Tuesday, December 07, 2010

Focusing on teachers will do virtually nothing to improve education

Imagine that all the supposedly poorly performing school teachers were magically replaced with teachers whose skills matched those of their most competent peers. What would that mean to educational attainment? The best research available says that at most variability in teachers' skills accounts for only about 15 percent of achievement outcomes. Given the focus on school teachers as the bane of educational outcomes that is a surprisingly small number. Sixty five percent of achievement outcomes is explained by student and family background characteristics:

Data for the above chart is from Teacher effects on student achievement, Rowan et. al., 2002.  That research found that:
"In review of this literature, Scheerens and Bosker found that when student achievement was measured at a single point in time... about 15-20% of the variance in student achievement lies among schools, another 15-20% lies among classrooms within schools, and the remaining 60-70% of variance lies among students."
Other studies put the effect of teachers much lower, even under 10 percent. Even if you take the highest number for classroom effect - 20 percent - and assign the entire classroom effect to its teacher,  then assume a high number of teachers are failures, i.e. 15 percent, that still renders bad teachers only three percent of the overall effect on student achievement (15 percent of 20 percent).  The pie graph above shows the effects of students (60-70 percent), schools (15-20 percent), classrooms - good teachers (85 percent of 20 percent), and "bad" teachers (15 percent of 20 percent) on achievement.

If so-called "bad teachers" represent only three percent of achievement impact, and students themselves represent 65 percent, why does the education deform movement focus on the three percent and not the 65 percent? Because it's easier and politically efficient. Focusing on teachers to improve educational outcomes is like trying to squeeze blood from a rock. Conversely, focusing on the socio-economic status of students and their families represents fertile ground for improvement seeing as how it represents such a large chunk of responsibility for achievement.

*Hanushek et al. 1998Rockoff 2003Goldhaber et al. 1999Rowan et al. 2002Nye et al. 2004

29 comments:

blogspotdog said...

"[F]ocusing on the socio-economic status of students and their families" requires introspection that our society and political system are unlikely to make any time soon.

Tom said...

Why does the best research available appear to be about ten years old?

Seriously, we're going to blame the students for their poor performance?  Not in today's society.  I'm pretty much with Spot on this one.

Adam Smith said...

This is a bizarre analysis, the bizarreness of which should not obscure a point that Mr. Levine usually attempts to hide but has finally put front and center.

The point he normally hides is that teachers make a huge difference. Let’s look at the pie chart The three biggest factors are schools, “student and family background,” and teachers.<span>  </span>Which of these factors can a school board control? Which can a principal affect?<span>  </span>The fact is that policy makers can only address controllable factors.<span>  </span>More than just politics is the art of the possible.<span>  </span>Principals can in principle attempt to control the family and student factors – that’s what they do when they discourage poor performers from showing up on testing day. That’s what they do when they discourage poor performers from sticking around schools.<span>  </span>But that is not honest improvement.<span>  </span>What school boards can do is to improve schools and teachers.<span>  </span>Levine’s pie chart shows that schools and teachers are the two biggest controllable variables.<span>  </span>They are the most important levers at hand.

To put it very simply, which of these two policy prescriptions makes the most sense:

1. Improve education by increasing teacher skills, or
2. Improve education by making sure that more students come from higher socioeconomic status families?

The bizarreness of the analysis is the 15% of 20% computation.<span>  </span>Regression does not work like that.<span>  </span>If variability in teachers' skills accounts for about 15 percent of achievement outcomes, then that’s what it is. One can’t cherry-pick that number and subdivide it out. (The cherry-picking is Levine’s assumption that “assume a high number of teachers are failures, i.e. 15 percent”.)<span>  </span>In regression, if a factor is worth 15%, then it’s worth 15%.<span>  </span><span> </span>And, following on the theme the first paragraph, teacher variability is about 50% of the controllable variables.

Rob Levine said...

Nobody's "blaming" students. This is research, not opinion. The point is that the attributes of the student and the family control most of the educational outcomes, not the school or teacher.

Rob Levine said...

<span>Adam says:   
 
"<span>To put it very simply, which of these two policy prescriptions makes the most sense:    
   
1. Improve education by increasing teacher skills, or    
2. Improve education by making sure that more students come from higher socioeconomic status families?  "</span>  
 
You just don't like the best answer to your own question because it costs too much and involves addressing issues like racism and poverty.  
 
Doing #1 doesn't accomplish much because a) Teachers are all pretty great to begin with, and b) Teachers are only responsible for about 15 percent of educational outcomes.  
 
I didn't even mention how the deformers' plans involve putting more POORLY trained teachers into schools with Teach For A while teachers.</span>

Alec Timmerman said...

Maybe you make some fair points Adam, but from every comment you've made on this site you have never, ever mentioned ideas to improve teachers or support good teaching. You've mentioned punishing "bad" teachers. You've mentioned getting rid of bad teachers. You've mentioned making it easier to become a teacher. You have never, ever mentioned how you might support better teaching. You are completely problem focuses and not solution focused. Now you want us to believe you care about increasing teacher's skills?

