The Gang Test

Social psychologist Malcolm Klein devised a test for Los Angeles that he says predicts how likely a child is to join a gang, reports the Wall Street Journal. The test, which can be found here in its entirety, asks kids questions like whether they have just broken up with a boyfriend or girlfriend and how many of their friends have used marijuana. The problem: the city won’t know for several years if the predictions are accurate. [%comments]


I suspect a high correlation between children who answer yes to questions 68-70 and gang activity. Can I be a psychologist now?


Methodological queries for the person who devised the test:

1) Is it fair to assume that kids won't be in gangs in the 4th grade? i.e., why is 5th grade the cutoff?

2) The "Race/Ethnicity" lines do not allow for gradients between "White" and "Black". If Prof. Roland Fryer's research is to be taken seriously (and I have excellent reasons why it should not), then the test is not picking up the "plighted mixed-race" children (namely, those that are part "Black" and "White")

3) Regarding the "Thinking back" question: Introspection is notoriously unreliable. Think of position effects in Psychology. If introspection is unreliable for adults, then wouldn't it be worse for children?

4) If question #53 was excluded, would that have a major effect on the predictive power of the test?

5) The same with question #70 - this seems like it is stacking the deck. This test is supposed to PREDICT how likely a child will be to join a gang; if they're already in a gang, the predictive power of this test is artificially increased. Major problem here.

6) What counts as evidence (for this test) that the subject is not honest?

I'm pretty skeptical that this test will have much fruitfulness, even with the questions in the survey that stack the deck for it's alleged predictive power.



75. Do you typically lie when you're given stupid questionnaires? Yes | No

Jeff S

Late. I've already given this to the people in the cubicles and 3 of them have already started throwing gang signs...

David Chowes, New York City

The "gang test" does not presently evidence usefulness. Though as it is being investigated, it may well.

Yes, it is being used -- but utilized to determine if it has reliability and validity. Reliability consists of the instrument giving consistent responses each time it is administered within the group which it is presently being administered to.

Validity means that the test is actually predicting the outcome -- in this case, gang membership in the future,

For a measurement devise to have validity it must have reliability; however, a test can have reliability but lack validity.


Things like "Have you recently broken up with a girlfriend or boyfriend" are temporary, though, even if they are repeated experiences. So a kid may score high one week, and low the next (depending on how one defines "recently"). This is likely to introduce a bit of noise...

Plus...virtually all kids break up with a boyfriend or girlfriend in school; the ones that don't often don't because their parents are putting substantial pressure on them not to date - i.e., there is home involvement. If they're looking for kids who never have relationships with this index, they are probably looking for those families - so in essence, family involvemetn may be a better predictor of non-gang involvement. No?

Have taxpayers in any way coughed up the money for this study?
Will the results be stored under the counter unused.
Be used but counter-productive?



Analysing this in conjunction with your post on 22nd about how telling college kids the loved asparagus as a kid will make them eat more, one would think that if a kid is told he is destined to be a gangster he will work towards it. I'm not very sure of this though, I was told I would be an engineer but I'm in law school.

John Morrison

"Social psychologist Malcolm Klein devised a test for Los Angeles that he says predicts how likely a child is to join a gang[.]"

Presumably, this psychologist has done the clinical research necessary to establish the prediction. It would have to involve thousands -- perhaps tens or even hundreds of thousands -- of children, and years of observation. But then, the necessary observation itself might alter the results, since people behave differently when under observation than not.


I think the commentors are too harsh on this test.

Even if it was based on common sense and experience rather than experimental results it could have some value, as it is distilling the experience of the test creator and providing his knowledge in a very easy to reuse way.

Predicting gang-membership might not be a very difficult task. This tool could provide teachers with a structured way to review the problem before it occurs and help children in need.

I would like my child to be subjected to a yearly test by the school to detect potential problems and help the teachers focus on certain areas where he might need help.

The quality of the test is not very important, what is important is creating a ritual where the teacher allocates the time to think about these non-academic issues and communicate with the parents if necessary.


#10 -

The quality of the test is very important, since it's trying to predict who will be in a gang and who will not.

You can't detect potential problems accurately with this test. Saying you're going to predict with a test when asking in the actual exam IF they are actually IN a gang already rigs the test, and predictive power.

You can create a ritual without having to take a rigged exam. That's what good parenting is all about.

Have taxpayers in any way coughed up the money for this study?
Will the results be stored under the counter unused.
Be used but counter-productive?