Bring Your Questions for Peg Tyre, Author of The Good School

Peg Tyre is a Pulitzer Prize-winning journalist who specializes in writing about education policy. In her 2008 book, The Trouble with Boys, she delved into the growing academic achievement gap between boys and girls to examine why boys are falling so far behind in the classroom. In her new book, The Good School: How Smart Parents Get Their Kids The Education They Deserve, Tyre mines education research data to find out which programs and strategies give kids the highest probabilities of academic success. The result is a concise handbook for parents, one that applies a macro-economic view of education in an effort to create a more rational market around school choice.

As another school year kicks off, Tyre has agreed to answer your questions about The Good School, and anything else education-related. So fire away in the comments section. Before you do, take a look at the table of contents from The Good School printed below, and also read Tyre’s adapted excerpt from the book on the merits (or lack thereof) of teaching to the test.

Read Peg Tyre’s answers to your questions, here.

Table of Contents:
Chapter One: The Preschool Scramble
Chapter Two: Testing
Chapter Three: Class Size
Chapter Four: Reading: What It Takes to Succeed
Chapter Five: When Mathematicians Get Angry
Chapter Six: The Right Balance
Chapter Seven: Teachers Matter
Chapter Eight: The Perfect School

What’s Really Wrong With Teaching to The Test

By Peg Tyre

For as long as there have been standardized tests, there have been people who’ve complained bitterly about instructors who teach to the test. But let’s ask ourselves: What’s wrong with that? I learned about the ancient Greeks in fifth grade. My teacher outlined his curriculum, and then gave us a test at the end of the year that consisted of harder and harder questions about those democracy and toga-loving people. Some of us did well, others less well. If your fifth grader is going to have to sit for a state mandated test on ancient Greeks, then it seems logical that the teacher should teach the material that he’ll be tested on.

But let’s take a closer look at the test itself. Since policy makers want to keep the test brief so that it’s easy and inexpensive to score, maybe they’ll choose three questions about, say, the formation of democracy. What do those questions consist of? One approach would be to ask three questions that reflect an increasingly more sophisticated grasp of those ideas. With this kind of test, each answer the child provides would help determine a two-pronged question: do these middle-schoolers know about the formation of democracy in ancient Greece? And how well do they know it?

But that’s not how standardized tests work. When state education authorities want to find out if your child knows a particular standard, they hire a group of test assessment professionals – mostly researchers and statisticians with a background in education. These test builders use chunks of the course material to fashion their questions. But the question or questions will never be able to test the depth of kids’ knowledge. It’s not meant to. Test builders design questions with one aim: to have roughly 40 to 60 percent of students answer it correctly. When they get those results, the test builders believe that the test question is a legitimate proxy for the material the kids learned. Why this quirky criteria? According to their statistical models, somewhere between 40 and 60 percent of kids are likely to understand any given material. So when they ask a question and about 40 to 60 percent get it right, it is (in the land of standardized testing) a legitimate, statistically defensible question.

If you can hold that in your mind, then teaching to the test starts to seem like a very bad idea. Standardized test questions are being pulled from the lower part of the middle range of what kids should be able to do. If teachers at a school are encouraged to “teach to the test,” they can probably show you test scores that are going up, but that means they are focusing instruction in the most basic part of the material. And that is not a particularly ambitious goal for a school.

The other and slightly more subtle problem with teaching to the test is that it disrupts the natural distribution of correct answers. If the teacher teaches to the test, all the kids might answer the question correctly. But remember, answering the question itself was not the point – it was just a small chunk of course material acting as a proxy for a bigger chunk of course material. And if a teacher successfully gets all kids to answer a test question correctly, it moves from a good, valid question to being statistical “noise.”

Which doesn’t stop schools from encouraging teachers to shape their instruction around getting kids to score well on tests.  These days, standardized test scores are being used for far more than they were ever intended—schools with poor test scores are being reorganized, while teachers who can produce high test scores are being given more pay. And politicians (I’m talking to you, Michael Bloomberg) are making rising tests scores part of their political campaign. The corruption of this delicate and specific form of measurement, most experts believe, is almost inevitable.


2 questions:

1) Looking back, what would say have been the greatest accomplishments and failures of the No Child Left Behind Act?

2) How important is early childhood education? What kind of change would we see if every child got 1-2 years of structured instructional time before entering kindergarten?


To what extent do you think parents have the ability to chose the right school for their kids? Specifically, for families living in poverty that often face bigger social and economic constraints, how much of a choice do parents of these families actually have?


Two questions:

1) Why are policymakers paying so little attention to the importance of quality parenting? Most children spend more time under the care of their parents than they do at school, and it's parents, not teachers/administrators, who are responsible for making sure that students do their homework, come to class prepared, get enough sleep, and eat a healthy diet.

2) What do you think about efforts to increase the length of the school year? One of the reasons that US students are falling behind their international peers is that they spend much less time in school. By the time a student in the US starts high school, they have been in school 1-2 years less than many of their peers in other countries due to the comparatively short length of the US school year.


Parents often look at test scores to determine good schools, but sending your child to a school (or district) with high scores does not mean your child will get a better education, or even a higher test score than he would at another school. What criteria should a parent look at when choosing schools?

Enter your name

"Test builders design questions with one aim: to have roughly 40 to 60 percent of students answer it correctly"

This is simply wrong. It's true if you're thinking about the ITBS from the 1970s, but have you actually tried talked to any of those test builders during the last 20 years? I know several of them, and they say two things:

One is that how the test is structured is determined by each state. What happens in Iowa (the major holdout for the relative-ranking style you describe) is not what happens in the other 49 states.

The other is that in recent years, nearly all of the states (Tennessee makes a fine example here) have said that they just want to know whether the kids know the material, without trying to find the trickiest way of handling it. So if the curriculum standard is "Student can correctly add one two-digit number and one one-digit number without carrying", then 12 + 4 works just fine, even if 100% of the students get the answer right. Actually, since NCLB came into force, the states would prefer than 100% of second-grade students got that answer right.



Peg - is it just teaching to the test, or teaching to any standard (i.e. State standards, core curriculum, etc) that get's this effect? Is this extended into an argument against standards (assuming if you have standards, you have to measure them)?