STA 9750 Mini-Project #03: Who Goes There? US Internal Migration and Implications for Congressional Reapportionment

Due Dates

Released to Students: 2026-04-02
Initial Submission: 2026-04-24 11:59pm ET on GitHub and Brightspace
Peer Feedback:
- Peer Feedback Assigned: 2026-04-27 on GitHub
- Peer Feedback Due: 2026-05-03 11:59pm ET on GitHub

Estimated Time to Complete: 13-15 Hours

Estimated Time for Peer Feedback: 1 Hour

Introduction

Welcome to Mini-Project #03! In this mini-project, we will explore patterns of internal migration within the United States with an eye towards forecasting the size of state congressional delegations for the 2032 midterm elections. Our analysis will rely on Migration Flows data inferred from the American Community Survey (ACS), the long-running premier national survey conducted by the US Census Bureau.¹ Unlike the decadal census, which occurs once every ten years, the ACS is constantly collecting new data, giving a up-to-date picture of the American public. ACS data is released across a range of granularities, from national averages to small census-block level estimates. High-level estimates, e.g., those for the country as a whole or individual states, are published on a year-by-year basis, in a data product often called “ACS-1” since they are derived from samples taken in a one-year window. In order to preserve respondent privacy and ensure statistical reliability at finer scales, ACS estimates for smaller regions are published based on a rolling five-year window: these estimates are known as “ACS-5”.² Larger regions are also available from ACS-5 to avoid alignment issues, but these are a bit less useful to us.

For this project, we will primarily make use of state-level ACS-1 flows for our demographic projections, but we will use metro-level ACS-5 to identify the primary endpoints of those flows. We will combine these with other ACS data via the tidycensus package to (roughly) predict state populations in 2030. These flows are estimated by asking respondents where they lived in the prior year, providing detailed data on population movement within the United States.

Student Responsibilities

Recall our basic analytic workflow and table of student responsibilities:

Data Ingest and Cleaning: Given a data source, read it into R and transform it to a reasonably useful and standardized (‘tidy’) format.
Data Combination and Alignment: Combine multiple data sources to enable insights not possible from a single source.
Descriptive Statistical Analysis: Take a data table and compute informative summary statistics from both the entire population and relevant subgroups
Data Visualization: Generate insightful data visualizations to spur insights not attainable from point statistics
Inferential Statistical Analysis and Modeling: Develop relevant predictive models and statistical analyses to generate insights about the underlying population and not simply the data at hand.

In this course, our primary focus is on the first four stages: you will take other courses that develop analytical and modeling techniques for a variety of data types. As we progress through the course, you will eventually be responsible for the first four steps. Specifically, you are responsible for the following stages of each mini-project:

Students’ Responsibilities in Mini-Project Analyses
	Ingest and Cleaning	Combination and Alignment	Descriptive Statistical Analysis	Visualization
Mini-Project #01			✓
Mini-Project #02		✓	✓	½
Mini-Project #03	½	✓	✓	✓
Mini-Project #04	✓	✓	✓	✓

In this project, I am no longer providing code to download and read the necessary data files. The data files I have selected for this mini-project are relatively easy to work with and should not provide a significant challenge, particularly after our in-class discussion of Data Import.

Rubric

STA 9750 Mini-Projects are evaluated using peer grading with meta-review by the course staff (GTAs and the instructor). The following basic rubric will be used for all mini-projects:

Course Element	Excellent (9-10)	Great (7-8)	Good (5-6)	Adequate (3-4)	Needs Improvement (1-2)
Written Communication	Report is very well-written and flows naturally. Motivation for key steps is clearly explained to reader without excessive detail. Key findings are highlighted and appropriately given sufficient context, including reference to related work where appropriate.	Report has no grammatical or writing issues.³ Writing is accessible and flows naturally. Key findings are highlighted and clearly explained, but lack suitable motivation and context.	Report has no grammatical or writing issues. Key findings are present but insufficiently highlighted or unclearly explained.	Writing is intelligible, but has some grammatical errors. Key findings are difficult to discern.	Report exhibits significant weakness in written communication. Key points are nearly impossible to identify.
Project Skeleton	Code completes all instructor-provided tasks correctly. Responses to open-ended tasks are especially insightful and creative.	Code completes all instructor-provided tasks satisfactorily. Responses to open-ended tasks are insightful, creative, and do not have any minor flaws.	Response to one instructor provided task is skipped, incorrect, or otherwise incomplete. Responses to open-ended tasks are solid and without serious flaws.	Responses to two instructor provided tasks are skipped, incorrect, or otherwise incomplete. Responses to open-ended tasks are acceptable, but have at least one serious flaw.	Response to three or more instructor provided tasks are skipped, incorrect, or otherwise incomplete. Responses to open-ended tasks are seriously lacking.
Tables & Document Presentation	Tables go beyond standard publication-quality formatting, using advanced features like color formatting, interactivity, or embedded visualization.	Tables are well-formatted, with publication-quality selection of data to present, formatting of table contents (e.g., significant figures) and column names.	Tables are well-formatted, but still have room for improvement in one of these categories: subsetting and selection of data to present, formatting of table contents (e.g., significant figures), column names.	Tables lack significant ‘polish’ and need improvement in substance (filtering and down-selecting of presented data) or style. Document is difficult to read due to distracting formatting choices.	Unfiltered ‘data dump’ instead of curated table. Document is illegible at points.
Data Visualization	Figures go beyond standard publication-quality formatting, using advanced features like animation, interactivity, or advanced plot types implemented in `ggplot2` extension packages.	Figures are ‘publication-quality,’ with suitable axis labels, well-chosen structure, attractive color schemes, titles, subtitles, and captions, etc.	Figures are above ‘exploratory-quality’ and reflect a moderate degree of polish, but do not reach full ‘publication-quality’ in one-to-two ways.	Figures are above ‘exploratory-quality’ and reflect a moderate degree of polish, but do not reach full ‘publication-quality’ in three or more distinct ways.	Figures are suitable to support claims made, but are ‘exploratory-quality,’ reflecting zero-to-minimal effort to customize and ‘polish’ beyond `ggplot2` defaults.
Exploratory Data Analysis	Deep and ‘story-telling’ EDA identifying non-obvious patterns that are then used to drive further analysis in support of the project. All patterns and irregularities are noted and well characterized, demonstrating mastery and deep understanding of all data sets used.	Meaningful ‘story-telling’ EDA identifying non-obvious patterns in the data. Major and pinor patterns and irregularities are noted and well characterized at a level sufficient to achieve the goals of the analysis. EDA demonstrates clear understanding of all data sets used.	Extensive EDA that thoroughly explores the data, but lacks narrative and does not deliver a meaningful ‘story’ to the reader. Obvious patterns or irregularities noted and well characterized, but more subtle structure may be overlooked or not fully discussed. EDA demonstrates competence and basic understanding of the data sets used.	Solid EDA that identifies major structure to the data, but does not fully explore all relevant structure. Obvious patterns or irregularities ignored or missed. EDA demonstrates familiarity with high-level structure of the data sets used.	Minimal EDA, covering only standard summary statistics, and providing limited insight into data patterns or irregularities. EDA fails to demonstrate familiarity with even the most basic properties of the data sets being analyzed.
Code Quality	Code is (near) flawless. Intent is clear throughout and all code is efficient, clear, and fully idiomatic. Code passes all `styler` and `lintr` type analyses without issue.	Comments give context and structure of the analysis, not simply defining functions used in a particular line. Intent is clear throughout, but code can be minorly improved in certain sections.	Code has well-chosen variable names and basic comments. Intent is generally clear, though some sections may be messy and code may have serious clarity or efficiency issues.	Code executes properly, but is difficult to read. Intent is generally clear and code is messy or inefficient.	Code fails to execute properly.
Data Preparation	Data import is fully-automated and efficient, taking care to only download from web-sources if not available locally. All data cleaning steps are fully-automated and robustly implemented, yielding a clean data set that can be widely used.	Data is imported and prepared effectively, in an automated fashion with minimal hard-coding of URLs and file paths. Data cleaning is fully-automated and sufficient to address all issues relevant to the analysis at plan.	Data is imported and prepared effectively, though source and destination file names are hard-coded. Data cleaning is rather manual and hard-codes most transformations.	Data is imported in a manner likely to have errors. Data cleaning is insufficient and fails to address clear problems.	Data is hard-coded and not imported from an external source.
Analysis and Findings	Analysis demonstrates uncommon insight and quality, providing unexpected and subtle insights.	Analysis is clear and convincing, leaving essentially no doubts about correctness.	Analysis clearly appears to be correct and passes the “sniff test” for all findings, but a detailed review notes some questions remain unanswered.	Analysis is not clearly flawed at any point and is likely to be within the right order of magnitude for all findings.	Analysis is clearly incorrect in at least one major finding, reporting clearly implausible results that are likely off by an order of magnitude or more.