The day you start supporting god teacing, Adam, is the day more kids can win. I have yet to see a deformer plan that supports good teaching. They are just reconfiguring the century old model of the isolated classroom kingdom that worked for them, but not for most kids.

Rob Levine said...

<span>Adam says: "..following on the theme the first paragraph, teacher variability is about 50% of the controllable variables." No - classroom effect is from 15-20 percent. School is 15-20 percent. Teacher effect is a subset of classroom effect. One study linked to gave the teacher effect as 7.5 percent. If the great majority of teachers are doing well, where is the opportunity for overall achievement by modifying the teacher variable?
</span>

Adam Smith said...

I do oppose putting religion in education, yes.  But where did I propose "punishing 'bad' teachers"?

Adam Smith said...

OK, let me get this right:
1. you said "<span> Teachers are all pretty great to begin with."</span>
2. teachers have a negligible effect on education ("<span> One study linked to gave the teacher effect as 7.5 percent.")</span>

Do you honestly believe those two points?  If you belive No. 1, then I think your beliefe are at odds with those of everyone who ever can read.    If you believe No. 2, then ditto. And if you don't believe No. 2, then there IS value in discussing improving the quality of teaching.  I thought that was one of your goals.

Rob Levine said...

<span>Teachers are pretty great. What evidence do you have that a large percentage are deficient?  And no, I'm not agreeing with the 7.5 percent figure - just saying that might be a lower bound. I'm definitely NOT dismissing the need for high-quality teachers, given that they ARE something that we can have an affect on. In fact, since teachers only control a relatively small portion of outcomes, it is important that they be professionally trained and comptent, which they already are, for the most part. Proposals from the deformers would have a negative effect on teacher quality.</span>

Here it is in a nutshell for you, Adam. If teachers have a relatively small impact, compared to other factors, and most teachers already are very competent, then focusing on them is not fertile ground for improvement.

Adam Smith said...

One of us isn't reading very well.  You wrote "<span>"<span> Teachers are ALL pretty great to begin with."</span> (I capitaliized the word "all" because that's what I want to draw your attention to.)  I asked you if you believe that.  You wrote back saying "Teachers are pretty great."  Now you're shifting your position. Then you ask me a question implying I said that "a large portion are deficient."  Where did I say that?</span>

I trust we agree that half of teachers are below average. I trust we agree that a few in the tail are pretty poor.  I bet we agree that the average could also be raised.  

Rob Levine said...

Average? Suppose 90 percent of a cohort is proficient at something. Half will still be "below average." Does that mean they can't do the job? No. Is there room for improvement? Some - but not much.

Adam Smith said...

I love to raise a challenge I know you can't meet (like asking you to tell me where I said "<span>"a large portion are deficient" or asking you if you flip flopped on whether ALL teachers are pretty good) and then raising some other issue (half are below average) just to see you flat out ignore the challenge you can't meet.</span>

Rob Levine said...

Adam - of course I don't believe ALL teachers are perfect. I made that pretty clear in my original post where I took the opposing position of "what if" 15 percent of all teachers were not proficient. BTW - that was an offhand statement in the comment about ALL teachers; obviously there are some who are not great. In the spirit of your method of argumentation, I presume you'll say my entire argument is total bunk because of that.

Alec Timmerman said...

Here's the chalenge Adam. Any reform that leaves the teacher in an isolated classroom kingdom, competing against other teachers, is no reform at all. It's a roconfiguration of what worked for those who were successful in schools themselves. Performance pay, standardized tests, NCLB. None of those reconfigue our education system.

Research will tell you that a collabortive focus on formative student data will work. Not a competitive focus on summative data. Giving teachers time to work together on student data takes time and money though.Cutting teachers pay, increasing their hours, and hiring untrained(cheap) teachers seems to be the solution du jour. Oh yeah, and "choice" will cure everything. We've had choice for 20 years, and it hasn't made a dent in the achievement gap, but somehow that is the magic bullet.

Alec Timmerman said...

And while we waste hours and hours addressing "bad" teachers, and being accused of defending bad teachers, the real problem and most expensive problem is we lose 50% of our teachers in the first 5 years. If the problem of keeping bad teachers is a bad storm, then losing promising teachers is a typhoon. Our focus should be on training and retaining the best teachers we can. If a teacher is struggling, come up with systems that make them better. I can tell you that with collabortive teams it is impossible to be one of those few teachers who sucks because your colleagues are always checking in with you, seeing how you are doing, sharing how they are doing, comparing student data, comparing teaching techniques, etc. This sharing and help also helps stave off burnout that happens in the isolated classroom kingdom. Support what supports good teaching

Adam Smith said...

Glad you have the answer.  Wish you had short-ciruited the above tussle with a note earlier.  Now let's see some cites.  Mr. Levine provide his (albeit after some prodding), so now it's your turn.

Adam Smith said...

part 1. 


