It's Only An Anecdote

In a recent post, I referred readers to a comment that had been left on another post. In the ensuing comment thread, I received a complaint that this was "only anecdotal evidence" . I should have cited some relevant literature to go along with it. That I needed to have "some science" in my post.

One of the many reasons for the existence of this blog is to tell stories about what happens in real women's lives - naming experience. Telling stories and naming the experience are worthwhile endeavors in and of themselves. It drives me nuts the way some people use "anecdote" as if it were equivalent to "uninformative, unreliable, meaningless". An anecdote is "a short account of a particular incident or event of an interesting or amusing nature, often biographical." Anecdotes - the biographical kind - are illustrative and sometimes just as powerful as all the cited studies in the world.

Even when you cite the literature, you can still find yourself dismissed. I've heard people respond, for example, to the stereotype threat literature "oh, that's all in an artificial setting, who knows if it happens in real life." Anecdotes are real life.

Seymour and Hewitt had this to say about anecdotes in their book Talking About Leaving:

In discussing our data with SME faculty, we sometimes encounter the objection that the state of affairs collectively portrayed in students' accounts is based on "anecdotal evidence". Strictly speaking (from its Greek root), an "anecdote" is an unpublished account. In more general usage, it is a story which is casually heard and has no coherent, patterned connection to other stories on the same theme. By either usage, the accounts which form the text data for this study are not anecdotes. Accounts which are gathered and analyzed in a systematic manner allow the investigator to discover things that cannot easily be discovered by any other means. In complex human affairs, noticing the patterns in the independent accounts of expert witnesses plays the same role as laboratory observations in the formation of hypotheses. As the reader will perhaps concede, there is much to be learned by treating such accounts with respect.

Now, I haven't systematically gathered and analyzed anecdotes like the one I referred my readers to. But over time, there have been many such anecdotes reported on this blog in the comments. The one I referred to seemed particularly poignant and worth attending to. We must, indeed, treat such accounts with respect. We must treat a woman's reflective description of her own experience as if it has some value. If we can't do that on this blog, then when and where is it going to happen? Unless you think the society we live in isn't really sexist and racist - in which case, no need to listen to the whiny bitches anyway.

Categories

More like this

Blargh to the kneejerk anti-anecdote folks.

Anecdotes can be used to:
- provide a counterexample,
- provide a cautionary example,
- provide an example of what is possible,
- illustrate aggregated data,
- suggest a new direction for query/research.

In the first three cases, anecdotes are perfectly appropriate to the task: that the anecdote is "only" a single example doesn't weaken what you're trying to accomplish with it.

In the fourth case, yes, it's good to cite the rest of your data. Depending on who you're trying to communicate with, however, the inclusion of anecdotes may be critical for success.

In that last case, the anecdote is a potential source for a vast new amount of information, but only if you don't dismiss it out of hand as "just an anecdote."

If it's 'science and stats', then deniers argue that it's just averages and impersonal statistics, and it doesn't speak to the experiences of an actual individual woman, who are presumably doing *just fine*. If it's anecdotes, then deniers argue it's not based in hard facts and on real distributions, and that that it doesn't speak to the experiences of an actual individual woman, who are presumably doing *just fine*. And thus we achieve a sort of Schroedinger's paradox for women's issues in science, in which any proof that women are having a shitty time is simultaneously neither quantitative nor anectodal enough. It's beyond 'moving goalposts', it's like a quantum singularity generating goalpost probabilities so indeterminate as to be inachievable.

Who was it who said, "The singular tense of data is not anecdote."?

By Jim Thomerson (not verified) on 12 May 2008 #permalink

If it's 'science and stats', then deniers argue that it's just averages and impersonal statistics...

In other words, reality unsoiled by whatever bias you are carrying. Can't have that. And what the hell would a "personal statistic" look like? Oh right, that's an anecdote.

That aside, I agree with the strategic sentiment of the poster in question. Anecdotes are too easy to dismiss by someone determined to not accept your case, regardless of subject. Better to do what my friend who just got her Masters in geology did to me: Tell me in no uncertain terms that the bias against women in the sciences was overwhelmingly supported by the data, particularly by simple studies of reactions to papers with and without female names attached to them (I don't have a cite but I'll bet Zuska does somewhere). That is tough to argue against even by the most mule-headed misogynist. And be sure you have an audience when you do this. It is a very good persuasion tactic. His reaction may help you sway them, if it appears sufficiently foolish.

It is good to remember that there is a certain level of intellectual dishonesty involved with anyone denying something well established by the data, and they will project this onto their interlocutors. They pick and choose their data and authorities, and make up anecdotes, so of course part of them will assume you do to.

The problem with anecdotes today is that by the means of modern communication you have an almost unlimitied supply of anecdotes for every claim, right or wrong. One could set up a blog about anecdotal evidence about UFOs, ghosts, angels, miracles and not go a single day of your life without a new post.

That is not to deny avery anecdote its truthfulness. It's just gotten harder to distinguish a false from a right claim if it's based only on anecdotes.
Which the gender bias case is not, just to be clear -- and anecdotes still supply life and personal accounts of what constitutes statistics.

It seems to me the pattern of behavior meh describes in comment #2 applies to other aspects of our culture, too. Many people seem to hold high-maintenance world views. Repairing their opinion sand castles sometimes looks like a full time job.

Our culture offers us many ways to distract ourselves by making our lives so much more difficult than necessary. Sometimes I think about what we might accomplish if that time and effort were directed toward lasting, sustainable projects. That's what keeps me going.