Note that the “Excellent” category for most elements applies only to truly exceptional “above-and-beyond” work. Most student submissions will likely fall in the “Good” to “Great” range.

At this point, you are responsible for the ‘Data Preparation’ portion of the project, but I am still providing a set of basic EDA activities. Accordingly, reports completing all tasks described under Data Integration and Exploration below should receive an automatic 10/10 for the ‘Exploratory Data Analysis’ rubric element.

Taken together, you are only really responsible for these portions of the rubric in this assignment:

Written Communication
Project Skeleton
Tables & Document Presentation
Data Visualization
Code Quality
Data Preparation
Analysis and Findings

Reports completing all key steps outlined below essentially start with 10 free points.

For this mini-project, no more than 10 total points of extra credit can be be awarded. Opportunities for extra credit exist for students who go above and beyond the instructor-provided scaffolding. Specific opportunities for extra credit can be found below.

Students pursuing careers in data analytics are strongly encouraged to go beyond the strict ambit of the mini-projects to

further refine their skills;
learn additional techniques that can be used in the final course project; and
develop a more impressive professional portfolio.

Because students are encouraged to use STA 9750 mini-projects as the basis for a professional portfolio, the basic skeleton of each project will be released under a fairly permissive usage license. Take advantage of it!

Submission Instructions

After completing the analysis, write up your findings, showing all of your code, using a dynamic quarto document and post it to your course repository. The qmd file should be named mp03.qmd (lower case!) so the rendered document can be found at docs/mp03.html in the student’s repository and will be served at the URL:⁴

https://YOUR_GITHUB_ID.github.io/STA9750-2026-SPRING/mp03.html

You can use the helper function mp_start available at in the Course Helper Functions to create a file with the appropriate name and some meta-data already included. Do so by running the following command at the R Console:

source("https://michael-weylandt.com/STA9750/load_helpers.R"); mp_start(N=03)

After completing this mini-project, upload your rendered output and necessary ancillary files to GitHub to make sure your site works. The mp_submission_ready function in the Course Helper Functions can perform some of these checks automatically. You can run this function by running the following commands at the R Console:

source("https://michael-weylandt.com/STA9750/load_helpers.R"); mp_submission_ready(N=03)

Once you confirm this website works (substituting YOUR_GITHUB_ID for the actual GitHub username provided to the professor in MP#00 of course), open a GitHub issue on the instructor’s repository to submit your completed work.

The easiest way to do so is by use of the mp_submission_create function in the Course Helper Functions, which can be used by running the following command at the R Console:

source("https://michael-weylandt.com/STA9750/load_helpers.R"); mp_submission_create(N=03)

Alternatively, if you wish to submit manually, open a new issue at

https://github.com/michaelweylandt/STA9750-2026-SPRING/issues/new .

Title the issue STA 9750 YOUR_GITHUB_ID MiniProject #03 and fill in the following text for the issue:

Hi @michaelweylandt!

I've uploaded my work for MiniProject #**03** - check it out!

<https://YOUR_GITHUB_ID.github.io/STA9750-2026-SPRING/mp03.html>

At various points before and after the submission deadline, the instructor will run some automated checks to ensure your submission has all necessary components. Please respond to any issues raised in a timely fashion as failing to address them may lead to a lower set of scores when graded.

Additionally, a PDF export of this report should be submitted on Brightspace. To create a PDF from the uploaded report, simply use your browser’s ‘Print to PDF’ functionality.

NB: The analysis outline below specifies key tasks you need to perform within your write up. Your peer evaluators will check that you complete these. You are encouraged to do extra analysis, but the bolded Tasks are mandatory.

NB: Your final submission should look like a report, not simply a list of facts answering questions. Add introductions, conclusions, and your own commentary. You should be practicing both raw coding skills and written communication in all mini-projects. There is little value in data points stated without context or motivation.

