[kansface]

I read a study a while ago that addressed this exact problem - a CS professor gave an intro class a test on CS. The students who had a consistent view of how a computer works (no matter if it was wrong or not) turned into programmers. Everyone else failed. There was essentially no movement between the groups.

[necubi]

That study was retracted by the author, who now deeply regrets it. See here: http://retractionwatch.com/2014/07/18/the-camel-doesnt-have-....

[pcthrowaway]

Here's Jeff Atwood on this: http://blog.codinghorror.com/separating-programming-sheep-fr...

My problem with this idea is that if you have no idea if the thing on the left (when assigning) takes on the value of the thing on the right, or vice versa, it seems reasonable to answer different questions with different assumptions to cut your losses.

[cmadan]

Interesting. Link to the study please?

[kansface]

I just found it: http://www.eis.mdx.ac.uk/research/PhDArea/saeed/paper1.pdf

[smt88]

This study is biased toward students who already knew CS before taking the class, which is a substantial number. If someone can explain how this methodology makes sense, I'll start to buy the argument.

People used to say women were less likely to have a "natural ability" at CS, despite the huge amounts of seminal CS work that women did in the 50s, 60s, and 70s.

That changed when the Apple II came out, and it was marketed at boys. Boys practiced CS before college and basically drowned out the true newbies in their 101 classes.

After schools realized this, they offered pre-101 catchup classes, and that has totally eliminated the gap that we saw before.

I've still never seen anything to indicate that aptitude to program (as a job, not as a researcher or theorist) exists.

[DanBC]

The study was retracted by the author. He has said that he was wrong and that the paper was wrong.