Economics of Charity at ANU

Professor John List, Head of the ANU John Mitchell Economics of Poverty Lab, is speaking on "The Economics of Charity" in Canberra, at the ANU College of Business and Economics. I am watching on-line. He is author of "The Voltage Effect: How to Make Good Ideas Great and Great Ideas Scale" (2022). 

Professor List related how in his early work he conducted experiments into getting donations for a university. He found that private donations in advance to a pubic call increases donations, as a quality signal. Taking an extreme example, mentioning there had already been a donation from The Bill and Melinda Gates Foundation greatly increased subsequent donations. However, traditional economic models on price effects don't work. In particular, different matching ratios (from 1:1 to 1:3) doesn't make a difference.

Another interesting point is that marketing gifts, only tend to work once. Men are more price sensitive than women.

Interestingly Professor List discussed the use of AI, both for experiment, and in practice. 

Overall an interesting talk, but I had some difficulty with the US outlook, terminology, and jargon. At times I had difficulty working out what were US colloquialisms, and what were technical terms from the field.

Gates Foundation report on Student Data for Personalized Learning

The Bill & Melinda Gates Foundation has released the 36 page report "Teachers Know Best" along with a two page summary "How Teachers Approach Data". The report advocates the use of "... student data to tailor and improve instruction for individual students..." (p. cov2). Such an approach, I suggest, may do more harm than good, by diverting resources away from the design of quality instructional materials and by setting unrealistic expectation as to the level of tailoring possible with the resources available to teachers.

The report divided teachers into Data Marvens 28%, Growth Seekers 20%, Aspirational users 17%, Scorekeepers 11 %, Perceptives 14 %, and Traditionalistss 10%. The report's authors clearly believe that the marven's data-driven personalized instruction is preferable. However, where are teachers going to get the data and will they be given the time to personalize student's instruction? Assuming no more resources are provided, the funding to provide data analysis tools will come from the education budget and reduce resources for course materials. Similarly, more time by teachers taken on a personalized approach will result in lass time for class teaching.

I suggest what is instead required is instructional design, which individual teachers do not have the time or  resources to do. Statistical analysis can be used to crunch the numbers on large numbers of students to see what educationally works and what does not work and what aspects of subjects students have difficulty with. These insights can be built into the educational materials and teacher training. Teaching can be given help in identifying what students will have difficulty with and how to help them. But each teacher does not need to become a statistician to do this.

There are ways to use personalized learning to help students. However, it also has to be done in a way which helps teachers and is affordable. As an example, last year I used peer assessment in my ICT Sustainability course. This helps students, as by having to assess their peer's work, students gain insight about their own work. As a by-product this reduces the assessment load of the teacher (and also reduces student appeals, as students are less likely to object to the marks their peers give).