iiinova — where your digital self keeps evolving

Nathan Favero @nathanfavero.com · 05/01/2026

My latest post: Why old school regression diagnostics are probably worth running... we really should check our variables for outliers, skew, multicollinearity, etc. open.substack.com/pub/nathanfa...

Brittleness: A new concept for statistical modeling
Factors like small samples, multicollinearity, and outliers may not themselves break our models, but they create conditions where a break can be dramatic
When learning statistical tools for empirical analysis, many of us were...
nathanfavero.substack.com

042

Nathan Favero @nathanfavero.com · 17/11/2025

open.substack.com/pub/nathanfa...

Reflecting on Null Results
What I have been learning and doing with null results over the past year

022

Nathan Favero @nathanfavero.com · 03/11/2025

open.substack.com/pub/nathanfa...

Substack tile: 3 rules for research presentations - how to avoid some common mistakes in academic presentations

011

Nathan Favero @nathanfavero.com · 23/08/2025

Equity is the value with most varied responses. Self-identified liberals indicate higher importance for equity (no surprise) but also representation.

Screenshot from article: Most notably, we find that the prioritization of equity exhibits the greatest variation across respondents from different backgrounds and social identities, suggesting that the importance of this value, more than others, remains contested and context-dependent. Similarly, representation—while generally held as a high priority by the public—is considered especially important by liberal-leaning individuals. Thus, when it comes to symbolic representation (see Keiser et al., 2022), pursuit of a representative bureaucracy may enjoy particularly widespread support among liberal constituencies. These examples suggest that public administrators should acknowledge citizens’ heterogeneous

110

Nathan Favero @nathanfavero.com · 23/08/2025

🧵I was surprised to find that people rate representation as more important than equity or efficiency, when asked to assess values for public services. From my latest work with @jaclynpiatak.bsky.social @colt-jensen.bsky.social & @minjungckim.bsky.social

Screenshot of Table 1 from article linked to later in thread. Summary stats (means). Equity 2.982. Efficiency 2.964. Effectiveness 3.199. Representation 3.136. N = 814; Response options: Not Important at all (0), Of little importance (1), Of average importance (2), Very Important (3), Absolutely Essential (4).

152

Nathan Favero @nathanfavero.com · 04/04/2025

Interested in structural equation models (SEM)? If you’re at #MPSA25, come see my talk tomorrow afternoon. Or better yet, go read the paper! (Link in next post)

Session info: Saturday, April 5. 3:20pm. Sandburg 3: 7th floor of Palmer

110

Nathan Favero @nathanfavero.com · 04/02/2025

Not bureaucrats running the show, AI running the show. I'm sure this will work great. www.nytimes.com/2025/02/03/u...

Screenshot from article: Mr. Musk’s allies now aim to inject artificial intelligence tools into government systems, using them to assess contracts and recommend cuts. On Monday, Thomas Shedd, a former Tesla engineer who has been tapped to lead a technology team at G.S.A., told some staff members that he hoped to put all federal contracts into a centralized system so they could be analyzed by artificial intelligence, three people familiar with the meeting said.

020

Nathan Favero @nathanfavero.com · 17/01/2025

Turns out that is also a great prompt for AI! Haha

AI overview after the search “what grade would a 4th grader get”. The overview says a 4th grader would get a “grade 4”

000

Nathan Favero @nathanfavero.com · 30/11/2024

As an aside, I often ponder the role of PA relative to other research communities in understanding government. This part of what I said is what might have gotten lost if it was only traditional "policy" or "politics" folks weighing in. Nothing groundbreaking but an important aspect to highlight IMHO

Screenshot of article text, which reads as follows: “Right now when we have the Department of Education, we have an entire agency where the leadership is thinking 100 percent about education, because that’s their whole mandate,” he said. “Whereas if they slice and dice this up and it goes to different agencies … now education is just going to be a very small part of the portfolio that that organization does, and I don’t think their leadership is going to have the same attentiveness to steer new initiatives.”

020

Nathan Favero @nathanfavero.com · 23/11/2024

Great to present my work with Weijie Wang on how NYC school expenditures change in response to performance signals at #2024APPAM

My coauthor Weijie wang presenting our conclusion slide

070

Nathan Favero @nathanfavero.com · 08/08/2024

Finally, I've posted my core lecture slides: github.com/favero-nate/... There are also videos for these slides available on YouTube: youtube.com/playlist?lis... I basically implement a "flipped classroom," but videos can also be used for supplemental study for particular topics.

screenshot showing pdf files for slides for 10 lectures, with a preview of Lecture 1: Graphing 1 variable. folder also shows sub-folders for data files and for Powerpoint conversions.

110

Nathan Favero @nathanfavero.com · 08/08/2024

3/ I also have posted various files that I've distributed to students over the years, including a class data collection project, Stata guides, and an Excel guide. github.com/favero-nate/...

showing files for a class data collection project (codebook, survey results template, homework instructions, policy memo assignment description, survey text, survey-based lab). a preview of the policy memo assignment description is also shown, indicating project should use data from survey that the class jointly collected. three drafts are due, and the memo should be structured with 3 section: (1) identifying a research question, (2) description of data & measure, and (3) analysis.

screenshot of excel guide showing how to create a histogram or box-and-whisker plot for continuous variables

screenshot of stata merging guide, showing the distinction between merge versus and append and how they can't always be mixed.

100

Nathan Favero @nathanfavero.com · 08/08/2024

2/ The labs are mostly Stata-based, but I've begun translating them into R (with 5 R scripts posted so far). I use a variety of datasets. Script files are generally annotated for students, with more detailed notes for some labs than others. github.com/favero-nate/...

screenshot of code from Stata lab showing how to use if statements and recode variables

a screenshot of R lab showing how to generate a binary indicator variable and then how to generate equivalent results (for a comparison of two means) using a t-test, anova, and regression

100

Nathan Favero @nathanfavero.com · 08/08/2024

🧵For anyone looking for classroom materials for intro stats/research methods, I've recently posted to GitHub a bunch of data labs and lecture materials that others are welcome to use.