Mini-Project #03: Who Goes There? US Internal Migration and Implications for Congressional Reapportionment

Longest Mini-Project of the Semester

NB: This is the longest mini-project of the semester. Mini-Project #04 is significantly shorter to give you sufficient time to complete your course project successfully. Unlike the previous two mini-projects, you are responsible for data import and cleaning in this project. Though this isn’t glamorous, this is the aspect of data analysis that typically requires the most time and effort and you should plan accordingly.

Data Acquisition

When downloading data from a web-based resource, it is important to be a polite and responsible user of that resource. Specifically, we want to avoid excessively and repeatedly re-downloading a file if it is not changing. Running a public data source is costly and downloading a file 1,000 times when you only need it once is simply abusing the good will of others.

We can ensure responsible usage by adopting a practice of file caching:

Create a “data directory” where data files relevant to the project can be stored
When attempting to load a data resource, first see if the relevant file is in the data directory:
- If the file is in the data directory, simply read and return the contents of that file to the user
- If the file is not in the data directory, download it into the data directory and save a permanent copy for later use before reading and returning the contents of the file to the user.

This process of “saving” an intermediate result (data file) to avoid an expensive step (downloading) is known as caching. Here, in addition to reducing load on the data server, caching will also make your analysis faster (as you can avoid download times) and allow you to work even when you don’t have reliable internet access.

Review the data download code I provided you in MP#01 and MP#02 to see how I implemented caching there. You should adopt similar patterns for this assignment.

We will use three different data sources for this assignment:

ACS-1 State-to-State Migration Flows
ACS-5 Metro-to-Metro Migration Flows
ACS-1 Baseline State Population Data

The last of these will actually be the simplest, so we will start there.

State Population - Package Usage

In order to make our predictions about how many residents each state will have in 2030, we need to know, as a baseline, how many residents that state has presently.⁵ Additionally, we want to know how many children are being born in that state. While fertility is a complex and deeply-studied topic in demography, we will use a very simple model below that holds the natural growth rate constant across the whole country: that is, we assume families have (on average) the same number of kids in each state, life expectancies are the same in every state, and the age breakdown is the same in every state. See Extra Credit #03 below for opportunities to build a more realistic analysis.

Census data is very high-quality, but accessing it can be somewhat tricky The census reports data in the form of “tables” which have systematic, but difficult to interpret identifiers. We can access these using the get_acs() function from the tidycensus package, which works like this:

library(tidycensus)
library(gt)

ny_income <- get_acs(
    c(median_household_income="B19013_001"), 
    state="NY", 
    geography="county", 
    year=2018, 
    survey="acs1"
)

In this case, B19013 is the table “Median Household Income in the Past 12 Months (in YYYY Inflation-Adjusted Dollars)” and the suffix “001” indicates that we want the first row of the table. Here, we asked for results from the 2018 ACS-1 county-level estimates for New York State. After some basic formatting, we get the following results, which can be verified against the Census Data Explorer website:

GEOID	NAME	Median Household Income
Highest Median Household Income Counties
36059	Nassau County, New York	$116,304
36079	Putnam County, New York	$102,525
36103	Suffolk County, New York	$100,468
36119	Westchester County, New York	$94,811
36087	Rockland County, New York	$89,812
Lowest Median Household Income Counties
36007	Broome County, New York	$51,125
36089	St. Lawrence County, New York	$49,681
36009	Cattaraugus County, New York	$48,179
36013	Chautauqua County, New York	$45,479
36005	Bronx County, New York	$38,467

Here, the GEOID, also known as the FIPS code, is a unique identifier used for different administrative units with the US (so states, counties, cities, etc all have distinct GEOIDs). Note that I’m only showing the top and bottom 5 counties here for brevity.

Note also the use of a named vector as the first argument to get_acs() lets us control the name the result is given in R. get_acs() has two additional features that will be useful:

the ability to automatically download shapefile information associated with each result;
the ability to cache results automatically.

You will want to use both of these.

Task 1: Data Acquisition - State Population and Birth Information

Use the get_acs() function from the tidycensus package to download state-level estimates of total population (B01003_001) for the years 2015 to 2024. Additionally, use get_acs() to automatically download state shape information and to appropriately cache your results locally.

Read the help page for get_acs() to identify the caching and shapefile download arguments.

You should not need to provide a census API key at this step. If you receive an error (not just a warning or message) indicating that an API key is required, please contact the instructor.

Once you have made your request, construct a tidy data frame with the following columns:

Total population
State Name or Abbreviation
Year

State-to-State Migration Flows - File Download

Since our primary focus is in predicting future state populations, we will principally work with state-to-state migration flows. The Census Bureau provides Excel files with these data here. While convenient for manual use, this site actually has several of the challenges associated with complex file-based data exchange:

The URLs of the individual files are not consistent
The formatting of files changes over time

Unfortunately, there is rarely a ‘simple’ way of dealing with these problems and we will instead need to write code to handle different cases separately. When dealing with scenarios like this, a good first step is to decide what you want your final result to look like and to work backwards from there. While census provides us with quite a bit, the actual data we need from this data set is reasonably straightforward:

state_current	state_1y	population	year_current	year_1y
AL	AL	4,960,560	2024	2023
AL	AK	387	2024	2023
AL	AL	4,920,737	2023	2022
AL	AK	780	2023	2022

Look at the 2024 and 2023 files and confirm for yourself how I computed these numbers.

Earlier files in this series have different formats and, while they could be helpful for our analysis, we are only going to use 2023 and 2024 to keep the scope of this project tractable.

Task 2: Data Acquisition - State-to-State Flows 2024

Write a function state_to_state_migration_2024() which extracts state flows from the 2024 migration file and formats them as shown above. Your function should, at a minimum,

Create the directory data/mp03 if it does not already exist. (See previous MPs for examples of how this is done.)
Check if the 2024 migration file is present in the data/mp03 directory and, if it is not, download it using the download.file function.
Read the file into R using the read_excel() function from the readxl package. (This package has two reader functions-read_xls() and read_xlsx()-and the wrapper function, read_excel() which will determine which is appropriate for a given file. Unless you have a specific reason not to use it, read_excel() should be your default import function.)

You actually will need to use this function twice, varying the sheet argument:
1. One time, set sheet="Table" to get the actual state-to-state flows from the first sheet of the xlsx file.
2. On a second time, use sheet="Supplemental - Current Res" so you can get the “stationary” population.
In these, you may also find the skip function or the range argument to import only those rows and columns that will be useful to your analysis.

