Log in

No account? Create an account

Journal of No. 118

September 19th, 2012

Flaws: Romney / Value Added Education @ 04:39 pm

Awright, Romney's already taken his lumps, but there's always room for more, especially since he seems to be doubling down on his sentiment. For reference:
There are 47% of the people who will vote for the president no matter what. All right, there are 47% who are with him, who are dependent upon government, who believe that they are victims, who believe the government has a responsibility to care for them, who believe that they are entitled to health care, to food, to housing, to you-name-it. That that’s an entitlement. And the government should give it to them. And they will vote for this president no matter what…These are people who pay no income tax…[M]y job is is not to worry about those people.

Now it does seem to be true that the nation is divided roughly 50/50, and each party can count on 47% of the nation to go their way no matter what, and then the fight is for the share of the remaining six percent.

But regardless of Romney's error in tone, and regardless of the subtleties about what kind of people are likely to be moochers, his logical error [assuming it was not a carefully crafted lie to appeal to his millionaire donors] is to identify the 47% of hard Obama support with the 47% who pay no federal income tax. The numbers may coincidentally be about the same, but these just aren't the same people. George Clooney and I are not moochers. Your average octogenarian Tea Partier mooching off social security and Medicare is not going to vote for Obama.

Democrats certainly do better among voters who depend on various government services, but it's by no means monolithic. If 63% of welfare recipients vote Democratic, it means something like 37% vote Republican.

Now, let's not make the same mistake Mitt did. The existence of Republican-voting moochers should not be a strange and humorous anomaly. We should be happy that people aren't making political decisions based solely on personal gain. They may be voting Republican for perfectly rational reasons, like the fact that that party's platform uses the word "God" more frequently.

One of the hot topics in the teaching world is the idea of merit pay. Personally, I support the idea that better teachers should receive better pay. But the devil is really in the details. The modern fad for more and more tests to keep evaluating the childern has resulted in the idea of using the test score results to evaluate the teachers as well. Better teachers would presumably help their students improve more than the students of worse teachers, and testing should uncover this fact, allowing merit to be calculated in an objective fashion.

But the Annals of Improbable Research pointed to a blog that pointed to a paper on the topic.

The researcher tried to validate certain assumptions about the value-added model by testing whether teachers had an observable effect on the previous year's scores of their students: "In data from North Carolina, each of the VAMs’ exclusion restrictions are dramatically violated. In particular, these models indicate large “effects” of 5th grade teachers on 4th grade test score gains."

One conclusion is that these teachers are educating faster than light. A more likely conclusion is that students are not being randomly assigned to classrooms.

One way to fix that would be, obviously, to assign students randomly, but this would mean that getting an accurate measurement of teacher evaluation is more important than actually teaching students. There are valid educational reasons for nonrandom assigment. If you know that Susan is good at teaching unruly boys, or Bob does well with students weak in reading, or that students A and B really need to be separated. (And of course there are invalid, but common, reasons for nonrandom assignement... principals playing favorites.)

The statistics in the paper are pretty hairy. I gave up trying to understand it when the word heteroskedasticity appeared. But I assume the results are valid. And thus it is important to not get hoodwinked into thinking that 'numbers can't lie' and that these value-added methods provide an objective measure of teacher merit. So if we want merit pay, maybe we can just dispense with using the testing and just go with old-fashioned subjective evaluations. Sure, principals can still play favorites, but this is different from how most of us get raises how?
Share  |  |


[User Picture Icon]
Date:September 20th, 2012 05:55 pm (UTC)
See if you can find the clip (it was on the Daily Show, I think) where Mitt "retracts" his 47% comment saying that he wants to represent "all" Americans. Check his body language and tone of voice while he says "all" --there's a disconnect there that is interesting to observe.

Journal of No. 118