Assignment 1: Census Data Quality for Policy Decisions

Evaluating Data Reliability for Algorithmic Decision-Making

Author

Kavana Raju

Published

September 29, 2025

Assignment Overview

Scenario

You are a data analyst for the Pennsylvania Department of Human Services. The department is considering implementing an algorithmic system to identify communities that should receive priority for social service funding and outreach programs. Your supervisor has asked you to evaluate the quality and reliability of available census data to inform this decision.

Drawing on our Week 2 discussion of algorithmic bias, you need to assess not just what the data shows, but how reliable it is and what communities might be affected by data quality issues.

Learning Objectives

Apply dplyr functions to real census data for policy analysis
Evaluate data quality using margins of error
Connect technical analysis to algorithmic decision-making
Identify potential equity implications of data reliability issues
Create professional documentation for policy stakeholders

Submission Instructions

Submit by posting your updated portfolio link on Canvas. Your assignment should be accessible at your-portfolio-url/assignments/assignment_1/

Make sure to update your _quarto.yml navigation to include this assignment under an “Assignments” menu.

Part 1: Portfolio Integration

Create this assignment in your portfolio repository under an assignments/assignment_1/ folder structure. Update your navigation menu to include:

- text: Assignments
  menu:
    - href: assignments/assignment_1/your_file_name.qmd
      text: "Assignment 1: Census Data Exploration"

If there is a special character like comma, you need use double quote mark so that the quarto can identify this as text

Setup

# Load required packages (hint: you need tidycensus, tidyverse, and knitr)
library(dplyr)
library(tidycensus)
library(tidyverse)
library(sf)
library(tigris)
library(knitr)
library(stringr)
library(scales)
library(kableExtra)

# Set your Census API key
census_api_key(Sys.getenv("CENSUS_API_KEY"), install = FALSE, overwrite = FALSE)

# Choose your state for analysis - assign it to a variable called my_state
my_state <- "PA"

State Selection: I have chosen Pennsylvania for this analysis because: the scenario explicitly involves the Pennsylvania Department of Human Services, and PA’s mix of urban, suburban, and rural counties provides useful variation in ACS reliability for evaluating algorithmic decisions.

Part 2: County-Level Resource Assessment

2.1 Data Retrieval

Your Task: Use get_acs() to retrieve county-level data for your chosen state.

Requirements: - Geography: county level - Variables: median household income (B19013_001) and total population (B01003_001)
- Year: 2022 - Survey: acs5 - Output format: wide

Hint: Remember to give your variables descriptive names using the variables = c(name = "code") syntax.

# Write your get_acs() code here
county_vars = c(
  med_hh_income = "B19013_001",
  total_pop = "B01003_001"
  )

county_data <- get_acs(
  geography = "county",
  state = my_state,
  year = 2022,
  survey = "acs5",
  variables = county_vars,
  output = "wide",
)

# Clean the county names to remove state name and "County"
county_data <- county_data %>%
  mutate(county_name = str_remove(NAME, paste0(", ", "Pennsylvania")) %>%
           str_remove("County")
  )
  
# Hint: use mutate() with str_remove()

# Display the first few rows
head(county_data)

# A tibble: 6 × 7
  GEOID NAME     med_hh_incomeE med_hh_incomeM total_popE total_popM county_name
  <chr> <chr>             <dbl>          <dbl>      <dbl>      <dbl> <chr>      
1 42001 Adams C…          78975           3334     104604         NA "Adams "   
2 42003 Alleghe…          72537            869    1245310         NA "Allegheny…
3 42005 Armstro…          61011           2202      65538         NA "Armstrong…
4 42007 Beaver …          67194           1531     167629         NA "Beaver "  
5 42009 Bedford…          58337           2606      47613         NA "Bedford " 
6 42011 Berks C…          74617           1191     428483         NA "Berks "

glimpse(county_data)

Rows: 67
Columns: 7
$ GEOID          <chr> "42001", "42003", "42005", "42007", "42009", "42011", "…
$ NAME           <chr> "Adams County, Pennsylvania", "Allegheny County, Pennsy…
$ med_hh_incomeE <dbl> 78975, 72537, 61011, 67194, 58337, 74617, 59386, 60650,…
$ med_hh_incomeM <dbl> 3334, 869, 2202, 1531, 2606, 1191, 2058, 2167, 1516, 21…
$ total_popE     <dbl> 104604, 1245310, 65538, 167629, 47613, 428483, 122640, …
$ total_popM     <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,…
$ county_name    <chr> "Adams ", "Allegheny ", "Armstrong ", "Beaver ", "Bedfo…

2.2 Data Quality Assessment

Your Task: Calculate margin of error percentages and create reliability categories.