Hint: Often, Excel’s “merged cells” are used as headers in complex data like this, but these are not easily brought into R. It is sometimes useful to change the col_names argument to provide headers yourself rather than trying to read them from the data set.
Manipulate the data into an appropriate format:
1. For this data, the "Table" sheet requires fairly little work, but you will want to remove the “non-moving” rows as these have X values instead of actual numbers.
Additionally, there are some data irregularities in the Table sheet you will need to handle. For state-to-state pairs with very little migration, a value of "N" is given to indicate that there are not enough values to preserve privacy. Use the replace_values() function from the dplyr package⁶ to replace "N" with "0" and then use the as.integer() function to ensure you have numeric values.

Finally, add the two year columns and make sure your data is in the format given above.
1. The Supplemental sheet only requires a bit more work. As before, use sheet, skip, and col_names to get only what you need. A useful trick here is to use select() to rename columns by position: e.g.,
```
library(dplyr)
penguins |> select(Species = 1, bill_length=3) |> head(5)
```
```
  Species bill_length
1  Adelie        39.1
2  Adelie        39.5
3  Adelie        40.3
4  Adelie          NA
5  Adelie        36.7
```
Add together two columns to get a “stayed within the state” total and manipulate everything else to ensure your data is in the format prescribed above.
Use the rbind_rows() function to combine the results from each sheet together into a single 2,500 ($=50 \times 50$)⁷ row data frame. Note that the rbind_rows() function combines two data sets rowwise (i.e., stacks them vertically) so you will need to make sure your data has consistent column names and types before calling it.

This analysis will be a bit tricky to figure out, but it shouldn’t require that much code when you get all of the pieces in place. I did this in about 40 lines of reasonably clean code, so if you find yourself using more than 75, step back and reconsider.

As you are cleaning data like this, you should constantly be printing output and comparing it to what you see when you look at the “raw” spreadsheets (keeping a copy open in Excel). This data is not quite small enough that you can check each cell individually, but you can definitely “spot check” data as you work.

Once you have read in the 2024 data file, it is time to read the 2023 data file. This is a bit trickier, but we can make it work.

Task 3: Data Acquisition - State-to-State Flows 2023

Write a function state_to_state_migration_2023() which extracts state flows from the 2023 migration file and formats them as shown above. Your function should, at a minimum,

Create the directory data/mp03 if it does not already exist. (See previous MPs for examples of how this is done.)
Check if the 2023 migration file is present in the data/mp03 directory and, if it is not, download it using the download.file() function.
Read the file into R using the read_excel() function from the readxl package. In this scenario, you will not need to supply the sheet argument to read_excel() since everything we need is on the "Table" sheet.

Given the complexity of this data, I recommend using the range argument to get only those cells you need. I recommend using two different calls to read_excel() to get this right:
1. First, read only the first 6 columns of the sheet and add together two columns to get the population that stayed within the state. Format your results into the five columns specified above.
2. Secondly, read in the bulk (columns J to DG) of the table to get the actual flows that we want to extract. The formatting of this sheet makes it difficult to get the column names, so it will be useful to provide the col_names argument directly. Thankfully, R gives us a built-in data vector with the names of each state (state.name) that can be used here, like this:
```
library(readxl)
states_with_dc <- c(state.name[1:8], 
                    "District of Columbia", 
                    state.name[9:50])

read_excel(..., # Fill in these dots
           col_names=vctrs::vec_interleave(states_with_dc, NA)) |>
    select(any_of(states_with_dc)) |>
    mutate(state_to=states_with_dc) 
```
After this second import, you will want to pivot_longer() this data into a format where each row is a pair of states. Finally, prepare this data for use by:
1. Dropping any rows with DC
2. Dropping any “within” rows (e.g., (Texas, Texas)) since that value is N/A and we get that data elsewhere.
3. Converting the numeric column to integer.
Combine the results of your two imports using rbind_rows() and confirm that you have a 2,500 row 5 column data set as before.⁸

Task 4: Data Acquisition - Combined State-to-State Flows

Write a function state_to_state_migration which invokes the two functions you wrote above and uses rbind_rows() to create a 5,000 row table of state to state migration over the previous two years. Confirm that your results match the example rows shown above.

Do Not Use tidycensus for Migration Flows

It is possible to get migration flows data using the tidycensus package: see documentation here.

Do not use tidycensus to access this data. You must use generic tools like the download.file() function or the httr2 (not httr) package.

The point of this portion is to practice getting data for which an easy wrapper is not available. Use of tidycensus to access migration data will receive an automatic 0 for the “Data Preparation” portion of this mini-project and significant penalization on the “Project Skeleton” portion.

You may however use tidycensus to cross-check your results: i.e., to see if the data you acquire directly matches values given by tidycensus.

Metro-to-Metro Migration Flows - API Usage

While we can complete our analysis using the state-level data acquired above, understanding movement between metro areas (cities) will be useful in gaining a deeper understanding of migration patterns (e.g., many people move from other states to NYS, but almost all of those move to NYC and suburbs, not to Albany). For this step, we will use a census API rather than reading formatted data files. As you will see, the API is typically much easier to parse in R than working directly with data files.

The latest metro-to-metro migration flows available from the API are those from the 2016-2020 ACS-5 window, and can be found here. Review that link and look at the example API calls.

Task 5: Data Acquisition - Metro-to-Metro Migration Flows

Write a function metro_to_metro_migration to get city-level migration flows from the Census API. Your function should, at a minimum,

Create the directory data/mp03 if it does not already exist. (See previous MPs for examples of how this is done.)
Check if the 2023 migration file is present in the data/mp03 directory and, if it is not, make an API request and save the resulting JSON file using the download.file() function.

Determine the URL of the API request using the example API calls linked above. Your request should request data at the metropolitan/micropolitan statistical area level for the following fields:
- MOVEDIN: The number of people who moved into FULL1 from FULL2
- MOVEDOUT The number of people who moved into FULL2 from FULL1
- FULL1_NAME: An MSA Name
- FULL2_NAME: An MSA Name
Read the saved JSON file using the read_json() function from the jsonlite library.
The census API response isn’t formatted quite properly, so read_json() will interpret the first row of the response as data instead of column names. This code snippet can be used to move the first row into column names and do some basic type clean-up:
```
library(tidyverse)
library(jsonlite)
move_row_to_colnames <- function(X, row = 1){
    X <- as_tibble(X)
    Xrow <- X[row,]

    X <- X[-row,]

    colnames(X) <- Xrow

    readr::type_convert(X)
}
```
Feel free to use it in your data preparation.
Simplify column names and drop unneeded columns.

