I know of one grocery store where the vast majority of the shoppers were white and a small minority were black. But virtually all the shoplifting was done by blacks (according to the store manager) so on rare occasions that a black entered they were watched like hawks.
It’s a bit more complicated in that instance, because the store was at the border of a middle-class white neighborhood and a ghetto-type black neighborhood, so it wasn’t just about skin color. But that’s how that manager identified the targets.
One thing to consider is that in some businesses (like groceries) the mark-up per item is relatively small, so one successful shoplifter can wipe out the profits of many many sales. This impacts the cost-benefit analysis of preventing shoplifters versus alienating shoppers.
That’s a bad example. El Al most definitely uses ethnicity as a big part of their profiling. It’s not all of it, but it’s definitely a big part of it.
I think there are very few people who would advocate (or practice) profiling where race is the only factor (e.g. more suspicious of a middle-aged black man with a suit than a trashy looking white teenager).
Meh. So economic conditions are correlated with crime statistics. No surprise.
While racial profiling may or may not be a system that works (if only psychologically) for the manager, it’s still inefficient, as the NYPD has found.
People can use whatever system they want to judge. That’s not really the debate. The debate is how “logical” it is. In this case, it’s not especially logical, except to the extent it’s also correlated with other factors, including economic.
Again, the high false positive rate is the key aspect. Most judiciously designed detection methods try to balance a low false positive rate with a sufficiently high detection rate. It’s a poor system that turns the knob all the way up and detects both at high rates.
It’s a part of the analysis but only a part. Using race or ethnicity as the sole proxy is inefficient. Using it (and other factors) in a multi-variate analysis may actually be efficient. That’s the point they have, and actually a criticism Israeli security officials have made about racial profiling systems in the US.
And that’s rather what the OP is suggesting - using race as the primary tool for identifying potential shoplifters. People actually ARE suggesting using race as the main factor.
If that’s how the manager identified the targets, it’s no wonder that those are the people he caught shoplifting.
This study (which I unfortunately cannot find available through a non-academic source) is called “WHO ACTUALLY STEALS? A STUDY OF COVERTLY OBSERVED SHOPLIFTERS.” It sets up a very rigorous methodology for watching for shoplifters with no opportunity for bias, at a drug store near Atlanta with a very diverse clientele.
I confess I don’t understand all their statistical analysis. However, some of their findings are very interesting:
Right. But as a practical matter, this guy was racially profiling, even if the underlying factors were economic. If a black guy walked into the store with no intention to steal anything he still got watched.
As noted, it depends on the circumstances. What may be inefficient for the NYPD might be efficient for that shopkeeper.
I don’t see where the OP said race was primary, as he specified a “suspicious” black guy which implies other factors.
But at any rate, much of the discussion in this thread seems to be aimed at precluding any use of race, to which your El Al example would be a counter-example.
Assuming that he began profiling them off the bat and not based on prior experience. We don’t know that.
Per your quote there were some examples of race/sex being valid predictors (though whether it would make their use efficient would depend on how strong the correlation is, which is not specified in your quote. E.g. if Hispanic women were 1% more likely than the other groups to shoplift, there’s not much you can do with that).
But the broader point is that even if shoplifting is exactly evenly distributed by race/gender, there could still be specific circumstances where it’s very much correlated, due to population subgroups which differ by class or other relevant factors which are harder to identify than skin color. So the dynamic is still present.
Yes–but not in the direction that stereotypes generally suggest, or even that trained observers in their study predicted. The point here is that when there are correlations, anecdotal observation is hugely unlikely to identify those correlations accurately.
Well, no. All the data gathered was through observation. The factors they discovered that were strong indicators were necessarily factors that could be observed.
Take, for example, the disparity between working class dress and middle-class dress. They found that folks dressed in middle-class clothes were less likely to shoplift. And it’s likely that in the OP’s scenario, white people are likelier to wear middle-class clothes. It might be tempting to use that correlation to justify laxer observation of white people.
That would be a mistake, even if it weren’t probably illegal. Since the clothing is the better indicator, use that as the indicator. Don’t find a less-accurate correlative indicator to use instead.
And again, the study found that far and away the best observable indicator of shoplifting was behavior: scanning the store, sampling products, messing with packaging, staying a short while and leaving without making a purchase. These are the factors that should be used to identify shoplifters.
Why? In an earlier post, I suggested four problems with racial profiling. Using behavioral profiling avoids all four problems.