Lab 1: basic graphing & labeling (data: college binge drinking)
Lab 2: basic data processing, descriptive stats, variable transformations, and more graphing (data: Texas school districts)
Lab 3a: bivariate data description, regression; assumes Lab 2 was completed (data: Texas school districts)
Lab 3b: basic regression examples, closely tracking Ch. 3 example (data: university GPAs)
Lab 3c: an alternative/extension to 3a, but less annotation so you may want to add your own; could also replace Lab 6 (data: U.S. states)
Lab 4: bivariate data description, confidence intervals (data: COVID-19 survey)
Lab 5: data processing; doesn't relate to Ch. 5, so could easily be another week (data: NYC schools)
Lab 6: simulations; additional simulations available with the Lecture 6 code/slides (https://github.com/favero-nate/minus-the-math/tree/main/lecture_slides)
Lab 8: comparing means through t-tests, ANOVA, & regression (data: U.S. states)
Lab 9: chi-squared

131

Nathan Favero @nathanfavero.com · 01/08/2024

I'll post some follow ups in the coming days, but hopefully some folks will find this useful. I put a lot of effort into it! minusthemath.com

200

Nathan Favero @nathanfavero.com · 01/08/2024

My main goal is for students to develop conceptual understanding and learn to interpret stats output (including graphs). I try to show underlying math when I think it helps students better conceptualize what's happening. But computers take care of the math, so it's secondary.

Table of contents:

Preface
1 Graphical Tools for Describing Variables
2 Statistics for Describing One Variable at a Time
3 Tools for Describing the Relationship Between Two Quantitative Variables
4 Estimation
5 Probability Distributions
6 Sampling Distributions
7 Hypothesis Testing
8 Comparing Means (How a Qualitative Variable Relates to a Quantitative One)
9 Comparing Groups (How Two Qualitative Variables Relate to One Another)
10 Causality and Experimental Designs
11 Models and Uncertainty
12 Regression with Qualitative Independent Variables
13 Regression with Qualitative Dependent Variables

100

Nathan Favero @nathanfavero.com · 01/08/2024

What makes this text different from all the other options? It's free! Also, I provide a small introduction to regression in week 3 so that students can get to work on quant readings and term papers.

1. It is a true introduction, not assuming any prior training in statistics.

2. I try to minimize use of math (beyond the very elementary) in initial explanations, instead focusing on conceptual description and interpretation.

3. I introduce regression very early on (Ch. 3) so that students (especially PhD students) can quickly get started on their term papers and better understand any quantitative articles they’re reading. The treatment of regression is further built out in the final chapters (11-13). Regression is, after all, the workhorse of applied statistics for the social sciences.

4. I skip a traditional treatment of probability theory because I don’t find traditional treatments to be very useful for students interested in applied statistics. Instead, I’ve written a brief chapter (11) on the logic and practice of building models that account for uncertainty.

5. There is a bit of Stata code in the final chapter, but otherwise all examples are presented apart from any software

100

Nathan Favero @nathanfavero.com · 25/07/2024

8/ Speaking of local revenue, which states are most mindful of perverse incentives with offsetting local revenue? Unfortunately, we have poor data on this Q. But from rough proxy we have for local tax rates, we find liberal governments do much more to reward local tax "effort"

110

Nathan Favero @nathanfavero.com · 25/07/2024

7/ Yet states don't appear very responsive to cost of living (COL) concerns! In appendix, we show that high-COL districts raise more local revenue; in liberal states they catch up with low-COL districts, but their revenue isn't higher

110

Nathan Favero @nathanfavero.com · 25/07/2024

6/ How does politics shape the ways states balance these priorities? Urban districts disproportionately benefit from the greater investment of liberal states in education.

110

Nathan Favero @nathanfavero.com · 25/07/2024

4/ We do see liberal state governments uniquely allocating extra funds to districts with large Black populations. Since we don't directly observe the funding formulas, we don't know exactly what mechanisms they use to accomplish this. But it is the end result of their policies

110

Nathan Favero @nathanfavero.com · 25/07/2024

3/ We find that (no surprise!) states spend more on k-12 education when their elected officials are more liberal. But surprisingly, liberal state governments don't exhibit a stronger preference for allocating funds toward high poverty districts

110

Nathan Favero @nathanfavero.com · 24/05/2024

Online seminar: Publishing in JPART from a Latin American perspective Stay tuned for more seminars across North America, Europe, Asia, and beyond! Register here: ruc-dk.zoom.us/webinar/regi... PAsky polisky

012

Nathan Favero @nathanfavero.com · 12/03/2024

3/ To evaluate, we asked participants to choose between hypothetical job arrangements, based on eight different job attributes, and respondents had to consider specific details for each job in making their choice.

100

Nathan Favero @nathanfavero.com · 31/01/2024

Thank you Texas Tech poli sci for the opportunity to visit and present my coauthored work (with Amanda Rutherford) on bureaucratic decision making among school-based law enforcement. I very much enjoyed learning about new research and appreciate the constructive feedback on our paper!

040

Nathan Favero @nathanfavero.com · 01/12/2023

What does it take to get expelled from Congress? Apparently, financially defrauding members of your own caucus will get you over the line. But even then, the party leadership will vote to keep you. x.com/JakeSherman/...

000

Nathan Favero @nathanfavero.com · 01/10/2023

Trying to experiment with learning something new. I painted this, mostly over the summer :)

Oil painting of a forest of trees changing colors in the fall

010