Monday, January 19, 2015

Informing parents about missing student work raises GPA and raises test scores for Math--but not for English

I read an interesting study yesterday. Here's my take:

The Study
A New York Times article yesterday referred to an interesting study suggesting that getting more information to parents about their kids' education is low-hanging educational fruit.

According to the study's author, Peter Bergman,  a professor at Columbia, there are three basic problems in this area:

1) parents don't have high-quality information about their kids' grades and work completion

2) parental bias means that parents think their kids are doing more homework than they really are

3) the short-term bias of children (Bergman says they have "higher discount rates") leads them to conceal poor performance and poor homework completion from their parents

The literature about parental involvement has been ambiguous, according to Bergman, because it has been difficult to distinguish between cause and effect. Studies have even found negative correlations between parental monitoring/involvement and student achievement. This could either be because parental nagging is counterproductive or because parents tend to monitor and nag kids who are underperforming. Bergman thinks it's the latter.

Bergman models the situation using some fancy-looking game theory, but his experimental design is simple: the treatment group, at a California school serving mostly poor kids, got messages about missing assignments, including specific information like the date and the page number of the textbook. The control group got the default amount of information that school and teacher provided parents: report cards four times a semester and teacher contact as the teacher chose (usually very little).

Positive Results: GPA goes up; math test scores go up
The results were positive: in the treatment group, both overall GPA and math test scores increased by about .2 standard deviations. That's a pretty big increase.

This is not surprising, but it does point to an obvious lesson: schools and teachers should move to as transparent a grading system as possible. Parents should have continuous access to their children's assignments and grades. As a teacher, this will push me to be more timely and thorough about updating my gradebook, but that's a good thing. Or--it would be a good thing if I were a math teacher.

Unfortunately--as usual--English test scores don't change
Another interesting result of the study--for me, more interesting than its headline finding--was that the English test scores of the treatment group did not go up. (Well, they went up .04 standard deviations, but that's not much).

As Bergman notes, this disparity is typical--many studies find it easier to raise math scores than English scores (2). But it is nevertheless a real puzzle, one that calls into question the effectiveness of English class.

Informing parents about missing work seems to have had some direct effects. Parents who were informed about missing assignments were in better touch with the school, and more of them attended parent-teacher conferences. Parents in the treatment group took away twice as many privileges from their children (mainly access to screens and friends). Students in the treatment group skipped class less, completed more of their assignments, and attended after-school tutoring more often.

Therefore, this experiment found that a group of students who had more engaged parents, who went to class more often, who did more of their schoolwork, and who went for help after school, nevertheless did not make any more improvement on their English test scores.

This is a striking result.  Going to math class and doing your math homework really matters. Going to English class and doing your assigned English work seems to matter much less.

Results like this should make English teachers very uncomfortable.

How might English teachers respond?
An English teacher at this school could interpret these results in three ways:

1) the purpose of English class is not really to improve reading ability, so this result is irrelevant, and I am going to keep doing what I'm doing.

2) the purpose of English class is to improve reading, but coming to class and doing the assigned work does not seem to serve that purpose, so I am going to change my practice.

3) the purpose of English class is to improve reading, and although coming to class and doing the assigned work does not seem to serve that purpose very well, changing my practice would be inconvenient, difficult, and unnatural, so I am going to keep doing what I'm doing.

I think most English teachers respond, if unconsciously, in the third way. Here's why:

Education is conceptualized in a way that fits math, not English
English teachers work within a system that is set up for teaching subjects, not for getting kids to read a lot. The other academic subjects, including math, are about teaching discrete concepts and techniques, while English is about continual gradual improvement in the exact same activity students have been practicing since they were in first grade. In English there is no proper "content," and the usual ways of thinking about education are inappropriate and inadequate for thinking about reading.

The usual way of thinking about education is that the teacher teaches a skill or concept, and then students practice or conceptualize on their own and get feedback from the teacher about their performance. For me, this model was most vividly represented by the Australian researcher John Hattie, who in his book Visible Learning says that the "heart of the model of successful teaching and learning" is like rappelling in that "the goals are challenging, specific and visible," the skill being taught does not come naturally, the experience of learning is "exhilarating," and "it is abundantly clear what the success criteria are". This may be a good model for learning a single skill like rappelling off the top of a building, and it may work to some extent for subjects like math, science, history or foreign languages, in which there is "material" to be learned or "skills" to be mastered, but it is not a good model for improving your performance in an activity, like reading, that you have been doing over and over for many years.

The best way for students to get better at reading is probably to spend a lot of time reading interesting material, and to spend some time discussing the meaning of interesting or difficult texts. Unfortunately, neither the reading or the discussing fits easily into education's standard conceptual framework. So instead English teachers often assign work and plan classroom activities that are neither actual reading of texts nor discussion of texts, and it doesn't really matter if our students do the work or take part in the activities.

Bergman's study shows that we should be more transparent about missing student work, but it also shows that we English teachers should be rethinking our practice.

"Parent-Child Information Frictions and Human Capital Investment:
Evidence from a Field Experiment" by Peter Bergman

(2) This phenomenon is discussed somewhat in Bergman's paper, and in more detail in a 2010 paper by Eric Bettinger. It's also touched on in a 1996 meta-analysis of summer learning loss by Cooper et al. The paper about summer learning loss is interesting, since it finds that although on average the summer learning loss is greater in math, for poor kids it is greater in reading. Food for thought.


  1. One problem with our whole current frame is internalizing the idea that math is one half of education and reading (plus maybe writing) is the other. I'd argue that in the big picture, math as we generally conceptualize it as a linear sequence of skills and techniques it is the outlier. "Education" is conceptualized like this insofar as we can forget about anything other than math, and maybe certain topics in science.

    The Common Core standards for ELA/Literacy were doomed from the start without a clear conceptual framework for English and the other core academic disciplines. The very idea that you could start writing standards without that framework is absurd.

    1. Hm. Maybe math is and outlier in the sense that it is more linear and cumulative than, say, science, but then--maybe not. Imagine month-long units in the standard academic subjects. In every subject but English, doing well in the monthlong unit would be very likely to improve your score significantly on a standardized test in that subject. The same is less true in English.

      In math you might spend a month learning some basic trig, in physics you might spend a month learning some basics about Newtonian mechanics/motion, in French you might spend a month learning the imperfect tense, and in history class you might spend a month learning about the Civil War. In all of these units you would have learned some material that was actually new, and if you did well on the end of unit test, you might expect to the unit to help you do better on a standardized test in that subject.

      In English, the matter is totally different. In English class, you might spend a month reading the Scarlet Letter. If you are like many students, you might spend the month doing a lot of activities related to the book's storyline, you might write a paper about Hester Prynne's proto-feminism, and you might do reasonably well on the unit test--ALL WITHOUT HAVING READ THE BOOK!

      If so, then you would probably not do much better on the standardized reading test than someone who had done poorly on that unit but had, say, read rock climbing magazines regularly.

    2. But I agree with you about the common core, of course. No clear conceptual framework for English, and basically no evidence that their chosen approach is any better than any other.

    3. Researchers should do what I did: interview top readers to find out how they because so good. I found out that they loved to read, and always made time for pleasure reading. Here's the irony: the typical English curriculum actually works against developing avid readers. Few kids read all of the assigned books, yet read nothing else, since assigned reading tends to drive out all other reading.

      Great post.