Differentiating for effective teaching.

At the ESERA conference in Istanbul (in September 2009), the results of preliminary analysis of my data are presented in a poster. This concerns a first analysis of the data of my first survey in September 2008.

The content of this poster is presented here:

The aim of my research is to study the influence of different didactical aspects of physics teaching in the pre-academic track of high school on the exam grades.

I conducted the survey in September 2008 among some 9 000 freshman students in academic science related fields of study in the Netherlands.

Response: 3032 students (34%); 50-50 male/female; 1306 regular physics; 1761 advanced physics.

Principal Component Analysis (PCA) is used to reduce the amount of variables in the dataset and to decrease colinearity problems in the rest of the analysis.  Initially, all students were taken together in this analysis. This procedure has some drawbacks which are discussed elsewhere.

Characteristics of the student (no results)



The most important indpendent variables in this category are:
  1. Dutch influence (parents/students living in the Netherlands)
  2. Highest education of parents
  3. Student reporting to be intelligent
  4. Track in high school (classical no/yes)
  5. Student reporting a handicap

Student results

These variables include exam grades in all subject accept physics and the amount of interesting and understandable lessons the student reports.

The were no important independent variables in this category.

Characteristics of the physics lessons



The most important indpendent variables in this category are:
  1. Teaching strategy
  2. Amount of homework given
  3. Labs outside of the classes
  4. Availability of answers to exercises
  5. Number of mandatory exercises

Characteristics of the physics teacher

These variables include teacher characteristics and their way of assessing the students together.

There were no important independent variables in this category.

Linear Regression is used to find the influential variables first on the physics exam grades and then on the most important of these first exam variables. Only the Standard Coefficients with a significance level of 1% have been retained for this report.

In the following graph one can find the significant variables correlating with the physics exam grades of the students. The variables in color are really components from the Principal Component Analysis.

The most important correlations are analysed separately with Linear Regression.

Grades or selection criteria for high physics exam grades

The correlation of hardworking non-tech girls with high grades is partly (!) due to the fact that these girls are selected on grades in their course of study because of the Numerus Clausus (see Dutch educational system).

Understanding and interest - the right teacher-pupil-combination

One has to be careful with conclusions from these correlations. Further analysis of the data suggests that students who are interested in physics and understand it, tend to rate their teacher as more pleasant. It seems to be safe to look at the correlation between teacher characteristics and student interest/understanding as a two way causal effect. In other words the teacher and the pupils influence each other as the teacher can increase the interest and the understanding of the pupils by being pleasant and keeping control. On the other hand interested students that understand the lessons rate their teachers higher on the 'pleasant scale' and the teacher could also respond to interested and understanding students in a more pleasant way.

Labs are motivating but take a lot of time and can be bad for grades.

Advanced physics students work minimally outside of the classroom.

Teacher pupil interaction.

Insightful students with high grades and pleasant teachers coincide. The influence could be reciprocal and high grades could also be the cause of a positive evaluation of the teacher.

Further analysis suggests that teachers give more test questions without calculations to a class with hard working but less insightful students.

Teaching strategy, i.e. lessons on how to solve problems, is the least ambiguous influential parameter on insight. Note that this only works for regular physics students.

Real influence, but how small it is...

Students ‘being technical’ and ‘working hard’ correlate negatively. Activating tasks ‘outside the book’ are an option.

Regular physics students tend to work hard. A text book can help them regulate this work.

Conclusions

Regular physics students on average work hard, but not always very effectively. Teaching them strategy can stimulate these students to learn more through insight and less by heart.

Advanced physics students, particularly the technical ones, tend to work minimally. To stimulate them to work harder extracurricular and/or practical tasks are advised. Using the lessons to work for themselves on the regular exercises only is counterproductive.

Regular and advanced physics students are not always in separate classes. If differentiation within the classroom is necessary, note that the easiest solution ‘leaving the advanced students to work by themselves out of the book while helping the others along’ is not the most optimal option.

In this sample, regular physics students are generally females and advanced physics students are generally males. From this preliminary study (and others) I have observed that some student characteristics correlate positively with achievement in regular physics and negatively in advanced physics. Similar observations can be made concerning female and male students. In order to be able to observe these influences though analyses using linear regression techniques, they should be analyzed separately. This results in four subsamples: male/regular physics (MR); female/regular physics (FR); male/advanced (MA); and female/advanced (FA).