Hacker News Reader

Comment by ljf | Hacker News Reader

On a similar note, I can't find the study now but I've seen the pretty much all social stratas and races commit crimes at the same rate. It's just that 'we' care about policing and charging 'others' more than ourselves.

As an example, heading East out of London the motorway is often heavy with speed cameras and often police cars (poorer end of London) . Heading West into the Cotswolds (the rich area) , there are few cameras and a high level of very fast speeding. But if you looked at a study it would appear that those from the 'poorer' areas committed more 'speed crime' as those are the only areas we are effectively policing.

I wonder if those looking for cheating also start from the position that foreign pupils are more likely to cheat and therefore spend more time looking for cheating?

View on HN

This makes no sense. At a given university, foreign and domestic students go to the same university and take the same classes. At best, you might argue bias with foreign-sounding names, but that would be an extraordinary claim to make and would require extraordinary evidence to explain away a 5x difference in cheating.

Extraordinary?

http://blogs.wsj.com/moneybeat/2015/02/16/study-finds-invest...

http://www.nber.org/digest/sep03/w9873.html

http://www.theguardian.com/money/2009/oct/18/racism-discrimi...

And again that is 5x more discovered cheating, what number of cheating occurred undiscovered for all groups? I'm not saying people don't cheat, because they do, I'm saying look at the environment to understand what other impacts on the number might occur.

One will note that you didn't respond to my first point, which is that your actual argument (about unequal resources devoted to policing cheating) is wrong.

So if you are going to argue the point that there is bias against foreign-sounding names, enough bias to explain a 5x gap, you'll need more than roundabout inference and implication. Extraordinary evidence means direct evidence, not links to employment-related articles.

Or it's ease of detection. Viz repetition between students in written assignments of characteristic ESL mistakes that are glaring to native speakers.

Where I went to school, for each CS course (eg algos 201, 301, etc) all programming homework was submitted with a department run submission utility. After submission, all programs for all time were compared for cheating. My guess is they did some basic comparisons of the AST.

The CS department produced well over 10x the cases of academic fraud as the rest of the college. We got lectured each and every semester about it. There were even locked disposal boxes in the CS labs to dispose of printouts; I got busted for cheating one semester because someone had stolen a printout I'd used to figure out bugs in an algorithm and copied the code! Fortunately, the other student had been caught cheating repeatedly and his programs regularly failed to work while I'd never come close and my program worked. The professor gave me the benefit of the doubt.

However, do I really believe CS produced 10x the cheaters? Of course not. They were just easier to detect.

That said, I was stunned at the amount of cheating in graduate courses. One professor teaching math stats was in the habit of posting distributions for each question on the tests. For a particular question, the distribution was bimodal -- I was one of two people in the middle, and the rest of the ~30 students got nearly 0 points or full credit. The question rested on a really neat integral trick, but if you didn't see it (on a timed test), or used the obvious substitution, it was about two pages of work. If you saw the trick, the problem fell apart.

I had been helping one of the students who got full credit for that question with her homework and her calculus was not her strong point. There were ex-physics students in the class who had won integration contests and I kind of expected them to get a trick like this, but not her. So I privately asked her how the hell she got that question. It turned out she had already seen the test -- the 15-ish Chinese kids in my class passed tests down between years, and the professor just twiddled numbers in them instead of rewriting tests. My solution, since I was graded on a curve with them, was to make sure I got to see the tests too. I'm not really sure if I'm proud of that, but you can't be graded on a curve when half the class sees the tests ahead of time.

>On a similar note, I can't find the study now but I've seen the pretty much all social stratas and races commit crimes at the same rate.

That sounds completely implausible. For instance:

For example:

https://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2013/...

Again, as the comment above, I said crime, the example I gave was speeding.

Speeding isn't usually considered a crime. It won't usually show up on background checks. Usually you can't go to jail for it. It's usually just a fine. When it isn't, you're talking about moving from simple speeding to reckless endangerment and the like.

Speeding (a) isn't a crime, and (b) is committed by just about everybody, making it a pretty bad example.

You can't find the study because it doesn't exist. In the United States and in the UK there is a huge disparity in violent crimes committed by race.

Did I say violent crime? I said crime.

http://m.huffpost.com/us/entry/8078586.html - 1) If a black person and a white person each commit a crime, the black person is more likely to be arrested. This is due in part to the fact that black people are more heavily policed.

http://blogs.channel4.com/factcheck/factcheck-black-american...

This does not prove that true crime rates are equal. That is an extremely strong claim to make. I agree the point about being more likely to be arrested, but that hardly leads us to believe therefore that the true rates are equal.

> all social stratas and races commit crimes at the same rate

Some communities have bars on their windows and others leave their doors unlocked.

Reply on news.ycombinator.com