Challenge 3: Extending Teaching Evaluation Investigations
Chi-Square Test of Independence
While a second course in statistics is a pre-requisite for this class, here is a refresher on Chi-square tests of independence.
Let’s compare the level of SET ratings for Question 3 (The professor used activities to make the class more engaging.) between senior instructors and junior instructors.
1. Using the original teacher_evals
dataset (not teacher_evals_clean
), create a new dataset that accomplishes the following with onedplyr
pipeline:
- includes responses for Question 3 only
- creates a new variable called
SET_level
that is “excellent” if theSET_score_avg
is 4 or higher (inclusive) and “standard” otherwise - creates a new variable called
sen_level
that is “junior” ifseniority
is 4 or less (inclusive) and “senior” otherwise - contains only the variables we are interested in –-
course_id
,SET_level
, andsen_level
- saves the mutated data into a new object named
teacher_evals_compare
.
Helpful functions: filter()
, mutate()
, if_else()
, select()
2. Using the new dataset and your ggplot2
skills, recreate the filled bar plot shown below.
Helpful geometric object and arguments: geom_bar(stat = ..., position = ...)
You should not have to do any more data manipulation to create this plot.
Note that getting the general structure and reader friendly labels is the first step. The next step is to figure out the labels of the bars and the theme of the plot.
3. Look up the documentation for chisq.test()
to carry out a chi-square test of independence between the SET level and instructor seniority level in your new dataset.
Note that the chisq.test()
function does not take a formula / data specification as we have seen before. You will need to extract the variables you wish to include in the analysis using a $
(e.g., evals$level$
).
4. Draw a conclusion about the independence of evaluation level and seniority level based on your chi-square test.
Study Critique
Part of the impetus behind this study was to investigate characteristics of a course or an instructor that might affect student evaluations of teaching that are not explicitly related to teaching effectiveness. For instance, it has been shown that gender identity and gender express affect student evaluations of teaching (an example).
5. If you were to conduct this study at Cal Poly, what are two other variables you would like to collect that you think might be related to student evaluations? These should be course or instructor characteristics that were not collected in this study.
6. Explain what effects / relationships you would expect to see for each of the two variables you outlined.