Use the following mutate command to pull out the (principal) state for a metro area. For metro areas contained within a single state, e.g., Los Angeles/Long Beach, this will just pull that one state (CA); for metro areas spread across multiple states (e.g., “New York-Newark-Jersey City, NY-NJ-PA Metro Area”) this will pull the first which I’m assuming is the largest:

pull_state_from_metro <- function(metro_name){
  str_extract(metro_name, ".*, (\\S{2})[-[:alpha:]]* Metro Area", group=1)
}

You can use it as follows:

example_metros <- c(
  "Los Angeles-Long Beach-Anaheim, CA Metro Area", 
  "Dallas-Fort Worth-Arlington, TX Metro Area", 
  "Riverside-San Bernardino-Ontario, CA Metro Area", 
  "Atlanta-Sandy Springs-Alpharetta, GA Metro Area",
  "Chicago-Naperville-Elgin, IL-IN-WI Metro Area", 
  "New York-Newark-Jersey City, NY-NJ-PA Metro Area", 
  "Outside Metro Area within U.S. or Puerto Rico", 
  "Houston-The Woodlands-Sugar Land, TX Metro Area", 
  "Washington-Arlington-Alexandria, DC-VA-MD-WV Metro Area", 
  "San Jose-Sunnyvale-Santa Clara, CA Metro Area"
)

data.frame(metro = example_metros) |> 
  mutate(metro_state = pull_state_from_metro(metro)) |> 
  gt(id="tbl_metro_states") |> 
  cols_label(metro="Metropolitan Statistical Area", 
             metro_state="Primary State")

Metropolitan Statistical Area	Primary State
Los Angeles-Long Beach-Anaheim, CA Metro Area	CA
Dallas-Fort Worth-Arlington, TX Metro Area	TX
Riverside-San Bernardino-Ontario, CA Metro Area	CA
Atlanta-Sandy Springs-Alpharetta, GA Metro Area	GA
Chicago-Naperville-Elgin, IL-IN-WI Metro Area	IL
New York-Newark-Jersey City, NY-NJ-PA Metro Area	NY
Outside Metro Area within U.S. or Puerto Rico	NA
Houston-The Woodlands-Sugar Land, TX Metro Area	TX
Washington-Arlington-Alexandria, DC-VA-MD-WV Metro Area	DC
San Jose-Sunnyvale-Santa Clara, CA Metro Area	CA

Do Not Use tidycensus for Migration Flows

It is possible to get migration flows data using the tidycensus package: see documentation here.

Do not use tidycensus to access this data. You must use generic tools like the download.file() function or the httr2 (not httr) package.

You may however use tidycensus to cross-check your results: i.e., to see if the data you acquire directly matches values given by tidycensus.

Data Cleaning and Preparation

You should have completed your data cleaning as part of the import steps above. While it is generally good to “spot check” data for accuracy, in this case we can trust the Census Bureau and move immediately to EDA, only performing additional checks if issues are identified.

Data Integration and Initial Exploration

Task 6: Exploratory Data Analysis

Answer the following questions to perform your Exploratory Data Analysis of the various data sets used in this project. For each question, answer using inline values, a table, or a graphic as you feel is most effective. Note that you will need to determine which data are best suited to answer each question: some questions might require multiple data sets to answer fully.

Which states have had the highest net population growth rates over the past decade?
If you meet someone who moved to New York State in the last year, what are the most likely states they moved from? If one of your friends announces they are moving out of the state, where are they most likely to move?
If you meet someone who moved to New York City in the last year, what are the most likely metro areas they moved from? If one of your friends announces they are moving out of the city, where are they most likely to move?
Which states have had the highest amounts of in-migration and out-migration? Furthermore, which states have had the highest amount of net migration?

To get total migration in a direction, add up the total migration in that direction: e.g., the total in-migration to Alabama is the sum of people who live in Alabama in 2024 but lived elsewhere in 2023 (this is a sum over 49 other states).

Net migration is simply total in-migration minus total out-migration.
Which metro areas have had the highest amounts of total in-migration and out-migration? Furthermore, which metro areas have had the highest amount of net migration?
Which states have the highest fraction of residents who lived there last year? (That is, which states have the lowest proportional total in-migration?)
Which state has the highest fraction of its population growth attributable to net internal migration? That is, determine the total population growth of each state and see what fraction of it corresponds to net migration (as opposed to deaths, births, or international migration).

These final questions are specific to the state you are focusing on (see final deliverable below for more).

Which state had the highest amount of migration into your state? Of the people who moved out of your state, what was the most popular destination state?
What is the largest metropolitan area in your state? (You can answer this just using background knowledge; you do not have to use Census data here.)

Which metro area, not located in your state, has the highest amount of migration into your state’s largest metro? To which metro area outside of your state do the most residents of your largest metro area move?
Are there any metro areas that have a particularly high connection to your state’s largest metro? (E.g., NYCers were over 15% of people moving into the Miami Metro area.)

Final Deliverable: Ensuring State Political Power in 2032

You are a policy analyst working for the governor of a state (possibly, but not necessarily, New York State). In light of recent news stories around redistricting, the governor has asked you to look ahead to the 2030 decennial census and the corresponding 2032 apportionment and redistricting cycle.

Under the US Constitution (Article 1, Section 2, Clause 3), seats in the House of Representatives are allocated to the states on the basis of state population. In particular,

Representatives and direct Taxes shall be apportioned among the several States which may be included within this Union, according to their respective Numbers […]. The actual Enumeration shall be made […] within every subsequent Term of ten Years, in such Manner as they shall by Law direct. The Number of Representatives shall not exceed one for every thirty Thousand, but each State shall have at Least one Representative.

As such, after each decennial (every 10 year) census, the Congress had to pass a new bill determining exactly how many seats each state was granted.⁹ Because this process became quite contentious, Congress passed the Permanent Apportionment Act of 1929 which fixed the size of the House of Representatives at 435.¹⁰ Given this fixed cap, states whose population is growing faster than others are predicted to gain more congressional seats, while states whose population is growing more slowly (or even shrinking) are expected to lose seats. Accordingly, estimates of state populations as of 2030 are quite politically interesting.¹¹

You have been tasked with i) estimating your state’s population in 2030; ii) estimating your state’s congressional apportionment as of 2032, based on this population; and iii) helping to design a marketing campaign to encourage more migration to your state.

