Class Size Doesn’t Matter

I prefer one-on-one instruction or small group trainings to learning from books. This might be because I went to a very small high school with a graduating class of 13 or because I like to explore topics by asking questions. Given my background, I’ve always believed that smaller classes are more effective; more individualized interactions lead to improved learning.

A recent Harvard study disproves my theory. In fact, the researchers found that none of the most commonly cited factors – class size, expenditure per pupil, teacher certification, or percentage of teachers with advanced degrees — are correlated with school effectiveness. Conventional wisdom is simply wrong.

Forty years of qualitative research at 35 charter schools suggests that five factors account for half of all differences in school effectiveness:

  • frequent teacher feedback,
  • the use of data to guide instruction,
  • high-dosage tutoring,
  • increased instructional time, and
  • high expectations.

Although the research was based on in-person schooling, I assume that similar findings would hold for Web-based instruction. High-dosage tutoring and frequent teacher feedback might be more difficult to administer via the Web but it’s possible. For example, technology exists to track which questions take longer for students to answer so that electronic courses could dynamically adjust, adding more information on these topics.

These results are also consistent with my own schooling experience. My teachers gave individualized feedback during every class, we were required to have regular tutoring in our weakest subject, and we spent two more hours per week in class than my friends in other schools.

While all of this might have been easier to accomplish due to smaller class sizes, the research shows that small classes alone are not enough.

, ,

9 Responses to Class Size Doesn’t Matter

  1. Annie Miu Hayward March 18, 2012 at 8:20 pm #

    I wonder if ‘high expectation’ bears some correlation to the ‘tiger mom’ principle …

  2. timoelliott March 18, 2012 at 10:13 pm #

    My first though was “a study of just 35 schools — very different from the vast majority, in a small geographic area — is supposed to mean something?”

    But the OECD data (which takes a LOT more data into account, although obviously not as fine-grained as the Harvard study) comes to the same conclusion:

    • Elizabeth April 24, 2013 at 8:09 pm #

      How can I access the Harvard study/

  3. Anonymous March 19, 2012 at 5:32 am #

    Those findings are consistent with my experience with corporate learning – regardless of the delivery method. When an employee and manager commit to learning, the manager holds the employee accountable for learning and then provides feedback on how the employee is applying their learning to their job, the impact to the business is much higher than if we simply “push” training and mandate that people complete it.

  4. Meta Brown March 20, 2012 at 3:27 am #

    Qualitative research? Where’s the quantitative research?

  5. Robert E March 21, 2012 at 12:36 pm #

    This is a perfect example of using a good strategy to drive better results rather than looking at measurements.
    The key to this is how the focus on measurements in education have more been about Proving, instead of Improving.
    As in business when the focus is on the wrong measures, resources in education keep whipsawing between the measurement de jour without showing improvement. The goal needs to be improve learning even if it means disappointing popular-held notions.

  6. Anonymous March 22, 2012 at 7:01 am #

    Isn’t frequent teach feedback a function of class size? The larger the class the less frequently a teach will be able to give meaningful feedback . . .

  7. Arun Krishnaswamy April 5, 2012 at 6:48 pm #

    Relating to personal experience, where I came from, schools typically had class sizes of 50. The teacher’s passion for the subject and love for teaching was probably the #1 factor on overall classroom performance especially not for the best kids (who did well despite the teacher), but how the back benchers did and how the teacher involved them. Individualized feedback was not practical or possible, however the other 4 of the 5 factors did apply in varying degrees. The study proves what seems intuitive to me, and also shows how measuring proxy variables (easy to measure) often take one away from the fundamentals which is the teacher-to-student interaction quality. How does one measure the emotional involvement so essential to learning and learning outcomes?

  8. anonymous June 9, 2014 at 11:50 am #

    Conventional wisdom is not wrong, in the conventional environment class size and expenditure per people are directly naturally correlated with the frequency of teacher feedback and the dosage of tutoring.

    Conventional wisdom maybe simply invalid in not conventional environments and hopefully we are getting closer to understand the driving factor behind it.

Leave a Reply