Thursday, November 29, 2007

Dr. Homeslice Value Added Model-- For Free!

Since the beginning of the school year, I’ve implemented my own value-added model in my classroom. For the uninitiated, value-added is basically a statistical, (most often proprietary) mathematical model that shows the “value” a teacher is able to “add” to student learning and proficiency based on a series of standardized tests. My model is called the DHVAM (Dr. Homeslice Value Added Model) and does not claim any patents (pending or otherwise) and is free for all to use.

To wit:

In my school district, I am given pre-packaged curriculum for my content area. It is not direct instruction (i.e. teaching without thinking, just reading from a script) but is somewhat flexible and opportunistic, so long as you “hit” the state standards.

We are given test questions (written by my district) that correspond to a unit that aligns to our state standards. I take all of the questions, pare them down to the best ones (there are some utterly crappy ones included) and then throw them at my students at the beginning of a unit.

What, you say? Why should you test them at the beginning of the unit? They don’t know the material, they’ve never seen it, never been taught it, you’ve never covered it. Why would you possibly test them over something they don’t know? You know they don’t know it, so why do you need to know they don’t know it, you know?

Reasonable question. Simply put, I want to know what they don’t know, and I want to know how little they don’t know what they don’t know.

So I give them a pre-test, with those district-provided questions. I tell them, this is a pre-test, it won’t hurt your grade, but I want you to take it.

“Why Dr. Homeslice?” asks a student. “Why would you give us a test that doesn’t count towards our grade?”

“Simple,” I answer. “When you go to the doctor because there’s something wrong with you, if the doctor doesn’t know what’s wrong, what do they do?”

“Well, they run some kind of test or something,” comes the customary reply.

“Precisely. That is what I’m doing with you, I want to figure out what’s wrong with you. Or in this case, I want to know what you don’t know.”

This usually elicits giggles from the student’s classmates.

Back to my methodology. I have them take the pre-test, and I tell them that when they take it, and they come to a question that they have no freakin’ clue on (sometimes I use those very words) to leave it blank. If they think they know, I tell them then they can guess. But if they have no idea, leave it blank.

Why no guessing, you ask? I don’t want them guessing and skewing my results. I want to know if they don’t know.

Once they take the test, I grade their answers. (They have an answer form that they fill in, and I don’t use scantron—I grade their responses by hand.) Once I’ve graded them all, I print out my chart and fill in for each question (for each section of students I teach) how many students got each question right. I also turn that number into a percentage of the total number of students who got that particular question right. There’s a lot of hand-work and number crunching, but it’s become somewhat interesting and I daresay comforting to me over the past few months.

Now I look at the numbers and figure out what my students know and what they don’t know. I look at the questions they were asked on the test. I think about prior knowledge, vocabulary, content and so forth. Then I plan for my next lesson, breaking down by each standard, how I’m going to raise those percentages of correct answers on each question. (You know I’m going to use the same freakin’ questions again, don’t you?)

Then I teach my lesson that’s specifically targeted to the areas where my students failed miserably and horribly. I ask for questions and assign homework that is targeted directly to the state standards that showed poor accomplishment on the pre-test.

The very next day (yes, the very next day) I give them a quiz of the same questions (not all of the original pretest, just the relevant ones) that they had for the pre-test. It’s been at least two days since they’ve taken the test, so hopefully they’ve forgotten about the questions, and not stayed up at night thinking about what they should have answered.

I give them the quiz (which this time counts for points) which is based on the lesson in class from the previous day and the homework from the night before. This way I will be able to assess the effectiveness of my in-class and out of class activities.

Those of you statistical wankers, er, I mean wonks will be wondering how I keep my results “pure” and reliable to show growth or no growth on the repeated pre-test items relative to the original pre-test. For me, it was actually very simple to do.

I only allow students who were present the day prior to take the quiz (that is, those who were there when I taught the lesson to overcome the shortcomings on the pre-test). If a student was absent on lesson day but present during quiz day, they don’t take the quiz. I either give them a zero on the quiz or exempt the grade based on whether or not they are able to come up with an excused absence for the day they were out.

I then go back and re-average the multiple choice items to see if there’s been any growth. You’re probably thinking, ‘You know Dr. Homeslice that the students who are taking the quiz aren’t the exact same students who took the pre-test and therefore your results are skewed’. To those folks, I say skew you. I use the pre-test as a baseline to establish a minimum competency assessment for my students.

So how’s it working, you ask?

I just “crunched the numbers” today from a pre-test and the resulting lesson and quiz. On average, students improved the number of correct answers on each item 20% or more.

I gotta tell ya, I’m feeling pretty damn good.

4 comments:

On the Edge said...

That much improvement in just one day! Well, you are a remarkable educator, sir.

Anonymous said...

All of this looks good, until the end. Each student gets each answer right or wrong. A kid can't do 20% better on a question, can they? Evaluating what's been added is the hardest part

Dr. Homeslice said...

Interesting, I didn't think of it this way. So, is it easier to say 20% of the class did 50% better? Dunno.

Anonymous said...

You stated that you dont allow guessing in the pretest because "I don’t want them guessing and skewing my results. I want to know if they don’t know."

But you obviously create incentives for guessing on the posttest, because it is an actual quiz. This disparity (no guessing and no incentives to guess on the pretest and guessing combined with incentives to guess on the post test) would upwardly bias the impact you have had on increasing the knowledge of your students. Say there you provide 4 options on the multiple choice pre and post test. With no guessing students are expected to receive a zero on all questions they have no clue but the same students are expected to get 25% of the questions that they have no clue on the post test. Thus a student who slept through the entire lecture would show on average an improvement of a zero to 25%, which is a massive increase despite gaining no increase of knowledge from your lecture.

Cheers,

Chris

Counters

Locations of visitors to this page