![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
Midnight Ruminations
Sep. 8th, 2006 11:10 pmhttp://writersweekly.com/contest/1stwinter04.html
So, nearly three years ago, I wrote a short story for a contest, and that was the title. It was about High Stakes Standardized Tests and stress and such. You can still read it if you like. Now, it'll be midnight before I'm through with this, and here I am, awake, because of High Stakes Standardized Tests.
I won't say my school is perfect. I won't say that public education is perfect. I won't even say that I think we're doing the best possible job we can do as either of those entities. But I do think some things, and I am going to say them.
I HATE High Stakes Standardized Tests. I don't think they're a good measure of student performance, overall. I think even the ones that we have are culturally biased. And I think that the sample size for a school like mine is so small as to make the results statistically useless.
But. More. I don't think we use what data they do provide in any kind of meaningful way.
I don't think the school does. I don't think the district does. I don't think the state does, and I don't think the NCLB people do.
To start, until two years ago, we had two different tests for different grade levels. One of those had such useful questions as: Which of the following is an example of an octagon: A) A Yield Sign, B) A Stop Sign, C) A Wong Way Sign, or D) None of the Above. Well, that's fine and dandy unless you live in a town of 600 people where there are no roads. No roads. No cars. No signs.
Now at least all the kids are taking a test made by the same company. And, it's made specifically for Alaska, so it's supposed to be more useful than that. Eh. It has good points and bad points. I think my biggest problem with it, as a test, is that regardless of whether you're taking the Reading, Writing, or Math sections, you have to be able to read very well to understand what the questions are asking. Not, read okay, but read very well. I've looked at the test. I had a student with accomodations and needed to read the math questions to him, and some of them I had to read multiple times, just so I could phrase them well enough that anyone would be able to understand what was being asked. Well, when 98%% of your students are English Language Learners (and the other two percent were absent when the test was given to make the determination), guess what - you're not going to get accurate information about math ability if they can't read the test.
Okay. Let's assume that all the kids are taking a reasonable test, and that each year the test is reasonably similar to the one before, such that a year's growth would show as the same score on the next test.
What do you do with the student numerical test scores? Well, you chunk them. You say a student either was Far Below Proficient, Below Proficient, Proficient, or Advanced. Okay... to a point (which I'll get to later).
But, what do you do with that information? Apparently, you look at the third grade, and you say, What percent of students in the third grade this year were not proficient? And then you compare that to the percent of students in the third grade last year.
Well. If you have 1000 third graders every year, that might give you some statistically interesting information.
What if you had 23 last year, and 8 the year before?
What if three of the 8 kids were children of teachers who had gone to posh pre-schools, and upper middle class white elementary schools for first and second grade?
What if instead 3 of the 8 kids were foster children just removed from their abusive homes in another village three months ago?
These aren't real examples (at least not numerically), but they are truly possibilities here.
So I don't think comparing percent of passing students from grade 3 2005 to grade 3 2006 is a statistically useful thing to do. At least not if what you want to know is whether your school is doing a good job.
But that's what NCLB does. And that's what the State is doing. (For now.)
Vic decided to do it differently. He compared the percent of passing students in grade 3 2005 to the percent of passing students in grade 4 2006. Well. that's /more/ useful.
Except when 80% of your students are below or far below proficient. Then all it tells you is that there were 80% far below proficient, and there are still 80% far below proficient.
It /doesn't/ tell you if they scored 10% correct the first year and 40% correct the second, or if they went the other way around. It doesn't tell you what's actually happening from student to student.
If we assume that the tests are valid, and that we want to measure whether our program is helping our students succeed, we need to know whether they're making progress, even if they haven't managed proficiency.
What /I/ want to do is take all the SBA (Standards Based Assessment) scores that we have and look at trends. I'd /rather/ look at individual student scores, because we have so few students, but I'd be happy taking averages. Heck. I'd even be happy taking the average SBA scores from the entire school for as far back as we have it. (Though it'd be a bit iffy comparing when we only had 4 grades take the SBA to when we had 9 grades taking it. And, I'm not really sure we can really compare HSGQE with SBA, but they're theoretically measuring the same skills, at least, and in similar formats.)