Requirements: - Calculate MOE percentage: (margin of error / estimate) * 100 - Create reliability categories: - High Confidence: MOE < 5% - Moderate Confidence: MOE 5-10%
- Low Confidence: MOE > 10% - Create a flag for unreliable estimates (MOE > 10%)

Hint: Use mutate() with case_when() for the categories.

# Calculate MOE percentage and reliability categories using mutate()
county_cat <- county_data %>%
  mutate(moe_perc = ((med_hh_incomeM / med_hh_incomeE) * 100) , 
           reliability = case_when(moe_perc < 5 ~ "High Confidence", 
                                   moe_perc < 10 ~ "Moderate Confidence",
                                   TRUE ~ "Low Confidence")
         )

# Create a summary showing count of counties in each reliability category
county_cat %>%
  count(reliability, name = "count") %>%
  mutate(share = scales::percent(count / sum(count)))

# A tibble: 2 × 3
  reliability         count share
  <chr>               <int> <chr>
1 High Confidence        57 85%  
2 Moderate Confidence    10 15%

# Hint: use count() and mutate() to add percentages

2.3 High Uncertainty Counties

Your Task: Identify the 5 counties with the highest MOE percentages.

Requirements: - Sort by MOE percentage (highest first) - Select the top 5 counties - Display: county name, median income, margin of error, MOE percentage, reliability category - Format as a professional table using kable()

Hint: Use arrange(), slice(), and select() functions.

# Create table of top 5 counties by MOE percentage
top5_uncertain <- county_cat %>%
  arrange(desc(moe_perc)) %>%
  slice(1:5) %>%
  transmute(
    county        = county_name,
    median_income = paste0("$", format(round(med_hh_incomeE), big.mark=",")),
    moe_dollars   = paste0("$", format(round(med_hh_incomeM), big.mark=",")),
    moe_percent   = sprintf("%.1f%%", moe_perc),
    reliability
  )

# Format as table with kable() - include appropriate column names and caption
top5_uncertain %>%
  kable(
    col.names = c("County","Median Income","MOE ($)","MOE (%)","Reliability"),
    caption   = "Top 5 Counties by Median Income MOE% (Highest First)",
    align     = c("l","r","r","r","l")
  ) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

Top 5 Counties by Median Income MOE% (Highest First)
County	Median Income	MOE ($)	MOE (%)	Reliability
Forest	$46,188	$4,612	10.0%	Moderate Confidence
Sullivan	$62,910	$5,821	9.3%	Moderate Confidence
Union	$64,914	$4,753	7.3%	Moderate Confidence
Montour	$72,626	$5,146	7.1%	Moderate Confidence
Elk	$61,672	$4,091	6.6%	Moderate Confidence

Data Quality Commentary:

Across Pennsylvania, the counties with the largest household income MOE are small counties such as Forest, Sullivan, Union, Montour, and Elk, with MOE between 6% and 10%. If an algorithm ranks counties by median income without accounting for this uncertainty, these counties can be misclassified because of sampling error and may cross decision thresholds incorrectly. Because they have smaller populations, they should be treated with caution and considered for manual review or aggregation to larger geographies.

Part 3: Neighborhood-Level Analysis

3.1 Focus Area Selection

Your Task: Select 2-3 counties from your reliability analysis for detailed tract-level study.

Strategy: Choose counties that represent different reliability levels (e.g., 1 high confidence, 1 moderate, 1 low confidence) to compare how data quality varies.

