I’ve recently experienced a total blogging blackout due to being in Northern Uganda to oversee the baseline data collection for an ambitious three-year RCT that is designed to evaluate the impact of a literacy program run by Mango Tree Educational Enterprises Uganda. My coauthor and advisor, Rebecca Thornton, and I also worked on the smaller, pilot RCT of Mango Tree’s program last year. The early results from that experiment are really exciting – this is an incredibly effective educational intervention – and we expect to learn even more from the current study, as well as using it as a platform to understand how parents, students, and teachers make decisions about how much (and in what way) to invest in schooling.
There will be more on those results in future posts, but for now I wanted to offer some potentially useful ideas for 21st-century Americans that I’ve picked up on during my visits to the Lango sub-region, which is the area in which Mango Tree works. When I lived in Malawi while running the experiment that underlies my job market paper, I occasionally wrote posts about things that America could learn from Malawi; this is in the same spirit. The Lango sub-region is home to the Langi, a tribe from the broader Luo ethnic group. They make up virtually all the kids in our study, and as a result I have spent an inordinate amount of time staring at different lists of Langi names. One thing I noticed quickly is that siblings almost never have the same surname. With a representative sample of thousands of kids, I quickly rejected the possibility that these were all cases of half-brothers and sisters. After asking around about it, I learned that instead of giving children one of the parent’s last names as their surname, they choose the surname of one of their relatives of the same gender. Another thing my friends explained to me, after I commented that nearly 50% of surnames start with “A” and a similar fraction with “O”, is that almost all names starting with an “A” are for girls, and most “O” names are for boys.
It gets better: sometimes it is difficult to decide between the many relatives you have whose surname you will assign to a child. Pick one brother, and another one might get jealous. One guy said that, in order to resolve this tension, he gathered all the relatives who wanted their name to be used into a room, wrote all the names on pieces of paper, and picked a name by lottery. It’s like a Freakonomist’s fantasy: want to know the effect of having a particular name on your success in the labor market? Just run the regression, they are literally random conditional on family background. (Unfortunately the lottery method seems to be the minority approach, with most people just choosing whichever name they like best for their kid.)
There are two possibly-useful ideas here for America as we adapt to a world where patrilineal naming traditions no longer hold. Our surname system sucks: couples agonize over who should take whose last name in a marriage, and if the partners keep their birth names then they have to figure out what to call their kids. We have no clear guidelines, and the emerging de facto norm of wives keeping their maiden names but children taking their father’s surname is fairly confusing and also male-biased. The Langi system offers two options for resolving this. The first possible solution is that we give up on the idea of trying to unite people through their family names: pick your children’s family names based on ancestors or relatives you want to honor. When you file the marriage paperwork, you could even do the same for yourself.
A second option is take my friend’s approach, and avoid resentment and anger by using a lottery to assign last names. If you want to use your surname or your spouse’s, but can’t come to an agreement, flip a coin! Or better yet, hold a lottery over a range of possible surnames. If you want to be smart about it, and realize that the coin flip will ensure fairness on average, but not for all possible actual outcomes, you can do a stratified lottery. For example, a “heads” could mean that your first kid gets your surname but that your second one will get your spouse’s, while a “tails” reverses the order. That would be a better system than our attempts to shoehorn modern families into a pseudo-traditional naming system by endlessly stringing together hyphenated names.
Jason what is the RCT studying (if you dont mind me asking)?
I don’t mind at all! The RCT is studying a literacy program that is centered around teaching students to read and write in their mother tongue (the one they grow up speaking at home) first, rather than English-first which is the norm in many former British colonies in Africa. Evidence from linguistic psychology suggests that, while becoming bilingual has cognitive benefits, jumping directly into reading and writing a language you don’t even speak yet (when you still can’t read or write in your own language) is too difficult. Students who do this tend to struggle to learn to read at all. The program also provides a curriculum, intensive teacher training and supervision, teaching and reading materials, and detailed lesson plans.
We are comparing the program to two alternatives. There is a control group that is just the same as any school in the local region. There is also a government-administered version of the program, where the training and supervision is provided by government officials instead of by Mango Tree experts. That version is cheaper and easier to scale up, and we are interested in how much the program’s effectiveness is reduced when things are done that way.
Membership in each group was determined by a public lottery, where representatives from the schools pulled tokens from a bag that had colors corresponding to group assignments. Mango Tree has capacity and funding constraints that limit how many schools they can go to, so holding a lottery was a fair way of picking among a large number of eligible schools. It also makes the program’s expansion into a randomized experiment: we can compare the treatment and control schools to see how much the program causes literacy test scores to rise.
Interesting stuff! Thanks Jason!