Let’s begin by constructing estimates of 2030 populations. We can use a very simple population forecast model:

\[P_{i}^{(t+1)} = P_i^{(t)} (1 + g_i^{(t)}) + \sum_{j} m_{j \to i} \sqrt{P^{(t)}_iP^{(t)}_j}\]

where

$P_i^{(t)}$ is the population of region $i$ in year $t$
$P_i^{(t+1)}$ is the population of region $i$ in year $t+1$
$g_i$ is the natural population growth rate (births and deaths) within region $i$
$m_{i \to j}$ is the migration rate (as a percent of population) from region $i$ to region $j$

This model is rather simple - notably, it ignores age and cohort effects - but it’s good enough for us to get started. To simplify this further, we’re going to assumes that the natural population growth rate is constant across states ($g_i = \overline{g}$ for all $i$). While this expression looks a bit complex, it is linear in both of the unknown parameters $g$ and $m_{i \to j}$. We can estimate this in two steps:

Compute $\overline{g}$ for the country as a whole as: \[\overline{g} = \frac{P^{(t+1)} - M^{(t)}}{P^{(t)}} - 1 = \frac{P^{(t+1)} - P^{(t)} - M^{(t)}}{P^{(t)}}\] where $M^{(t)}$ is the total net migration (all sources) into region the country between year $t$ and year $t+1$.
Compute $m_{i \to j}$ for each state pair as

So, for example, if we have three regions (A, B, C) with the following population data:

	Total Population (Year 2)	Total Population (Year 1)	Location in Year 1
	Total Population (Year 2)	Total Population (Year 1)	Previously in Region A	Previously in Region B	Previously in Region C	Previously Living Abroad
A	1100	1050	1000	0	0	50
B	1500	1295	25	1250	0	20
C	2000	2050	25	0	1975	50
Total (Nationwide)	4600	4395	1050	1250	1975	120

From these, we compute the following key statistics:

Total Population in Year 2: $P^{(2)} = 4600$
Total Population in Year 1: $P^{(1)} = 4275$
In-Migration: $120$¹²

From these, we see that

\[\overline{g} = \frac{4600 - 4275 - 120}{4275} \approx 4.8\%\]

so this population is growing at a very healthy clip. (For reference, the actual US growth rate has never surpassed 1% in the past 25 years. See this Census news release.)

Next, we want to determine the migration factors $m_{j \to i}$ for each pair of regions:

\[m_{j \to i} = M_{j \to i}^{(t)} / \sqrt{P_i^{(t)}P_j^{(t)}}\]

This model assumes that the number of people who move from region $j$ to region $i$ depends on both the population in region $j$ and region $i$, so as $j$ becomes larger it has more out-migration and as $i$ becomes bigger, it has more in-migration. This is not terribly realistic, but it will do for our purposes.

Applying this calculation to our data, we get the following migration factors:

$m_{B \to A} = m_{C \to A} = m_{C \to B} = m_{B \to C} = 0$
$m_{A \to B} = 25 / \sqrt{1000 * 25} = 0.16$
$m_{A \to C} = 25 / \sqrt{1000 * 25} = 0.16$

From here, we can uses these to compute population projections for Year 3:

\[P_{A}^{(3)} = P_{A}^{(2)}(1 + \overline{g}) + \sum_{j \in \{B, C\}}m_{j \to A}\sqrt{P_{A}^{(t)}P_j^{(2)}} = 1100 * (1 + 0.048) + \sum_{j \in \{B, C\}} 0 * \sqrt{1100 * P_j^{(2)}} = 1153\]

\[P_{B}^{(3)} = P_{B}^{(2)}(1 + \overline{g}) + \sum_{j \in \{A, C\}}m_{j \to B}\sqrt{P_{B}^{(t)}P_j^{(2)}} = 1500 * (1 + 0.048) + 0.16 * \sqrt{1500 * 1100} = 1778\]

\[P_{C}^{(3)} = P_{C}^{(2)}(1 + \overline{g}) + \sum_{j \in \{A, B\}}m_{j \to C}\sqrt{P_{C}^{(t)}P_j^{(2)}} = 2000 * (1 + 0.048) + 0.16 * \sqrt{2000 * 1100} = 2333\] If we wanted to predict further ahead to Year 4, we would simply repeat this calculation using $P^{(3)}$ as inputs, holding $\overline{g}$ and $m_{j \to i}$ constant.

Task 7: Population Projections

Fit the above model to the US and use it to make population predictions for 2030. Because we have two data sets (migration flows 2023 and 2024), let’s fit our parameters twice and average them to improve the accuracy of our model.

First, for 2024, use the migration flows data to see how many people were living outside the US in the previous year (2023) to determine $M^{(2023)}$.
Using the national population $\overline{P}^{(2023)}$ and $\overline{P}^{(2024)}$, as well as $M^{(2023)}$, determine $\overline{g}^{(2023)}$.
Using the migration flows data and the per-state populations in 2023 and 2024, estimate $m_{j \to i}^{(2023)}$ for 2,450 ($=50 \times (50 - 1)$) state pairs.
Repeat Steps 1-3 for 2022 to 2023.
Get parameter estimates by averaging over each of the previous two years, e.g., $\overline{g} = \frac{\overline{g}^{(2023)} + \overline{g}^{(2024)}}{2}$.
Using your two year estimates, predict per-state populations for 2025, 2026, 2027, 2028, 2029, and finally 2030.

A word of warning: you will compute approximately 7,650 numbers in this calculation. ($2,451 * 3$ parameters and 300 future populations.) Even though none of these are too hard to compute, that is still quite a lot of bookkeeping. Think carefully about how you want to organize your calculations. You will want to use a combination of group_by calculations and joins to make this work. If you find yourself defining scalar variables outside of data frames, you are almost certainly doing down a difficult path.

