Selecting for the wrong variable

There is a famous anecdote about the search for genes that govern behavior. A scientist wanted to try to breed a type of smart drosophila melanogaster, a common fly used in genetics experiments. He designed a 3D version of the mazes used to test intelligence in rats and mice. He placed one hundred flies in the maze and bred whichever ones came out first, assuming they were the smartest. He did the same with their offspring, finding that with each generation the speed of the fastest flies were increasing. Surely he was on the right track towards developing really smart maze negotiating flies. Finally after several generations he saw a drop-off in the improvement from the last generation and assumed the flies had become as smart as they possibly could be. He took one of these super-smart flies and put it in the maze alone for the first time. Ten, fifteen, twenty minutes passed, but the fly didn't come out. He sent another in after him and the first one quickly flew out.
A little more experimenting showed that in putting so many flies in the maze and breeding the ones that escaped the fastest, he hadn't been selecting for maze-running behavior. Rather, he had been breeding the flies that most wanted to avoid the other flies. He had essentially bred them for antisocial behavior. So, in the pursuit of smart flies, he wound up with crowd-hating flies.

The Korean school year starts in March, and my wife and I were saddled with the task of dividing our two kindergarten classes into two new first grade classes. Although there had been a fairly broad difference in abilities among the children, they were assigned to classes based on their schedules rather than their levels. Of 30 kindergarten students, 16 remained in our first grade classes. We gave the children level tests and combined that information with our own judgment to make the lineups for each class. A1 class would be the children who could read and understood phonics, A2 would be the children who need to focus more and catch up. Both classes have now been running for 6 weeks, and yesterday I came to the discovery that, like the apocryphal scientist in the above story, we have selected for the wrong variable.
We thought we were selecting the students for the speed at which they learn. What I have instead come to realize is that, almost to a man, we have selected students for the presence of a 'good parent'. I define a good parent as someone who spends time with them daily making sure they do homework and maintain an active interest in their child's achievement.
Here's the breakdown:

A1

Simon: Mother frequently consults with my wife about his progress.
Steve: Father studies with him every day.
Penny: Does homework with her mother.
Judy: Mother is an amateur English teacher, they do homework together and she speaks with me regularly.
Manny: Parents have an interest in his education. Does well although not naturally gifted.
Craig: Although he is distracted often, his mother cares a lot about his education.
Lisa: Mother, although busy, checks her homework every day.
Helen: Mother is interested in her progress.

A2

Mark: Started much later than the other children. Parents are fine but don't offer any extra help, which is clearly needed to play catch-up.
Paul: Lisa's twin brother. Despite mother's great interest is not naturally gifted.
Sam: Mother works full time, father recently passed away. Frequently doesn't do homework.
Jessie: Mother is combative, negative and constantly asks me why her daughter is not smart right in front of her.
Sarah: Mother is petty, ditzy and generally unliked. Daughter shares negative personality traits despite sweet general demeanor.
Jason: Very smart boy, but he and his older brother are latchkey kids and their folks are gone all day.
Peter: Not naturally gifted. Parents are aware but unwilling to give him the extra help needed.
June: Good parents and good study habits. Narrowly missed the cut-off point for A1 class, a decision I regret. We are considering moving him up.

So it's clear that, although natural gifts have played a role in our selection, the major factor seems to be the attitude or ability to help of the children's parents. A boy like Jason in A2 would be at the top of A1 with a little help from his mom or dad, while kids like Manny or Craig in A1 would sink to the bottom without good parents. Of course the monkey wrench is Lisa and Paul, fraternal twins with very different abilities, and the saddest case is Sam, who both is not a strong student and only has one very busy parent to look after him. This realization has been very helpful in dealing with A2 class, in that I now know that I can't count on every note that gets sent home to be read and I can't tell the kids to ask their parents for help, because they'll likely get none. Unfortunately the separating of the children has led to vast differences in the pace of study. A1 class is progressing in their new phonics book twice as fast as A2, and there no end to the trend in sight