3.3. Rare events

If we accept the need for controls, then it becomes quite difficult to

provide a causal explanation for very rare or unique events. For example,

after someone commits a horrible crime, there is the inevitable spate of

‘‘explanation’’ stories in the media. These reports focus on some sordid or

unhappy aspects of the person’s background—a taste for music with violent

lyrics or violent video games, a broken home, distant parents, adolescent

alienation, or a childhood tendency to be cruel to animals. Such

themes make for a narrative that seems to culminate naturally with the

crime. These are often very good stories—memorable, dramatic, and they

satisfy our subjective sense of understanding. But do they accurately identify

the causal factors that led to the crime? After all, those who come

up with these narratives (and those of us who believe them) are seldom

constrained by any knowledge of base rates or serious empirical research

on violence.

How many of us have lived such fortunate lives that if a gaggle of

reporters were set loose on us, not one could construct a narrative on the

basis of our youth that plausibly culminated in a horrible crime (fictitious,

we trust)? In our cases, the greenest cub reporter could whip out a bloodcurdling

narrative along the lines of ‘‘quiet alienated youth goes bad’’ in a

day. As pillars of our communities, we intend to keep these stories to

ourselves. But it’s not just that lots of kids have stories uncomfortably similar to those told in the media about criminals. It’s also that we don’t

even know if those whose pasts are free of alleged ‘‘danger signs’’ really are

less prone to violence than the rest of us.

The lesson we would draw is not that narrative modes of inquiry are

irreducibly or intrinsically unsound. All of us offer narrative explanations,

particularly for people’s behavior. Rather, we would contend that storytelling

forms of inquiry are less reliable than most of us believe; and narrative

explanations deserve much less confidence than they typically get.

Without knowledge of base rates and without an accurate causal model,

we can only rely on sweet anecdote and the subjective plausibility of a

good story. On some occasions, these stories may act as useful heuristics.

But too often these stories rely on background knowledge that has not

risked the painful test of disconfirming feedback. Without awareness of

our own boundedness and frailty, then, we can only proceed with blithe

innocence, assuming that a good story is good enough. But as the literature

we have considered so far indicates, good stories come easy. True

stories are harder to find.

If we accept the need for controls, then it becomes quite difficult to

provide a causal explanation for very rare or unique events. For example,

after someone commits a horrible crime, there is the inevitable spate of

‘‘explanation’’ stories in the media. These reports focus on some sordid or

unhappy aspects of the person’s background—a taste for music with violent

lyrics or violent video games, a broken home, distant parents, adolescent

alienation, or a childhood tendency to be cruel to animals. Such

themes make for a narrative that seems to culminate naturally with the

crime. These are often very good stories—memorable, dramatic, and they

satisfy our subjective sense of understanding. But do they accurately identify

the causal factors that led to the crime? After all, those who come

up with these narratives (and those of us who believe them) are seldom

constrained by any knowledge of base rates or serious empirical research

on violence.

How many of us have lived such fortunate lives that if a gaggle of

reporters were set loose on us, not one could construct a narrative on the

basis of our youth that plausibly culminated in a horrible crime (fictitious,

we trust)? In our cases, the greenest cub reporter could whip out a bloodcurdling

narrative along the lines of ‘‘quiet alienated youth goes bad’’ in a

day. As pillars of our communities, we intend to keep these stories to

ourselves. But it’s not just that lots of kids have stories uncomfortably similar to those told in the media about criminals. It’s also that we don’t

even know if those whose pasts are free of alleged ‘‘danger signs’’ really are

less prone to violence than the rest of us.

The lesson we would draw is not that narrative modes of inquiry are

irreducibly or intrinsically unsound. All of us offer narrative explanations,

particularly for people’s behavior. Rather, we would contend that storytelling

forms of inquiry are less reliable than most of us believe; and narrative

explanations deserve much less confidence than they typically get.

Without knowledge of base rates and without an accurate causal model,

we can only rely on sweet anecdote and the subjective plausibility of a

good story. On some occasions, these stories may act as useful heuristics.

But too often these stories rely on background knowledge that has not

risked the painful test of disconfirming feedback. Without awareness of

our own boundedness and frailty, then, we can only proceed with blithe

innocence, assuming that a good story is good enough. But as the literature

we have considered so far indicates, good stories come easy. True

stories are harder to find.