect at least *one* to have lost two children to the phenomenon. 1 in 73 million does not mean that the event does not occur, it means that it *does*, and you will start seeing multiple occurrences once your population pool is large enough. 73 million is large, but it is not unfathomable for a population. It was inevitable that *someone* would show up with this situation at some point or another.

By Professor Meadow’s logic we could look at every mother on earth, one at a time, and for each individual conclude that it is too improbable to believe she has lost two children to SIDS. And thus, we would go through the entire population, believing none of them could have suffered that ordeal, when by Meadow’s own statistics there would have been over a hundred women who actually did.

It is the same with addiction recovery. Statistics can define the pattern for a group of addicts, as every group will inherently have a certain likelihood for certain behaviors. But when we apply those likelihoods to the individual we make the subtle, but damning mistake of saying that if something is improbable for *everyone*, then it is improbable for *anyone*.

### Groups Within Groups)

Statistics can model the group, but they cannot model you. And they especially cannot model you when you consider that inside every group there are more specific subgroups, each with their own accompanying statistics.

By this I mean that most statistics on recovery cover a very broad spectrum of addicts. The odds of sobriety that they give tend to include individuals who have been working a 12-step program for twenty years, and individuals who showed up for the first time today. It includes those that have come of their own volition and those who came only because a judge ordered them to. Clearly not everyone in this broad group is as likely to remain sober as every other.

Every time you take the next step in recovery, you come into a new subgroup, which is represented by better and better odds. Maybe 1-in-10 of all addicts will stick with recovery, but just by returning for the second week you might now belong to a group with a 1-in-9.5 success rate. Do your “step 4 inventory” with another member of your program and you come into another subgroup with even higher levels of recovery. Earn your one-month chip and enter yet another higher-recovery subgroup.

Not to suggest that you are improving your own odds, only that the *group* statistics are converging more and more to who you actually are. Or at least they would be if the research were conducted down to such granular levels. But you don’t have to take my word for it. Start going to recovery programs for any sustained amount of time and you will quickly see that there is a clear correlation between amount of time working on recovery and length of sobriety.

### You Are You)

Hearing discouraging statistics might make you feel like you don’t have a chance of recovery. But never forget that in order to even define the odds researchers must first find a number of successes. They couldn’t say 1-in-10, or 1-in-100, or 1-in-anything until they had found that 1. Never mind what the group pattern might be, the fact remains that there are people who *do* achieve sobriety, and whatever their methods to get there, it’s safe to assume that they weren’t putting too much stock in “what their odds” were. They got better because they got better. They weren’t 1 in a group, they were 1 in themself, and they decided for themself what that themself was going to be.

Governing your life by statistics is not only a misunderstanding of the science, but also a dangerous game of self-fulfilling prophecy, one that can ironically change the statistics on the subject. There is no statistics on you as an individual. There are statistics on the group, and they are useful for understanding the group, but the group is not you. The group can be documented, but to quote Lawrence of Arabia, for you as an individual “nothing is written.”