Rates and risk for trafficking
August 31, 2006 12:54 AM   Subscribe

I'm working on a grant, and one of the only 'hard' numbers we have is from a qualitative study in the community we work in, which estimated that 30-40% of under-16s are trafficked at some point. I'm trying to take that likely-conservative 30-40% and figure it out as a per-year risk.

We deliberately select the high-risk among that community, so the kids in our project are even more likely to be trafficked. We have a 60-kid capacity and have tracked 50+ kids for coming up to a year now. In that time, we've seen at least one of them trafficked despite intervention.

Do I just divide 35% by 16 years for an average 2.18% risk a year for each child? How many children based on that stat could I say were likely to be trafficked in a year without intervention?

There just isn't any more hard data - even the number of children in the community is in flux. This is an estimate based on an early 2006 research project that calculated based on the number of neighbours' children reported trafficked in the same community in Phnom Penh, Cambodia.
posted by viggorlijah to Science & Nature (7 answers total)
Could you define "trafficked"? I assume it has something to do with changing homes, but the first things that came to my mind were sex slave trafficking and drug trafficking.
posted by chrisamiller at 1:07 AM on August 31, 2006

Sorry, trafficked as in children who are sold/bought (as opposed to legal change of custody as in an orphanage relinquishment or legal adoption) for sex work, domestic labour, begging rings or illegal adoption, the major demands.

Trafficking as in a parent or guardian sells outright or hires out their child (under-16) for an illegal or unethical job.

We have one parent who has forced her son to drop out and do shoe-shining and odd jobs to make money, but we don't consider that trafficking - she's still his guardian and it's not by local standards cruel work. There's another family where the girl is likely being hired for sex work once a month, which is trafficking, and another where a young girl was sold to a neighbour as a housemaid/nanny and treated very badly - that's trafficking as opposed to child labour. These are three greyish area examples I know of first-hand. Most are just outright trafficking - a newborn sold for illegal adoption, young teens sold to a brothel.

It's for this - RiverKids.
posted by viggorlijah at 1:27 AM on August 31, 2006

When a child "becomes trafficked", do they remain "trafficked" for the purpose of your study? And do they drop out of your base group or not?

So, if I have 50 children one year, and one is trafficked:
- Do I now have 49 children?
- Next year, does the trafficked child reset to not-trafficked? Or
- Do I get more 1 more untrafficked child to make up for the vanished trafficked children? Or
- Do I now have 1 trafficked and 49 untrafficked children? If so, can a trafficked child be trafficked again? If so, (and this is quite key politically), can we assume the odds of being trafficked, having been trafficked, are the same? (Probably not, but that position is important)

No moral or operational comment, by the way, just trying to figure the stats.
posted by alasdair at 1:55 AM on August 31, 2006

How many children based on that stat could I say were likely to be trafficked in a year without intervention?

Number of children X annual risk of being trafficked. For example, 50 children, 2% chance, 50 X 0.02 = 1 child. But you're getting ahead of yourself: you have to work out if that annual risk % is right first - see questions above.
posted by alasdair at 2:04 AM on August 31, 2006

By your data, there is a 60-70% chance for a child to go 16 years without being trafficked. Suppose that we knew the average chance of not being trafficked (call it q). In order for a child to go all 16 years without, the chance is q^16. So:
                  q^16 = 60-70%
    ln(q^16) = ln(.6)       ln(q^16) = ln(.6)
    16*ln(q) = ln(.6)       16*ln(q) = ln(.6)
    ln(q) = ln(.6)/16       ln(q) = ln(.6)/16
    q = e^(ln(.6)/16)       q = e^(ln(.6)/16)
    q ~= .969               q ~= .978
But, of course, you are looking for the opposite probability:
    p = 1 - q = 2-3%
I don't belive that your quoted data is precise enough for anything more than a single digit precision here and am way too lazy to do that math.

FINE PRINT: This makes some simplifying assumptions, notably that the chance of a child being trafficked any given year is independent of other years and that the chance of being trafficked is constant across the range of ages. These assumptions are almost certainly wrong as per alasdair's questions above. The calculation, however, is probably (p > .5) close enough.
posted by Mr Stickfigure at 5:29 AM on August 31, 2006

Alasdair -

When a child "becomes trafficked", do they remain "trafficked" for the purpose of your study? And do they drop out of your base group or not?

They remain trafficked. Even if they were rescued, they would have to go to a specialist recovery NGO that specialises in reintegration and so on.

We've just marked four kids as inactive after four months absence so that their slots can go to another four on the waiting list. One of them has definitely been trafficked, the other three dropped out for unknown reasons.

Multiple trafficking happens for sex work. A high proportion of teenage sex workers after they're 'rescued', return to the trade - family pressure, few alternatives, addiction, shame etc. But I don't think it's relevant for us given that we work with kids only before.

There's just so little known about trafficking. Children under 2 are big for illegal adoptions, girls just at puberty are big for sex work. But how much it varies year by year is unknown.

This is hugely helpful so far. Thanks.
posted by viggorlijah at 7:02 AM on August 31, 2006

Mr Stickfigure has it, then, I believe (and his or her maths is better than mine!) and I'll second his one-digit range. If you want to look more convincing stick some extra digits in (e.g. 2.26%): this is dishonest, but will impress non-mathematical people with your great knowledge.

Now, you're going to want to compare the rate without intervention you've obtained here with the rate for your organisation, and draw some conclusions. This means comparing the expected rate - in this case about 2-3% - with the observed one - which is your fifty-plus kids with three-ish being trafficked, or 3 out of 50, or 3/50, or 6%. So kids are much better off without your intervention, according to the statistics!

Well, no. Your expected rate (2-3%) is a real finger-in-the-air job, and you haven't been going very long, and your identification of kids who have been trafficked is clearly flexible (as one would expect in such a setting). So you could muddle around the figures (assume 40% are trafficked, or claim 50% for your target group as opposed to this study, then reduce your count of trafficked kids by removing the grey-area ones, or not counting kids you only had a short time and were trafficked...) and make yourselves look better. But really, you don't have the data to draw any strong statistical conclusions, in my opinion.

I would try "Source X says 30-40% get trafficked, over their lives, which means about 2-3% for the population of that study. However, we're looking at a higher-risk group, so you would expect the trafficking rate to be higher. We've looked after 50+ kids, and here are the trafficking stories (detail horror stories), and that's more than 2-3% - see, we're dealing with a much more problematic group. Here is where we've made a difference (detail successes). We're clearly doing great work."
posted by alasdair at 12:10 PM on August 31, 2006

« Older Help me be an active robot!   |   Current TFTs and games. Newer »
This thread is closed to new comments.