# Use filter() to select 2-3 counties from your county_reliability data
# Store the selected counties in a variable called selected_counties
county_cat <- county_cat %>% mutate(county_code = substr(GEOID, 3, 5))

selected_counties <- county_cat %>%
  filter(reliability %in% c("High Confidence","Moderate Confidence","Low Confidence")) %>%
  group_by(reliability) %>%
  slice_head(n = 1) %>%
  ungroup() %>%
  select(county_code, county_name, med_hh_incomeE, med_hh_incomeM, moe_perc, reliability)

# Display the selected counties with their key characteristics
selected_counties %>%
  dplyr::mutate(
    `Median Income` = paste0("$", format(round(med_hh_incomeE), big.mark=",")),
    `MOE (%)`       = sprintf("%.1f%%", moe_perc)
  ) %>%
  dplyr::select(
    County = county_name, `Median Income`, `MOE (%)`, Reliability = reliability
  ) %>%
  kable(caption = "Selected Counties for Tract-Level Study") %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

Selected Counties for Tract-Level Study
County	Median Income	MOE (%)	Reliability
Adams	$78,975	4.2%	High Confidence
Cameron	$46,186	5.6%	Moderate Confidence

# Show: county name, median income, MOE percentage, reliability category

Comment on the output: Both chosen counties are at the high and moderate confidence levels. The absence of any low confidence counties suggests that county level income is measured with fair stability across the state, though smaller rural places may still sit near the threshold and deserve careful review.

3.2 Tract-Level Demographics

Your Task: Get demographic data for census tracts in your selected counties.

Requirements: - Geography: tract level - Variables: white alone (B03002_003), Black/African American (B03002_004), Hispanic/Latino (B03002_012), total population (B03002_001) - Use the same state and year as before - Output format: wide - Challenge: You’ll need county codes, not names. Look at the GEOID patterns in your county data for hints.

# Define your race/ethnicity variables with descriptive names
race_vars <- c(
  total    = "B03002_001",
  white    = "B03002_003",
  black    = "B03002_004",
  hispanic = "B03002_012"
)

# Use get_acs() to retrieve tract-level data
# Hint: You may need to specify county codes in the county parameter
selected_codes <- selected_counties$county_code

tract_raw <- get_acs(
  geography = "tract",
  state     = my_state,
  county    = selected_codes,
  variables = race_vars,
  year      = 2022,
  survey    = "acs5",
  output    = "wide"
)

# Calculate percentage of each group using mutate()
# Create percentages for white, Black, and Hispanic populations
# Add readable tract and county name columns using str_extract() or similar

tract_data <- tract_raw %>%
  mutate(county_code = substr(GEOID, 9, 11)) %>%
  separate(NAME, into = c("tract_name", "county", "state"),
           sep = ";\\s*", remove = FALSE, fill = "right") %>%
  mutate(
    pct_white   = (100 * whiteE    / totalE),
    pct_black   = (100 * blackE    / totalE),
    pct_hisp    = (100 * hispanicE / totalE)
  ) %>%
  select(-state)

3.3 Demographic Analysis

Your Task: Analyze the demographic patterns in your selected areas.

# Find the tract with the highest percentage of Hispanic/Latino residents
# Hint: use arrange() and slice() to get the top tract

top_hispanic_tracts <- tract_data %>%
  group_by(county) %>%
  arrange(desc(pct_hisp), .by_group = TRUE) %>%
  slice(1) %>%
  ungroup() %>%
  transmute(
    Tract = tract_name,
    County = county,
    `Pct Hispanic/Latino` = pct_hisp,
    `Total Pop` = totalE
  )

top_hispanic_tracts %>%
  kable(
    caption   = "Top % Hispanic/Latino Tract in Each Selected County",
    col.names = c("County","Tract","Pct Hispanic/Latino","Total Pop"),
    align     = c("l","l","r","r")
  ) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

Top % Hispanic/Latino Tract in Each Selected County
County	Tract	Pct Hispanic/Latino	Total Pop
Census Tract 315.02	Adams County	20.880246	3908
Census Tract 9601	Cameron County	2.867203	1988

# Calculate average demographics by county using group_by() and summarize()
# Show: number of tracts, average percentage for each racial/ethnic group

county_avgs <- tract_data %>%
  group_by(county) %>%
  summarise(
    Tracts = n(),
    `Avg % White` = mean(pct_white, na.rm = TRUE),
    `Avg % Black` = mean(pct_black, na.rm = TRUE),
    `Avg % Hispanic/Latino` = mean(pct_hisp, na.rm = TRUE)
  ) %>%
  rename(County = county)

