Outsourcing TAs?

8 April 2010 at 11:06 am Peter G. Klein 5 comments

| Peter Klein |

An interesting make-or-buy decision for colleges and universities. Best line, from Roosevelt University B-school dean Terry Friel: “Faculty have this opinion that grading is their job, . . . but then they’ll turn right around and give papers to graduate teaching assistants. . . . What’s the difference in grading work online and grading it online from India?”

Entry filed under: - Klein -, Education, Theory of the Firm.

Westhoff’s The Economics of Food A Hayekian Story About Taxis

5 Comments Add your own

1. MandarKagade | 8 April 2010 at 6:40 pm

I suspect there is a difference between grading them through TAs and outsourcing the same to these virtual-TAs. The distance between the agent (Edumetry Inc) and the principal can mean supervision (monitoring) is an issue- this can generate moral hazard and that will mean that effort in assimilating the argument made by the student in her papers will be sub-optimal.

Even if there are no shirking costs, even with all the diligence, this can impose significant agency costs because the agent may look to develop a set of best “practices”(answers) for each subject–to target economies of scale, for example, as the task is repetitive. This may deter really innovative answers from students (from the second iteration onwards).

With “real-time” TAs, the accessibility to the faculty (and vice versa) as also the lack of incentives to generate a formal “key” to questions (apart from some rules of thumb) during assessing makes “real-time” TAs more capable of assessing students. In addition, TAs are likely to be present when the classes take place; so, they can judge students from a more richer perspective than a far away agent. Finally, TAs accept “TA-ship” out of intrinsic motivation while the virtual TA is only responding to monetary incentives. Intrinsically motivated agent may positively look around for an alternative perspective in the student answers while an agent purely motivated by monetary incentives is unlikely to have that drive.
2. MandarKagade | 8 April 2010 at 7:05 pm

I read the argument from Dean Friel again and must warn faculties in the United States who may be thinking about using these virtual-services, against having Utopian visions of the intrinsic quality of education in India.

I am an Indian myself and completed my undergrad in India (will be attending Columbia Uni. from this fall). The Indian examination system is hardly comparable to the way courses are taught in the US; memorizing and rote-learning is the norm here from high school to graduate degree. Virtual-TAs socialized in Indian educational system might not be the most capable agents for assessing student work from jurisdictions where examination pattern is more context-rich and decentralized. So arguments about “harnessing the capability” have to be made with caution. Training on the job is a poor substitute to cure this gap. Habits that are ingrained from High School onwards will necessarily hold sway.
3. FC | 9 April 2010 at 12:05 am

1. Most US university students learn neither facts nor analysis. Things aren’t so swell over here.

2. By having direct subordinate TAs, professors raise their own status.

3. Hiring people outside the US at wages below the US minimum is considered exploitative (despite Item 2.), while not hiring them at any wage is considered to be social justice.
4. ProfDC | 9 April 2010 at 10:57 am

Full disclosure: I’m advising a team of entrepreneurs who are starting such a business “on-shore” in the USA.

All of us have had graders who were “too good to be true,” who could turn 100 student papers around in one week, every week, with comments that helped the students learn the material; the goal is to have ALL papers graded by such professionals. There are massive benefits to having a professional grading service available (e.g., year-to-year continuity, economies of scope, etc.). Since it creates value, the feasibility of the business is not dependent on wage arbitrage (except, perhaps, between the professor and the grader — an arbitrage present in the TA relationship as well.) Students can get better comments, more consistent grading while universities can get guaranteed costs. (And yes — professional graders can get health insurance coverage and other benefits.)

The keys for running such a business are personnel selection, setting up the correct incentives for quality performance, and a reasonable level of supervision while still leaving graders some autonomy — not unlike the medical field, for example. To claim that grading and student feedback will inevitably be outsourced to the lowest bidder, regardless of quality or qualifications, is a straw man of epic proportions.
5. David Gerard | 9 April 2010 at 11:14 am

My model predicts that assignments specific to a certain class are less likely to be done in house, whereas more generic assignments are more likely to be outsourced. Shall we call this level of “specificity” K for convenience?

I opined elsewhere that this could help in fairness in grading, as the TA is less likely to play favorites.

On the other hand, the lower K is, the higher the probability that the student can buy it off the internet. (The first time I taught managerial econ I gave an assignment to write up a five-forces analysis on the industry of your choice, and two people ended up getting tossed out of school).