The GiveWell Blog

Two questions about education

Researching Cause 4 has been frustrating. On one hand, I immediately ran right into a pretty terrific overview of what’s known about how to improve high schools. On the other hand, I have a couple burning questions that I’m struggling to answer, and I’d appreciate any thoughts.

1. How does the achievement gap break down by year in school?

What I’d really like to know is what % of students is below grade level in reading and math as of each year (entering K, entering 1st grade, etc.), broken down by income (and race, if possible). I’ve seen scattered statements along these lines, but what I’m describing seems so essential for deciding where intervention is most needed.

2. What’s the connection between education and later life outcomes?

Initially, I hadn’t planned to spend much time on this question – it seems intuitively obvious that a college education leads to much more opportunity than a high school diploma, which in turn is much better than dropping out. But as I look through reports like the one above, I’m starting to question whether targeting academic performance is the same as targeting better outcomes. For example, it looks like one of the best ways to improve test scores is through extremely intensive instruction … is someone’s likelihood of becoming a doctor rather than a drug dealer really more impacted by their knowledge of math after having it drilled into them, or by, say, mental health? (I’d still guess that it’s the former, actually, but I’d like to see some evidence – especially because I’m trying to decide on the relative value of getting dropout risks through high school vs. getting non-dropout risks ready for college.)


  • Doug S. on August 26, 2007 at 12:53 pm said:

    “Grade level” is a fiction. There are many contradictory definitions of grade level, and many of them are not what you would want.

    Here’s one way of measuring grade level. Imagine that your school has 10 grades and 15 students in each grade. Then give each student a standardized test. The top 15 scoring students students are said to perform at a 10th grade level, the next 15 perform at a the 9th grade level, and so on. It’s a standard that can be completely unconnected to objective performance.

    Another definition of grade level is the 50th percentile for students of that age. Using that definition, half of all students will be below grade level, regardless of how proficient the average student is!

    Grade level is a vague and sometimes contradictory standard. Under the No Child Left Behind Act, each state gets to define “grade level” however it wants. This means that success can be achieved simply by lowering standards until they are already met.

    Before one can use “grade level” as a standard for achievement, one needs a specific, rigorous, and useful definition of grade level.

  • Holden on August 27, 2007 at 9:32 pm said:

    Thanks for the comment – this isn’t something that had occurred to me, and you’re right that I should be watching out for how “grade level” is defined.

    Do you have any thoughts on New York State’s scoring system? It’s briefly characterized here – not in enough detail for me to form a strong opinion, but the wording seems to imply objective performance levels rather than excessive use of “curves.” This is the main thing we’ve been using as a proxy for a “grade level” in our research so far; I’m interested in your thoughts on it.

  • Doug S. on August 29, 2007 at 5:00 pm said:

    I really don’t know much about New York State’s scoring system. I read a book about problems with education in the US that was written maybe 15 years ago, and one chapter was devoted to criticizing “grade level” as a standard, and that’s why I’m always skeptical of “grade level.” They also said that all 50 states managed to claim “above average” performance because they compared their current students to an average taken years ago. There are lots of ways to lie with statistics, and one is to, well, just lie.

    Basically, my philosophy on education comes down to this:

    1) Figure out what you want students to be able to do
    2) Make a test that accurately measures student’s ability to do that
    3) Have teachers teach the students how to pass the test

    If “teaching to the test” is bad, then you need better tests. If you want to see if students can write papers, it might be better to have them mail a 10 page thesis paper to a centralized grading center, instead of making them write five paragraph timed essays and sending them to a centralized grading center. If you want to measure mastery of the mechanics of writing and don’t care much about content, you could present students with editing tasks: give them a badly written essay and ask them to clean it up.

    Developing a good writing style is probably the single hardest thing schools want students to do. It’s easy to tell the difference between competent writing and incompetent writing, but defining competent writing is nearly as hard as defining good art. I wish I had solutions, but I don’t.

  • Holden on August 30, 2007 at 9:26 pm said:

    I think from a policy perspective, what you’re saying makes sense – the problem is that there’s no consensus on the question of what we should be teaching children in the first place. What skills do we really need to make sure people pick up as children, rather than later? I have my theories, but I’ve seen relatively little discussion of this and certainly no agreement. It’s obvious that education opens doors to better careers, but far from obvious how much that’s because it teaches useful skills vs. because it’s an employer’s heuristic for finding the most capable people.

    From a charity perspective, we aren’t getting caught up on this question. We’re trying to help kids who could have great careers if not for stupid things like their school system and neighborhood. That largely means finding charities that get them to succeed on the things that are currently considered important (i.e., by colleges, both in terms of admissions and grades/diplomas), and I think reading & math test scores are as good a proxy for that as is available.

Comments are closed.