Friday, November 12, 2010

Friday article summary - is cooperative group work bad?

This term, I'm actually supposed to being taking a course which entails me to read and think about research articles in the teaching of mathematics. So to help me stay on track I'll use Fridays to read and comment articles.
First out: 

Relationship between student and instructional factors and algebra achievement of students in the USA and Japan: an analysis of TIMSS 2003 data. 
J. Daniel House and James A. Telese. (2008)

In this study, House and Telese aim to discover whether different student factors (such as self-confidence and interest in mathematics) and instructional factors (such as group work or alone work) correlate with achievement in the US and Japan. 
To do this, the researchers used data from the 13 year-olds who participated in the TIMSS 2003 study, and carried out a statistical analysis of student responses and their subsequent test scores. 
Results indicate that there are several differences in which factors are influential in Japan and the US. 
In Japan, but not in the US, liking mathematics was found to be positively correlated with achievement. Also in Japan, but not in the US, spending class time doing mathematical computations without the calculator had a strong correlation with achievement - whereas the opposite relationship was found in the US. 

Some similarities emerged as well. In both countries, students who made negative comparisons of themselves compared to other students tended to score lower. In terms of instructional strategies the results were more 
surprising: in both countries cooperative learning was significantly negatively associated with achievement. This is against a background of other studies which have shown mixed and inconsistent results of cooperative learning.  
Homework was positively associated with achievement - which is a kind of "duh" result - but working with "everyday applications of math" in class showed negative correlation with achievement. 

Comments: It's not really clear to what extent we can infer causation and not just correlation. With attitudes, it's probably the chicken and the egg problem, but maybe that's what's happening in instructional strategies as well - with teachers lowering expectations and changing instruction strategies with weaker students. 

Also, this article does not take into account the differences in weaker and stronger students in terms of what they need to develop skill and understanding. Maybe it's better for stronger students to work alone, but weaker, especially unmotivated students, need more group work?

In any case, these results offer reason to suspect that the current trends in collaborative learning need to be carefully evaluated. I suspect that what we are seeing is that many of the strategies modern teacher training encourages have the consequence of temporarily (in class) lowering the standards for student achievement. Instead of demonstrating understanding one by one - students are encouraged to cooperate, thereby masking their lack of understanding from both themselves and their instructor. On exams, when individual work is required, the students suddenly find themselves unable to solve the things they could do in class. 


  1. At a guess (what do you think?) doing a lot of collaborative group work is correlated with being in a class with some seriously low-skilled or disaffected students. If the teacher knows that the result of asking the students to work on problems alone is that some children will do nothing, she won't choose that strategy. Maybe she lectures - also correlated with low scores - or maybe she uses group work in the hope that this will help the low-achieving students to catch up. Members of the class have a low TIMSS average not because of what the teacher did but because of the reason she did it. I guess the "everday applications of maths" correlation is for the same reason - you're more likely to bend over backwards to find such applications if your students are struggling.

    I do think group work, valuable though it is, can be overused. It looks good because you don't have whole groups making no progress, but the difference between real synergy where everyone's learning and a group where some members are not successfully contributing is hard to spot and harder to fix.

  2. Perdita, that's exactly what I mean by "infer causation and not just correlation". But it does make me wonder whether the "cooperate with a classmate on this investigation"-activities I do in class are as beneficial as I like to think.