<span>I appreciate that you think collaborative work is the key for you and your school. It may be the ultimate solution.</span>
<span> </span>
<span>But a word or two to temper your enthusiasm. The theme of my comments is “you are sole-sourced to a for-profit business, one that does not seem to provide the focus on collaboration you emphasize.”</span>
<span> </span>
<span>These are not two independent<span>  </span>sites you gave me. Both are by “Learning Tree,” which looks like a for-profit concern<span>  </span>touting the ideas of “[a] leading researcher in education, . . . a speaker, trainer, and author of more than 30 books and 150 articles on topics such as instruction, assessment, writing and implementing standards, cognition, effective leadership, and school intervention.”<span>  </span>http://www.marzanoresearch.com/About/about_dr_marzano.aspx.<span>  </span>And ‘"Dr. Marzano is a true icon in the field of educational research. “<span>http://www.marzanoresearch.com/About/partnership.aspx</span>.<span>  </span>Sounds great.<span>  </span></span>

Alec Timmerman said...

Adam,
   Hope you are checking back. Here is an independent paper on Peer Learning. Peer Learning, Collaboration, and Professional LEarning Communities (PLCs) are all pretty synomynous. The concept is the same. No one can know everything and not everything works with all kids. The best way to learn different strategies is from ech other and multiply your effectiveness.

the following is from the Ntional Bureau of Economic Research.

http://www.nber.org/papers/w15202.pdf



Professional Learning Community is just their trademark name for collaboration. They are synomynous. For successful schools you have to have great teachers and great structure. Marzanno focuses, mostly, on the teaching, and PLC's focuses on the structure. I agree you have to be suspect of for profit folks tooting their own horns. However, every other structural reform is just a reconfiguration of the isolated classroom teacher. PLC's is the only realy reform I have seen.

Alec Timmerman said...

Some more on the same theme:

http://www.nber.org/papers/w15202.pdf

http://www.teacherleaders.org/node/463

Adam Smith said...

Thanks!  That reference requires a subscription, but I get the point.

Phoenix Woman said...

Nailheadtom, that was really slick how you took data that showed that economic standing -- over which students have no control -- was the prime arbiter of student achievement, and twisted it around to create a "Oooh, oooh, you're blaming the kids!" strawman to attack, all the while pretending that strawman is Rob.

Phoenix Woman said...

Not when no one in the corporate media in the key decision-making levels has ever had to worry about choosing between rent and food.

Phoenix Woman said...

Of course to a Con, busting a teachers' union is much more fun than addressing the primary root causes of student underachievement -- namely, being poor.

Teacher-bashing: It's the new black! http://phoenixwoman.wordpress.com/2010/12/09/fill-in-the-blank/

Adam Smith said...

10 years old, way back to 2000?  Students and teachers were pretty much the same then, no?  

Check this out; http://www.washingtonpost.com/wp-dyn/content/article/2010/12/10/AR2010121004276.html?hpid=moreheadlines   Undercuts the title of the post w/o slamming unions.

"<span>The central finding indicates that teachers with 'value-added' ratings are able to replicate that feat in multiple classrooms and in multiple years.</span>
<span>

Other findings suggest that teachers with high 'value-added' ratings are able to help students understand math concepts or demonstrate reading comprehension through writing; that the average student is able to recognize effective teaching; and that multiple sources of data can help teachers improve."   

I love that "the average student is able to recognize effective teaching." To me, that shows that what we all remember - that some of our teachers were great, others were OK, and some were terrible - undercuts the argument that "all/most teachers are pretty good" and there is little to be gained by focussing on teachers as being a variable that can be optimized. (Sorry to sound so mechanistic.) Further, I think it enhances Alec's point (perhaps I wrong that he believes this) that if one finds lots of little hints for each individualized teacher's weaknesses, then each will become more effecitive.  This point matches what we know about ourselves playing tennis or the piano or doing our jog, that a little improvement here and a little there, and sooner or later we are dang decent at the task. That, I think is what Gates' program is looking for - improved teaching techniques; they are not looking for less union actvitiy, barring any union "job security" programs that establish work rules.
</span>

blogspotdog said...

The central organizing principle of buzzards like Adam Smith is the destruction of unions of any kind. That teachers might desire to band together to protect themselves from bug-eyed control freak helicopter parents - does this echo with any familiarity to you, Adam Smith? - is something incomprehensible to AS. This is principally because AS, like Bill Gates, was born missing the gene for empathy. Regrettably, there is no cure for this.

"Concern for students" is an elaborate ruse to cover the bilious, poisonous self-absorption and overwhelming sense of grievance and resentment at the core of the modern conservative.

Rob Levine said...

Read the comment thread to see what complete baloney that "study" is. The first comment:

"Amazing. We've discovered that teachers who raise kids test scores are the most successfull.....at raising kids test scores!! If life we only that simple."

Rob Levine said...

You know what else is interesting  - Adam Smith complains about our demeanor (education deformers) but starts his comments on this thread complaining about how the post itself is a "bizarre analysis." Hardly.

Adam Smith said...

Well, Mr. Levine - take your statistical analysis to a statistician, and see what she calls it.  You can't take Pearson r's or r-squareds and multiply them by a fraction to get their explanatory power.  The r-squared IS the percent of explanation.  

You are right that "bizarre" was a bad word choice, and I apologize. Sometimes my middle initial is "S."