“Best And Worst Teachers Can Be ID’d Early”
Posted: March 5th, 2013 | Author: Michael Goldstein | | 5 Comments »
Sarah Sparks at Edweek writes up an interesting new study.
New teachers become much more effective with a few years of classroom experience, but a working paper by a team of researchers suggests the most—and least—effective elementary teachers show their colors at the very start of their careers.
Hmm. Interesting how top quintile teachers maxed out in Year 2. Thoughts?
I can imagine a version where Year 2 teachers keep burning midnight oil at the same max capacity as Year 1, with big increases in efficiency over Year 1. So spike.
Then in Years 3 and beyond, they shift some hours towards personal sustainability. You know, “extras” like sleeping 8 hours a night, eating food while not grading papers, hitting the gym once in a while, calling Mom on a regular basis. Maybe even dating.
I think the most pertinent point is that 95% of the original sample were out of the experimental group by year 5. This might in particular explain the year 2 spike somehow.
Hey, Mike,
Actually our experience with VAM has been quite different. Our grads continue to become more effective in raising MCAS scores into years 4-5. Granted it’s a really small sample. On the other hand the typical pattern in BPS does sort of mirror what the CALDER folks found. Novice teachers plateau fairly early.
But I think we have to remember that these patterns reflect the status quo in terms of the quality of teacher support, school organization, and formal/informal induction. Given the current ways in which we support (or fail to support) teachers, it’s not particularly surprising that there is a limit to how much teachers improve via individual trial and error. But this says nothing about how much they could grow with high quality coaching.
By “we” I mean the field, not our specific program.
Yes, Ed.
But high quality coaching, like high quality pre-school, and high quality charters, has proven rare and hard to replicate.
Interesting study. I was perplexed by this graph at first as well. It looks like the spike in the Year 2 (and drop in Year 3) is just a statistical issue of regression to the mean. If you click through to the original article, it turns out that the teachers were grouped into these quintiles based on their performance in the first two years. The results of some previously top-quintile teachers will drop (regressing to the mean) in Year 3 and beyond. These results are getting averaged in to the “Quintile 5″ group on the graph, pulling down the average of that group. In the same way, some of the previously bottom-quintile teachers will get better results in Year 3 and beyond, and are again averaged in to the “Quintile 1″ group, which markedly raises the average in Year 3.
So it makes sense that there is a shift in how the average statistics look between Year 2 and Year 3. That spike is just a result of how the teachers were locked in to quintile groups from their first two years of results. I think the point of the graph is to show that the group of teachers that were in Quintile 5 based on their first 2 years are *on average* still better than the teachers who started out in Quintile 4. It is not suggesting that an individual teacher will spike in Year 2 and then return to his/her Year 1 performance.
There is also an interesting graph in the original article that models future results of the bottom x% of teachers. For example, if a school adopts a policy of terminating the new teachers in the bottom 10%, how many would have gone on to get results in the top 10% or 20%? I found the results to be pretty suprising. Since value-added model results are so random from year to year, it is very hard to make these predictions for individual teachers.