Cheers

Jim, did you even read the post?

No, he didn't.

This has been another in our series of Simple Answers To Easy Questions.

(Also, what DrugMonkey said.)

Another thought about anecdotes-if you have no data but a few anecdotes, they can be grounds for convincing others that research needs to be conducted. For instance, in situations such as Love Canal, where there were anecdotes of more people getting sick than average. I also have no problem even with people arguing for things like a link between vaccines and autism-if no research has been done. Once well-designed extensive studies have been done you should defer to the evidence, of course.

And anecdotes are great for communication, because that's how our brains work in general.

As I explained in the other thread I wasn't disagreeing with the point being made or the use of anecdotes in blogging. Also, the comments being made in this thread and the response by DrugMonkey have made me reconsider my views of the use of anecdotes in science in general.

My complaint was more based on the 'strategy' of using anecdotes in this way. As Zuska said this is the perfect place for sharing experience, but my impression was that when Zuska decided to elevate her appreciation of that comment and make a new post referring people to it (rather than just replying to it in the comments) it would seem like an 'official rebuttal' to the previous post. I thought that people who don't already share your view and are unaware of the research backing it up would find this unconvincing. Now, a commenter in the previous thread said that the purpose of this blog wasn't to convince 'naifs' but considering that one comment by a naif has spawned three posts with 70+ comments between them it seems that purpose or not that does go on here.

By chemniste (not verified) on 14 May 2008 #permalink

As the first commenter said, anecdotes have their place. My purpose in posting my anecdote was not to say "AH HA! See? Proof!" Rather, it was to illustrate what Zuska had posted about and share my own experience regarding the topic. It was also to show that, at least for one woman, what Zuska had said was true.

And honestly, even if it's only true for me, for one woman, that's still a problem. No human being should ever be shunned out of an interest or academic subject because of stereotypes. If it happens only once, it is something that needs to be addressed. We serve no one by sweeping my anecdote (as well as the anecdotes of hundreds and thousands of other women) under the carpet simply because we have only our voices and not statistical data to back us up.

Which, now that I think of it, is rather ironic. For so long, we have been shut out of academic discourse and the fact that we had no academic literature to back up our claims of this has been held against us! Well, we're in now, and there are finally studies being conducted to show that our interactions with people may shape our understanding as gender as much as, if not more than, our genetics. But not all of us have the benefit of access to that research. I have my voice, that's all. I shared that. Do with it what you will.

By Grimalkin (not verified) on 14 May 2008 #permalink

If this isn't terribly meta, I have an anecdote about a collection of anecdotes becoming a new research inquiry. A friend of mine is a school bus driver. Over a period of time, he noticed that an awful lot of his coworkers were showing up to work with bandages on, saying they'd had melanomas removed, more than he would have expected. He counted up the incidents he could remember, and reported it to a statistician working for the (Canadian) federal government. There's now apparently a study being designed to determine whether school bus drivers have a higher risk of melanoma than people in other occupations, and possibly what other factors might be causing it.

So in this specific case, you might say that anecdotes were data precursors. (Actually, I think that might be more generally true than in just this one case.)

By Interrobang (not verified) on 15 May 2008 #permalink

Interrobang: which side? If there's an occupational risk connection, my money's on the left side (assuming you're in the US). You could use taxi drivers as a control, and extend to a country where they drive on the left for futher control.

Anecdotes are rhetorical devices - they are not suitable for reasoned argument. And because rhetoric tends to overwhelm reasoned argument, resorting to anecdote is frequently interpreted as an attempt to sabotage the discussion.

Whether they are appropriate depends on what you're trying to accomplish.

By Caledonian (not verified) on 22 May 2008 #permalink

Anecdotes are rhetorical devices - they are not suitable for reasoned argument.

Bzzt!!! This is total fucking non sequitur.

The status of anecdotes as rhetorical devices is neither here not there in relation to their suitability for reasoned argument.

yeah. what interrobang said. a bunch of anecdotes about the same thing (as distinguished from *saying* the same thing) can be analyzed as aggregate data. "aggredotes."

and -- a single anecdote, if verified, can potentially negate an "always" statement, or a "never" statement. aaaaaaaall by its little lonesome.

and -- isn't it funny how we write off the anecdotes we *don't* like as hearsay, or as not statistically significant; whereas we don't construe those we *do* like in the same way at all.

The issue, I suppose, is in verification. If an anecdote can't be verified, then there is always a risk that it is nothing more than an individual's individual perception, and we all know that this can be flawed.

But if multiple sources report the same phenomena, then it most certainly deserves being investigated (preferably with some controls) and recorded -- and ultimately, this is what makes anecdote into "data", I would think.

It's a line you have to walk, because individuals most certainly can be wrong in their perceptions, but you also can't just blow something off because it appears to violate what you already believe.

By Luna_the_cat (not verified) on 07 Jun 2008 #permalink

If it's anecdotes, then deniers argue it's not based in hard facts and on real distributions, and that that it doesn't speak to the experiences of an actual individual woman, who are presumably doing *just fine*. And thus we achieve a sort of Schroedinger's paradox for women's issues in science, in which any proof that women are having a shitty time is simultaneously neither quantitative nor anectodal enough. It's beyond 'moving goalposts', it's like a quantum singularity generating goalpost probabilities so indeterminate as to be inachievable.

It is good to remember that there is a certain level of intellectual dishonesty involved with anyone denying something well established by the data, and they will project this onto their interlocutors. They pick and choose their data and authorities, and make up anecdotes, so of course part of them will assume you do to.