Anyway. That's what I think about test scores at 11:45.
So, nearly three years ago, I wrote a short story for a contest, and that was the title. It was about High Stakes Standardized Tests and stress and such. You can still read it if you like. Now, it'll be midnight before I'm through with this, and here I am, awake, because of High Stakes Standardized Tests.
I won't say my school is perfect. I won't say that public education is perfect. I won't even say that I think we're doing the best possible job we can do as either of those entities. But I do think some things, and I am going to say them.
I HATE High Stakes Standardized Tests. I don't think they're a good measure of student performance, overall. I think even the ones that we have are culturally biased. And I think that the sample size for a school like mine is so small as to make the results statistically useless.
But. More. I don't think we use what data they do provide in any kind of meaningful way.
I don't think the school does. I don't think the district does. I don't think the state does, and I don't think the NCLB people do.
To start, until two years ago, we had two different tests for different grade levels. One of those had such useful questions as: Which of the following is an example of an octagon: A) A Yield Sign, B) A Stop Sign, C) A Wong Way Sign, or D) None of the Above. Well, that's fine and dandy unless you live in a town of 600 people where there are no roads. No roads. No cars. No signs.
Now at least all the kids are taking a test made by the same company. And, it's made specifically for Alaska, so it's supposed to be more useful than that. Eh. It has good points and bad points. I think my biggest problem with it, as a test, is that regardless of whether you're taking the Reading, Writing, or Math sections, you have to be able to read very well to understand what the questions are asking. Not, read okay, but read very well. I've looked at the test. I had a student with accomodations and needed to read the math questions to him, and some of them I had to read multiple times, just so I could phrase them well enough that anyone would be able to understand what was being asked. Well, when 98%% of your students are English Language Learners (and the other two percent were absent when the test was given to make the determination), guess what - you're not going to get accurate information about math ability if they can't read the test.
Okay. Let's assume that all the kids are taking a reasonable test, and that each year the test is reasonably similar to the one before, such that a year's growth would show as the same score on the next test.
What do you do with the student numerical test scores? Well, you chunk them. You say a student either was Far Below Proficient, Below Proficient, Proficient, or Advanced. Okay... to a point (which I'll get to later).
But, what do you do with that information? Apparently, you look at the third grade, and you say, What percent of students in the third grade this year were not proficient? And then you compare that to the percent of students in the third grade last year.
Well. If you have 1000 third graders every year, that might give you some statistically interesting information.
What if you had 23 last year, and 8 the year before?
What if three of the 8 kids were children of teachers who had gone to posh pre-schools, and upper middle class white elementary schools for first and second grade?
What if instead 3 of the 8 kids were foster children just removed from their abusive homes in another village three months ago?
These aren't real examples (at least not numerically), but they are truly possibilities here.
So I don't think comparing percent of passing students from grade 3 2005 to grade 3 2006 is a statistically useful thing to do. At least not if what you want to know is whether your school is doing a good job.
But that's what NCLB does. And that's what the State is doing. (For now.)
Vic decided to do it differently. He compared the percent of passing students in grade 3 2005 to the percent of passing students in grade 4 2006. Well. that's /more/ useful.
Except when 80% of your students are below or far below proficient. Then all it tells you is that there were 80% far below proficient, and there are still 80% far below proficient.
It /doesn't/ tell you if they scored 10% correct the first year and 40% correct the second, or if they went the other way around. It doesn't tell you what's actually happening from student to student.
If we assume that the tests are valid, and that we want to measure whether our program is helping our students succeed, we need to know whether they're making progress, even if they haven't managed proficiency.
What /I/ want to do is take all the SBA (Standards Based Assessment) scores that we have and look at trends. I'd /rather/ look at individual student scores, because we have so few students, but I'd be happy taking averages. Heck. I'd even be happy taking the average SBA scores from the entire school for as far back as we have it. (Though it'd be a bit iffy comparing when we only had 4 grades take the SBA to when we had 9 grades taking it. And, I'm not really sure we can really compare HSGQE with SBA, but they're theoretically measuring the same skills, at least, and in similar formats.)
Anyway. That's what I think about test scores at 11:45.