# Create a nicely formatted table of your results using kable()
county_avgs %>%
  knitr::kable(
    caption   = "Average Demographics by County (Selected Counties)",
    col.names = c("County","Tracts","Avg % White","Avg % Black","Avg % Hispanic/Latino"),
    digits    = c(NA, 0, 2, 2, 2),      # one decimal place for % cols
    align     = c("l","r","r","r","r")
  ) %>%
  kableExtra::kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

Average Demographics by County (Selected Counties)
County	Tracts	Avg % White	Avg % Black	Avg % Hispanic/Latino
Adams County	27	88.33	1.31	7.14
Cameron County	2	93.19	0.04	2.10

Part 4: Comprehensive Data Quality Evaluation

4.1 MOE Analysis for Demographic Variables

Your Task: Examine margins of error for demographic variables to see if some communities have less reliable data.

Requirements: - Calculate MOE percentages for each demographic variable - Flag tracts where any demographic variable has MOE > 15% - Create summary statistics

# Calculate MOE percentages for white, Black, and Hispanic variables
# Hint: use the same formula as before (margin/estimate * 100)
# Create a flag for tracts with high MOE on any demographic variable
# Use logical operators (| for OR) in an ifelse() statement

tract_moe <- tract_data %>%
  mutate(
    moe_pct_white = ifelse(whiteE    > 0, 100 * whiteM    / whiteE,    NA_real_),
    moe_pct_black = ifelse(blackE    > 0, 100 * blackM    / blackE,    NA_real_),
    moe_pct_hisp  = ifelse(hispanicE > 0, 100 * hispanicM / hispanicE, NA_real_),
    high_moe_any  = (moe_pct_white > 15) | (moe_pct_black > 15) | (moe_pct_hisp > 15)
  )

# Create summary statistics showing how many tracts have data quality issues
summary_tract_moe <- tract_moe %>%
  group_by(county) %>%
  summarise(
    `Tracts (total)`    = n(),
    `Tracts (high MOE)` = sum(high_moe_any, na.rm = TRUE),
    `Share high MOE`    = sprintf("%.2f%%", 100 * `Tracts (high MOE)` / `Tracts (total)`)
  )

summary_tract_moe %>%
  kable(
    col.names = c("County","Tracts","High MOE","Share high MOE"),
    caption   = "High-MOE Tracts by County",
    align     = c("l","r","r","r")
  ) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

High-MOE Tracts by County
County	Tracts	High MOE	Share high MOE
Adams County	27	26	96.30%
Cameron County	2	2	100.00%

4.2 Pattern Analysis

Your Task: Investigate whether data quality problems are randomly distributed or concentrated in certain types of communities.

# Group tracts by whether they have high MOE issues
# Calculate average characteristics for each group:
# - population size, demographic percentages
# Use group_by() and summarize() to create this comparison

tract_pattern <- tract_moe %>%
  mutate(high_moe_any = ifelse(is.na(high_moe_any), FALSE, high_moe_any)) %>%
  group_by(county, high_moe_any) %>%
  summarise(
    Tracts                 = n(),
    `Avg Pop`              = mean(totalE, na.rm = TRUE),
    `Avg % White`          = mean(pct_white, na.rm = TRUE),
    `Avg % Black`          = mean(pct_black, na.rm = TRUE),
    `Avg % Hispanic/Latino`= mean(pct_hisp,  na.rm = TRUE),
    .groups = "drop_last"
  ) %>%
  mutate(Group = if_else(high_moe_any, "High-MOE tracts", "Other tracts")) %>%
  ungroup() %>%
  rename(County = county) %>%
  select(County, Group, Tracts, `Avg Pop`, `Avg % White`, `Avg % Black`, `Avg % Hispanic/Latino`)

# Create a professional table showing the patterns
tract_pattern %>%
  kable(
    caption = "Comparing Tracts With vs. Without High MOE (Any Demographic Variable)",
    digits  = c(NA,NA, 0, 0, 2, 2, 2),
    align   = c("l","l","r","r","r","r","r")
  ) %>%
  kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

