May Contain Lies by Alex Edmans;

Author:Alex Edmans;
Language: eng
Format: epub
ISBN: 9780520405851
Publisher: University of California Press

A shock to the system

Given the success of RCTs across a variety of fields, you might hope to use them for the debate on whether parental choice improves school performance. You’d randomly merge some school districts (the test group ), keep your nose out of others (the control group) and compare their outcomes.

But such an experiment would be both costly and risky. The expense of merging school districts is huge, and if competition does improve performance, thousands of kids in the merged districts will suffer worse education. This is the key limitation of RCTs: they’re effective where you can use them, but sometimes the consequences of putting people in the wrong group are so severe that you can’t take the risk. To test if smoking causes cancer, you can’t recruit volunteers and force half of them to smoke; to verify Belle Gibson’s advice, you can’t sign up cancer patients and order 50% to forgo chemotherapy for clean eating.

In situations like this, you can’t conduct an intervention study – you can’t intervene and change the input yourself – so what you need is something that already does so in the real world. This is known as an instrument.** An instrument causes the input to change, but for random reasons that have nothing to do with the output. In simple terms, it’s a shock to the system. You can then observe what happens after the input was shocked and conduct an observational study that requires no intervention.

A famous paper by economist Caroline Hoxby used rivers as an instrument for school choice to solve our education conundrum.4 In the US, school districts were formed in the eighteenth century, when crossing a river was difficult because there were no cars and few bridges. As a result, districts rarely crossed rivers, so that children wouldn’t need to do so to get to school. Metropolitan areas with several rivers thus had multiple districts; since districts haven’t changed much over time, these areas still have many today.

Hoxby decomposed the input – school choice – into two parts: the exogenous part that can be attributed to the instrument (rivers) and the endogenous part that can’t (and instead arises from common causes like parental engagement).†† Then, she linked only the exogenous part to the output: student performance. Common causes don’t affect exogenous school choice – there’s no arrow between them – so they can’t be the reason for any correlation between exogenous school choice and performance.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.