At this point, you have (estimates) for the 2030 population of each state. We will use these to estimate congressional apportionments for the 2032 election. In the US, this is done using the Huntington-Hill method. The Wikipedia entry is reasonably clear, but I will also demonstrate how this was done for the 2020 reapportionment, a.k.a, the apportionment currently in effect.

library(fs)
library(tidyverse)
library(gt)
library(readxl)

# CA has 52 districts, so if we put in a hard max of 100 we're safe
MAX_DISTRICTS <- 100 
# Fixed per Reapportionment Act of 1929
N_DISTRICTS <- 435

data_mp03 <- fs::path("data", "mp03")
data_file <- fs::path(data_mp03, "apportionment-2020-tableA.xlsx")
if(!fs::dir_exists(data_mp03)) fs::dir_create(data_mp03, recurse=TRUE)
    
if(!fs::file_exists(data_file)){
    download.file("https://www2.census.gov/programs-surveys/decennial/2020/data/apportionment/apportionment-2020-tableA.xlsx", 
                  destfile=data_file)
}    

data_file |> 
    read_excel(skip=4) |> 
    select(state=1, 
           population=2) |> 
    filter_out(state %in%  c("District of Columbia", "U.S. Total")) |> 
    expand_grid(cd=seq(0, MAX_DISTRICTS)) |> 
    mutate(hh_den = sqrt(cd * (cd+1)),
           population=as.integer(population),
           hh_weight = population / hh_den) |> 
    slice_max(hh_weight, n=N_DISTRICTS) |> 
    group_by(state) |> 
    summarize(population = first(population), 
              n_districts = n()) |> 
    mutate(pop_per_district = population / n_districts) |> 
    arrange(desc(n_districts), desc(population), state) |> 
    gt(id="tbl_2020_apportionment",
       rowname_col="state") |> 
    cols_label(state=md("State"), 
               population="Apportionment Population", 
               n_districts="Number of Congressional Districts", 
               pop_per_district="Approximate Population per District") |> 
    fmt_integer(c(population, pop_per_district)) |> 
    data_color(columns=pop_per_district, 
               palette="PuOr") |>
    grand_summary_rows(columns=c(population, n_districts), 
                       fn=list("Total" ~ sum(.)), 
                       fmt = ~ fmt_integer(.)) |> 
    tab_header("Congressional Apportionments from the 2020 Census") |> 
    tab_source_note(md("Apportionments computed using the 
[Huntington-Hill Method](https://en.wikipedia.org/wiki/Huntington%E2%80%93Hill_method).
Data from the US Census Bureau 
[Reapportionment Table A](https://www2.census.gov/programs-surveys/decennial/2020/data/apportionment/apportionment-2020-tableA.xlsx)
(Resident Population + Assigned Overseas Population).")) |> 
    tab_footnote(
        locations=cells_column_labels(pop_per_district),
        "Colors indicate population per congressional representative, ranging from
almost 1 million residents per representative (DE, purple) to just over 540 thousand 
per representative (Montana, brown)."
    )

Congressional Apportionments from the 2020 Census
	Apportionment Population	Number of Congressional Districts	Approximate Population per District¹
California	39,576,757	52	761,091
Texas	29,183,290	38	767,981
Florida	21,570,527	28	770,376
New York	20,215,751	26	777,529
Pennsylvania	13,011,844	17	765,403
Illinois	12,822,739	17	754,279
Ohio	11,808,848	15	787,257
Georgia	10,725,274	14	766,091
North Carolina	10,453,948	14	746,711
Michigan	10,084,442	13	775,726
New Jersey	9,294,493	12	774,541
Virginia	8,654,542	11	786,777
Washington	7,715,946	10	771,595
Arizona	7,158,923	9	795,436
Massachusetts	7,033,469	9	781,497
Tennessee	6,916,897	9	768,544
Indiana	6,790,280	9	754,476
Maryland	6,185,278	8	773,160
Missouri	6,160,281	8	770,035
Wisconsin	5,897,473	8	737,184
Colorado	5,782,171	8	722,771
Minnesota	5,709,752	8	713,719
South Carolina	5,124,712	7	732,102
Alabama	5,030,053	7	718,579
Louisiana	4,661,468	6	776,911
Kentucky	4,509,342	6	751,557
Oregon	4,241,500	6	706,917
Oklahoma	3,963,516	5	792,703
Connecticut	3,608,298	5	721,660
Utah	3,275,252	4	818,813
Iowa	3,192,406	4	798,102
Nevada	3,108,462	4	777,116
Arkansas	3,013,756	4	753,439
Mississippi	2,963,914	4	740,978
Kansas	2,940,865	4	735,216
New Mexico	2,120,220	3	706,740
Nebraska	1,963,333	3	654,444
Idaho	1,841,377	2	920,688
West Virginia	1,795,045	2	897,522
Hawaii	1,460,137	2	730,068
New Hampshire	1,379,089	2	689,544
Maine	1,363,582	2	681,791
Rhode Island	1,098,163	2	549,082
Montana	1,085,407	2	542,704
Delaware	990,837	1	990,837
South Dakota	887,770	1	887,770
North Dakota	779,702	1	779,702
Alaska	736,081	1	736,081
Vermont	643,503	1	643,503
Wyoming	577,719	1	577,719
Total	331,108,434	435	—
¹ Colors indicate population per congressional representative, ranging from almost 1 million residents per representative (DE, purple) to just over 540 thousand per representative (Montana, brown).
Apportionments computed using the Huntington-Hill Method. Data from the US Census Bureau Reapportionment Table A (Resident Population + Assigned Overseas Population).

Note that this analysis uses the so-called Apportionment Population which includes ‘actual’ residents as well as assigned overseas populations (e.g., mapping active duty military back to their home states). You do not need to follow this subtlety in your analysis.

Task 8: Reapportionment

Apply the Huntington-Hill method to your population forecasts in order to estimate the number of congressional districts your state will have as of the 2030 redistricting cycle.

We are now finally ready for our final deliverable.

Task 9: Developing an Advertising Strategy to Induce Internal Migration and Increase State Political Power

Your boss, the governor, wants to increase the number of congressional seats your state has following the next census. To do so, you have been charged with developing an advertising strategy to convince more people to move to your state. Using your analysis up to this point (and any additional analysis you find helpful), develop a proposal for a new state advertising strategy. This strategy should:

Target common sources of migration into your state and its major metro areas in an attempt to increase in-migration.
Target common destinations of migration out of your state and its major metro areas in an attempt to convince people to move back.

For additional value, you might want to target metro areas in states that are near the low end of the population per resident statistic. If their population drops much lower, they might loose a congressional seat and you might hope to take it. E.g., in the 2020 apportionment shown above, Montana has just an apportionment population just over 1 million and two congressional seats. If just 10,000 Montanans were to move to New York, NY would gain an additional seat (going to 27) and Montana would have only one seat.¹³

To help ‘size’ your advertising campaign - and to get legislative support to pay for it - estimate how much you would internal migration you would need to net an additional congressional representative. (You do not need to do this analytically, rough numbers will suffice.)

