In May 2004, I surveyed the CS department TAs. I announced the survey to all MS and PhD students, and asked everybody who had TAed in the preceding year to respond; 39 TAs filled out the survey. You can see everything that I asked on the survey form.
This page summarizes some of the most interesting results of the survey. It is intended to inform and stimulate discussion about course staffing policy and the role of the TA. For full details of the results, you should look at the raw data. (I have removed course number and general comments to preserve anonymity.)
One of the most interesting parts of the survey responses was the general comments section, where I asked TAs to write anything they wanted the committee on curriculum and course staffing to read.
A 50% TA appointment is for 20 hours/week and a 25% appointment is for 10 hours/week. Among survey respondents, 50% TAs reported workloads pretty close to this official level, but 25% TAs reported working one and a half times as much as they should have.
The error bars show 95% confidence intervals, but you should take them with a grain of salt, because the sample was self-selected. (More detail on the range of workloads for different TAs can be seen in the scatter plots below.)
This next chart shows a list of common TA duties, with the length of the bar proportional to the percentage of TAs that reported doing each activity. The chart does not show how much time they spent on each activity.
A goal of the survey was to see if there was any pattern to the situations in which TAs feel that a course is understaffed.
This first chart shows the relationship between the number of hours per week that a TA works and her perception of the staffing level in her course. Among survey respondents, the more hours a TA works, the more understaffed she feels her course is. No big surprise there. You can see the range of workloads reported by 25% and 50% TA in the horizontal distribution of the points on this chart. There is a lot of spread and overlap between them. Ideally, the organge dots would all be bunched around 10 hours/week and the greens around 20.
This second chart shows the relationship between the Student-TA ratio in a course and the TA's perception of how over- or under-staffed their course was. In courses without a heavy programming component, the relationship is not surprising: the more students each TA is responsible for, the more that TA feels the course is understaffed. In contrast, in courses with a lot of programming, there is no strong connection between the number of students per TA and the feeling of over- or under-staffing. The strongest pattern is the separation of non-programming from programming courses. Most TAs in non-programming courses felt that their courses were properly-staffed or over-staffed, while most TAs in programming courses felt their courses were under-staffed. Ideally, both the blue dots and the browns would overlap and both be centered on "just right".
This third chart helps to explain the patterns in the second. It shows the relationship between the student-TA ratio in a course and the average number of hours a TA in that course worked. The patterns from the second chart are repeated: In non-programming courses, more students per TA means more hours of work per TA, but in programming courses, there is no connection between the student-TA ratio and the amount of work. An average across each of the course types reveals that TAs in programming courses work more (22 hrs/week) than TAs in non-programming courses (18 hrs/week). I left out CS106 from all three of these charts, because the use of undergraduate section leaders in that course gives it an artificially high student-TA ratio.
I asked TAs several questions about their communication with professors and the professors' understanding of TA work. The results were positive. TAs and profs communicate frequently. TAs are generally comfortable talking with profs about their workload, especially when their course's prof is also their advisor. Professors have a good understanding of how much time TA duties require, and TAs and professors agree on the responsibilities and time commitment of the TA position:
|How often did you communicate with the prof?|
|How comfortable were you talking with the prof about your workload? (1="not at all", 5="very")|
|How well did the prof understand how much time each of your responsibilities took? (1="not at all", 5="very well")|
|How well did your understanding of a TA's responsibilities and time commitment match with the prof's understanding? (1="prof expected less", 3="same", 5="prof expected more")|
My general impression is that TAships in the department are in pretty good shape. For most students, TAing is a rewarding experience. Relationships between TAs and faculty are strong. A survey like this one inevitably ends up focusing on perceived problems, but before I address those, let me make the point that these concerns are minor compared to those at some other universities.
It is important not to conclude too much from the quantitative results of the survey. Although the respondents form a large enough sample size and cover a broad range of course types, they are a self-selected sample. TAs who have concerns were more likely to respond, so the survey results probably overstate problems like overstaffing and understaffing. Nevertheless, I have heard similar concerns brought up elsewhere, so while I am skeptical of the full quantitative magnitude of the results, I do not doubt their qualitative message.
I believe that TAs are mostly concerned about how the department distributes TA assignments among its courses. Specifically,
I think that both of these concerns arise from the same basic problem with our TA assignment policy: The number of students enrolled in a course does not reliably predict the amount of TA support required by that course. (See section 4.) It works OK as a first approximation in classes with a lot of grading, but in courses where many of the TA duties do not depend on the number of students, and especially in classes with heavy programming, the current formula misallocates TAs.
A related problem that I have seen is general confusion and uncertainty about the TA assignment policy. Most TAs know that more TAs are assigned to lab courses and courses with heavy programming, but they don't know how much more or which courses have received this designation.
These recommendations have two sources. In part, they echo the ideas of TAs given in the general comments section of the survey and that I have heard raised in many informal discussions. The recommendations also represent my response to some of the numerical results of the survey, especially with regard to distributing TA resources better. I have left out the most common suggestion of hiring more TAs and only listed those things which would not cost more money.
a*n, the calculations should be of the form
a*n+b, with a constant term to capture the fact that some of the work for a course does not depend on the number of students enrolled.