They need to compare it to what people with actual experience running the specific stores have experienced, and not to stereotypes or “trained observers” (whatever that means in this context).
I would think most stereotypers do include clothing, as above. But clothing is very imprecise even if it’s more highly correlated, since it’s much more frequently indeterminate.
These factors are much more useful for researchers sitting back and studying videotape after the fact than to a shopkeeper who needs to decide what to do as soon as someone enters the store.
More seriously - we, as a society, have decided that some forms of discrimination (hebeism? youthism) are OK, some (racism) are not. Kids can’t drive, drink, fuck. They are most definitely not, in that sense, a protected class. Race is.
This is all fine and well in the abstract. But what if the ACTUAL numbers at the store in questions are what was presented. How would you answer the OP for that on particular store, which we have stipulated, suffers from shoplifting in the proportion presented?
That’s the thing – any ACTUAL numbers at any store will only be what the store sees. They won’t catch all shoplifters – they probably won’t even catch most. Unless they have a superhuman inventory tracking system and inventory themselves daily, much and probably most of their shoplifting will be unnoticed for far too long to have any idea who actually did it.
How are you determining the ACTUAL numbers at the store in question? If it’s through anecdotal observation, this study pretty persuasively indicates that’s not an accurate means for determining the ACTUAL numbers.
Same question: the people “with actual experience running the specific stores” are humans subject to racist assumptions and confirmation bias, just like everyone else. The beauty of science is that experiments can be devised to eliminate such variables. The people with actual experience running stores are unlikely to have an accurate sense of the actual demographics of actual (not just caught) shoplifters, unless they have been engaging in very rigorous data-keeping.
Or, you know, an average, decent person who objects to being drafted into enabling racial profiling. Which may be synonymous with “SJW” in certain circles, I suppose.
Since my opinion of people who use “SJW” as a pejorative really doesn’t belong in GD at all, I’ll leave it at that.
That’s life. You can’t always be absolutely certain that your experience is completely accurate and doesn’t reflect, to one extent or another, some hidden bias or cultural indoctrination on your part.
You do your best. It’s not reasonable to expect people to toss out everything they’ve learned in life on the off-chance that some study will prove it to have a faulty basis.
And it’s especially so in this case, where - as indicated earlier - the correlation will be highly specific to the store in question, and it’s extremely unlikely that any valid study exists which would be better than actual experince relating to the store in question.
You can, however, choose to recognize that sometimes science shows your raw perceptions are faulty. The stars at night aren’t actually tiny; wearing my lucky socks won’t help my baseball team win; vaccines don’t cause autism; my perceptions about who is shoplifting don’t match who’s actually shoplifting.
In this case, again, the evidence is that perceptions do not match reality. We can always wave our hands at scientific research and choose to reject it. But if you really want to do your best, you pay attention to the research.
Which, again, is what professionals in the field are doing. I earlier linked to a long review of shoplifting, written for retail managers; their advice dovetails with what the research shows. Folks with genuine expertise in the field, not just anecdotes based on racist preconceptions, agree.
As general knowledge it’s useful to keep in mind. But anyone who dismisses everything they’ve learned from experience because science sometimes shows that experience to be wrong is making a mistake.
If someone researched your particular store, this makes a lot of sense. Otherwise, a lot less.
(Useful to know anyway, as it may shed some light on it even at your store, but you can’t over-ride all your specific experience based on a study of some other situation.)
First, to fight the hypothetical: with perfect knowledge, I have no reason to stereotype, I just ban the actual shoplifters.
Second, to fight it a little more: you really mean slightly less than perfect knowledge, i.e., perfect knowledge of the race of shoplifters but nothing else. In that case, I wonder what kind of creepy racist god insists on giving me that level of knowledge.
Third, to stop fighting the hypothetical: remember before when I listed four problems with racial stereotyping? Given your hypothetical, one of those problems no longer exists. The other three remain (going forward you’ll still suffer from confirmation bias, unless Creepy Racist God continues to feed you numbers; you’ll be a pain in the ass to innocent black people; and it’s illegal).
As for the continued rejection of science in favor of anecdote, at this point I become more concerned with a follow-up question: how do we prevent shopkeepers from breaking the law in this manner? It’s pretty clear that our measures in place currently are ineffective; fuzzy_wuzzy has even offered advice in this thread how to circumvent these laws. What do we need to do to enforce anti-discrimination laws?
What I’d like is to put some of the people in this thread in that store as an awned, where their livelihood depends just on how well the store does. Something tells me their scrutiny wold stop being so idealistically egalitarian.