Answer: Excellent college performance is only weakly associated with professional performance.
Malcolm Gladwell wrote about the similarities between NFL Quarterbacks and grade school teachers. The NFL game is so vastly different, and the environment is so much more dynamic that many, even most, great college quarterbacks cannot make the transition. In a similar way, the teaching classroom is a dynamic, shifting kaleidoscope that is vastly more complicated than the sequential, ordered instruction/examination environment of the college classroom.
It is a bit like swimming. You can instruct people on the skills of swimming using Powerpoint presentations, videos and breathing drills....but the only way you can find out if they can swim is to throw them into the water.
It is widely recognized that the upper 90% of the teacher population is effective. They deliver the required materials, monitor the progress, maintain classroom order, comply with the applicable laws and keep adequate records.
The bottom 10% is a disaster. Not only do they sometimes fail to teach. Sometimes they teach material that is completely wrong. They complicate the lives of the upper 90%.
I am not banging teachers. I am not saying that teachers are worse than other professions. I am saying that education draws from the same population that every other profession draws from and there will be some mismatches.
One reason that teachers are attracted to the profession of teaching is because of the autonomy and the status. Recent efforts to identify and re-career the bottom 10% have the upper 90% in a dither. They feel their autonomy and status ("I am a professional, I don't need anybody looking over my shoulder!") is threatened. The plain reality is that every professional, except Congressmen, is now scrutinized. Ask your doctor.
Another concern teachers have is "How can it be objective? How can I be sure that I will not be punished when I am given The-Class-From-Hell?"
The short answer is to consult with mathematicians and with people who use numbers to control other complex processes.
One common strategy to eliminate "noise" like The-Class-From-Hell (TCFH) is use a filter. A simple one is to look at runs of sub-standard performance. A "run" is a number of batches in-a-row.
A crude way to filter out the bottom 10% is to take one year's learning growth, rank from top-to-bottom and then re-career the bottom 10%. Done deal. Unfortunately, you end up releasing some really competent teachers who were given a batch with very challenging kids. The teachers I know tell me that the toughest kids routinely get routed to the best teachers. So those teachers would be double punished. First with the extra work, then with the additional risk of job loss.
There would also be the temptation to disown those kids. Those kids are a threat to your livelihood, so there would be a temptation to run those kids down the discipline gang-plank and out of the school.
A better way to skim the sediment would be to tag every teacher in the bottom 25%, put them on an improvement plan and then release them if they were in the bottom 25% the following year. The odds of randomly falling into the bottom 25% two years in a row is approximately 6%. Please be mindful that language like "randomly falling into" implies that teacher ability has no effect on outcomes.
If teacher ability has a strong effect on learning outcomes, then the chance of an improper "cull" goes way down.
Still too risky? Pick a number. Remember that this value is a prediction of the upper bound of improper "cull" picks and it assumes that teacher ability has no effect on outcomes. 1%? Microsoft Excel informs me that two years in-a-row at sub-10% performance (relative to peers) or three years in-a-row at sub-21% performance will hit that number. Leaving that system in place for 11 years will find, and re-career that bottom 10%.
And tweaks can be put in place to blank out first-year teachers and to find appropriate peer groups for specialty subjects. It is not likely that a remedial math teacher wants to lumped into the same pool as the Advanced and Gifted Calculus teacher. Also, a bigger in-a-row number (accompanied by an appropriate performance threshold) can be selected for upper grades that are on a semester or trimester system because more batches will be processed through in a given school year.
Remember that it is kinder to be re-careered earlier rather than later. It is kinder to the kids. It is kinder to the person being re-careered. It is a balancing act to design a system that is both insensitive to noise yet sufficiently responsive to either re-career or re-motivated the bottom 10% in a timely way.
Finally, I think it really does matter. In our small school district of 3000 kids, a student will likely have seven homeroom teachers and 6 "specials" teachers by the time they hit 6th grade. Then the number of teachers explodes. They will likely encounter 15 more teachers by 9th grade and another 20 by graduation. By my math, that adds up to 48 teachers. Four or five of those teachers are likely to be in the bottom 10%.
And the sooner the child encounters a teacher in that bottom 10%, the more damaged they are likely to be for future learning.