Daniel Rodriguez-Segura

My practical tips for designing and analyzing powerful experiments

My practical tips for designing, implementing, and analyzing powerful experiments. In today's blog I summarize a new paper I've written for a special issue on power calculations. A key message is that it does not make sense to talk of “the” power of an experiment. blogs.worldbank.org/en/impacteva...

DISTRIBUTIONS: 16 ways to visualize them

Since you folks seem to like lengthy threads, let's look at visualizing distributions. I'll visualize one data set 16 ways and give some other examples of each chart type. #dataViz

📚📊 NEW @cgdev.org RESEARCH: Children who master basic reading and math in primary school earn significantly more as adults. Evidence that foundational literacy and numeracy skills have real economic returns throughout life.

🚀 #Stata #ridgeline ( #joyplot) is now bumped to v1.9 with new features:

👍 Sorting on marked peaks is now allowed.
👍 Peaks can now be labeled and fully customized.

Test it out! More info here:
github.com/asjadnaqvi/s...

Up soon on #SSC!

Last week Professor Stefan Dercon visited the IDB and talked about how to give economic policy advice that is actually useful to policymakers and beneficial to their constituents. It often isn't the "first-best" advice!

You can read the open-access paper here academic.oup.com/jae/article/...

Mucha suerte, Matt!

I had a great time presenting our research on *between*-school heterog. in foundational learning in LMICs (& its policy implications for thoughtful curr. design) at AEDC by @ADBInstitute this weekend in 🇰🇷. Great conference! If you haven't read the paper, please check it out ⤵️!

We hope that this piece serves as a conversation starter to better understand between-school heterog., to put it on researchers' and practitioners' radars, and to get the broader community to think of potential ways to address this issue when it is indeed a policy challenge.

that a large portion of classroom instruction time is currently being wasted as the curriculum only caters to a small share of students and schools — even if it is targeted towards a more realistic level within the system (i.e., even if the curriculum is not "overambitious").

But I fear that in the presence of meaningful bw-school heterog. —absent some intervention— either teachers need to improvise their approach to the curriculum given the learning levels in their own class w/o the support that central curricular guidance is meant to provide, or...

…(e.g., deciding on assessments used, how to deal with measurement error, ensure data fidelity etc.), designing proper grade-on-grade continuity of the content for specific cohorts of children, among other serious concerns.

W/my practitioner hat on: I'm the first one to admit that curriculum targeting for diff schools sounds great, but it's likely v. hard to pull. Think of: implementation capacity (e.g., getting the right books to the right classes, diagnosing precise levels at the school-level...

4) Finally, we show that when between-school heterog. *is* a serious challenge (e.g., some G3 classes at a G1 lvl, some at a G4 lvl), more-fine grain targeting of the curricular expectations can lead to a massive boost in the number of children reached by classroom instruction.

All of these findings suggest that the 1st step towards diagnosing how pressing of a policy challenge between-school heterog. is in a given context —perhaps frustratingly— is to collect context-specific data on learning outcomes, as typical predictors might not be very precise.

3) We also find that between-school heterogeneity is only weakly predicted by average system-wide performance and not very well predicted by local geographic areas or urban/rural status.

2) We find cases where within-class heterogeneity is a more serious challenge than between-school heterogeneity — suggesting very different policy prescriptions (e.g., cross-grade ability grouping-type interventions vs. differentiated curricula at the school-level).

In short, we find that 1) between-school heterogeneity *can* be a serious challenge to setting local curricula. Yet, this is not always the case, as sometimes kids in the same grade across a system are equally behind grade-level expectations.

We use a unique dataset representative of six public educational systems across West Africa and South Asia to better understand the potential extent and potential policy implications of between-school heterogeneity in foundational literacy and numeracy (FLN) outcomes in LMICs.

Yet, at a macro-level, policymakers face a similar problem: they observe a range of school-level performance, and typically need to select a single level to pitch the curriculum at. So, ⬆️ between-school heterogeneity = harder to reach all schools with the same curriculum.

Plenty of research has explored the challenges of within-class heterogeneity: teachers observe a range of levels in their class, and then choose a level to pitch instruction at. So, ⬆️ within-class variation = harder to reach all children through regular classroom instruction.

How can policymakers set national or regional curricula, especially for foundational literacy and numeracy (FLN) in LMICs, when learning outcomes vary significantly between schools? Savannah Tierney and I have a new working paper exploring this question. edworkingpapers.com/ai24-927

It’s happening simultaneously in several places then haha

Ha! Nigeria?