Finally, write up a proposal for your advertisements including:

Where you want to advertise (see, at least, the two target metros above)
How much migration you are hoping to induce
A new slogan or marketing pitch for your state.¹⁴

AI Usage Statement

At the end of your report, you must include a description of the extent to which you used Generative AI tools to complete the mini-project. This should be a one paragraph section clearly deliniated using a collapsable Quarto “Callout Note”. Failure to include an AI disclosure will result in an automatic 25% penalty.

E.g.,

AI Usage Statement

No Generative AI tools were used to complete this mini-project.

AI Usage Statement

GitHub Co-Pilot Pro was used via RStudio integration while completing this project. No other generative AI tools were used.

AI Usage Statement

ChatGPT was used to help write the code in this project, but all non-code text was generated without the use of any Generative AI tools. Additionally, ChatGPT was used to provide additional background information on the topic and to brainstorm ideas for the final open-ended prompt.

Recall that Generative AI may not be used to write or edit any non-code text in this course.

These blocks should be created using the following syntax:


::: {.callout-note title="AI Usage Statement" collapse="true"}

Your text goes here. 

:::

Make sure to use this specific type of callout (.callout-note), title, and collapse="true" setting.

Please contact the instructor if you have any questions about appropriate AI usage in this course.

Extra Credit Opportunities

There are optional Extra Credit Opportunities where extra points can be awarded for specific additional tasks in this mini-project. The amount of the extra credit is typically not proportional to the work required to complete these tasks, but I provide these for students who want to dive deeper into this project and develop additional data analysis skills not covered in the main part of this mini-project.

For this mini-project, no more than 10 total points of extra credit may be awarded. Even with extra credit, your grade on this mini-project cannot exceed 80 points total.

Extra Credit Opportunity #01: Expanding the House

The (relatively) small size of the US House of Representatives is often cited as a cause of many systemic problems in US politics. For up to two points, write a brief 3 paragraph argument for upsizing the House focusing on i) equilibration of population per representative across states; and ii) the additional (relative) power gained by your state in a larger House. You may assume that the basic contours of the system remain unchanged (minimum one representative per state, the Huntington-Hill allocation) and that only the number 435 is up for debate.

Extra Credit Opportunity #02: Spatial Visualizations

When analyzing data like this, maps and other data visualizations can be very helpful. For up to three points, add additional visualizations to your analysis. You should include:

two maps
one chord diagram.¹⁵

The geometry=TRUE argument to tidycensus functions will be very helpful to you here.

If you omit the chord diagram, but have at least two maps, you can only get up to two points here.

Extra Credit Opportunity #03: More Realistic Growth Models

The population growth model described above is rather limited and assumes a lot of temporal and cross-sectional (between state) homogeneity. For up to four points, build a more realistic growth model that takes additional demographic data into account.

For one point, fit a different growth factor $g$ for each state instead of a national growth factor.

For up to two more points beyond that, further modify your model to take advantage of additional census variables. There are many ways that you might choose to modify this model, but some possibilities include:

Forecasting on smaller levels than the state level (e.g., county or after dividing into a rural/urban split)
Using demographic variables (especially age) to forecast how growth and death rates will change. (Younger populations have more children.)

For one more point, use a longer history to improve estimation of long-run growth parameters. This may involve parsing more data files.

Extra Credit Opportunity #04: Model Validation

Finally, for up to four points, evaluate the performance of your population forecast model retrospectively. E.g., if you had used the same model in 2020, how well would it have predicted populations in 2022 to 2025? There are many ways you can go with this, but at a minimum, you should

Make at least one additional set of predictions based on historical data
Compare the accuracy of those predictions with the realized values on a one-year, two-year, and three-year horizon.
Use that error estimate to put some sort of margin of error on your 2030 population predictions.

Footnotes

The American Time Use Survey (ATUS) from Mini-Project #02 is a downstream product from the ACS.↩︎
Specifically ACS-5 estimates are computed over a retrospective (looking backwards) five year window: e.g., the 2022 ACS-5 results are obtained by combining survey responses from 2018 to 2022.↩︎
This the level of “ChatGPT-level” prose, without obvious flaws but lacking the style and elegance associated with true quality writing.↩︎
Throughout this section, replace YOUR_GITHUB_ID with your GitHub ID from Mini-Project #00. Note that the automated course infrastructure will be looking for precise formatting, so follow these instructions closely.↩︎
As of the time of writing this assignment, the latest ACS data are 2024, so interpret words like “current” or “present” to refer to 2024. If 2025 data are available when you are working on this project, please use those instead to make more accurate forecasts.↩︎
This function is new in dplyr 1.2.0, released in Februrary 2026. If you do not have this function, restart R and run update.packages() to ensure you have the latest versions.↩︎
This data includes Abroad or Foreign country as a “one year ago” destination, so you have $50 * (50 + 1 - 1) = 2500$ total rows, not $50 * (50 - 1) = 2450$.↩︎
For consistency with the prior table, use the total residents previously living abroad (sum of Puerto Rico, U.S. Island Area, and Foreign Country) from column DH as your abroad factor.↩︎
This is the process of (re-)apportionment, or determining the total number of seats given to each state. After apportionment, that state determines how those seats are distributed within its boundaries (“redistricting”), via state legislative action, independent commission, judicial decree, etc.↩︎
An interesting, but somewhat obscure, idea for increasing the representativeness of the House of Representative (and, by extension, the electoral college) is to increase this limit from 435 to something much higher. See e.g.,

from the political left or
from the political right. You can explore this possibility in more detail in Extra Credit #01 below.↩︎
Due to the timing of the census, the federal election two years after a census is typically the first to use this updated apportionment. (So, e.g., results of the 2000 census were first used to determine parameters of the 2002 election.)↩︎
You might ask about out-migration (leaving the country) in this data. As a general rule, we do not get out-migration from US Census data because out-migrants are, by definition, no longer within the US. For our purposes, we are simply treating “out-migration” as a part of the death effect in the natural growth rate. After all, what is death but moving to the great census tract in the sky? (Cf., my childhood dog who moved to a nice farm upstate.)↩︎
In 2020, NY could have actually gained an extra seat with just a few hundred Montanans, but that seat would have come from Minnesota, not Montana. The apportionment algorithm is not entirely intuitive under small perturbations.↩︎
E.g., if you were trying to draw New Yorkers who retired to Florida back into the state, a slogan of “At least we keep our gators in the sewers” might be appropriate.↩︎
The ggraph package, especially the linear layout with circular=TRUE might be helpful, as well as this page from the R Graph Gallery.↩︎