Comparing Tracts With vs. Without High MOE (Any Demographic Variable)
County	Group	Tracts	Avg Pop	Avg % White	Avg % Black	Avg % Hispanic/Latino
Adams County	Other tracts	1	2416	98.68	0.00	0.00
Adams County	High-MOE tracts	26	3930	87.93	1.36	7.42
Cameron County	High-MOE tracts	2	2268	93.19	0.04	2.10

Pattern Analysis: In Adams County, almost all tracts are flagged as high MOE, and the tracts with high MOE also show more racial and ethnic diversity than the single tract without issues. In Cameron County, every tract is flagged as high MOE, reflecting the effect of very small population size. This pattern confirms that high MOE is concentrated both in small rural places and in tracts where subgroup counts are small, which increases the risk of unstable algorithmic classifications.

Part 5: Policy Recommendations

5.1 Analysis Integration and Professional Summary

Your Task: Write an executive summary that integrates findings from all four analyses.

Executive Summary Requirements: 1. Overall Pattern Identification: What are the systematic patterns across all your analyses? 2. Equity Assessment: Which communities face the greatest risk of algorithmic bias based on your findings? 3. Root Cause Analysis : What underlying factors drive both data quality issues and bias risk? 4. Strategic Recommendations: What should the Department implement to address these systematic issues?

Executive Summary:

Overall Pattern Identification: Most counties in Pennsylvania fall into the high confidence group, and only a handful are moderate confidence. At the county scale, ACS estimates for income are fairly stable. At the tract scale, however, many estimates exceed the reliability threshold, showing that data quality declines as geography becomes more granular.
Equity Assessment: Communities most at risk of algorithmic bias are those with smaller populations and greater diversity. These are the places where sampling variability makes the data least reliable. If an algorithm ranks tracts or counties without accounting for uncertainty, these communities could be wrongly deprioritized or overrepresented in resource allocations.
Root Cause Analysis: The reliability issues are driven by sample size and subgroup sparsity. Smaller counties and tracts have fewer responses, and race and ethnicity subgroups within them have even smaller counts, leading to higher margins of error. County level data appears stable because the sample is larger, while tract level data shows greater variation. These patterns reflect how ACS sampling works rather than errors in the data.
Strategic Recommendations: Algorithmic prioritization can move forward in high confidence counties, with standard monitoring in place. In moderate confidence counties, margins of error should be factored into scoring and human review should be required for borderline cases. At the tract level, reliability weighting, aggregation, and the use of complementary administrative data can reduce the risk of misclassification. All tracts flagged with high MOE should be subject to human oversight, and the Department should establish monitoring routines to catch disparities introduced by uncertainty.

6.3 Specific Recommendations

Your Task: Create a decision framework for algorithm implementation.

# Create a summary table using your county reliability data
# Include: county name, median income, MOE percentage, reliability category

# Add a new column with algorithm recommendations using case_when():
# - High Confidence: "Safe for algorithmic decisions"
# - Moderate Confidence: "Use with caution - monitor outcomes"  
# - Low Confidence: "Requires manual review or additional data"

rec_table <- county_cat %>%
  transmute(
    County          = county_name,
    `Median Income` = scales::dollar(round(med_hh_incomeE)),
    `MOE (%)`       = sprintf("%.2f%%", moe_perc),
    Reliability     = reliability,
    Recommendation  = case_when(
      Reliability == "High Confidence"     ~ "Safe for algorithmic decisions",
      Reliability == "Moderate Confidence" ~ "Use with caution — monitor outcomes",
      TRUE                                 ~ "Requires manual review or additional data"
    )
  )

# Format as a professional table with kable()
rec_table %>%
  knitr::kable(
    caption = "County-Level Decision Framework for Algorithm Implementation",
    align   = c("l","r","r","l","l"),
    col.names = c("County","Median Income","MOE (%)","Reliability","Recommendation")
  ) %>%
  kableExtra::kable_styling(full_width = FALSE, bootstrap_options = c("striped","hover","condensed"))

County-Level Decision Framework for Algorithm Implementation
County	Median Income	MOE (%)	Reliability	Recommendation
Adams	$78,975	4.22%	High Confidence	Safe for algorithmic decisions
Allegheny	$72,537	1.20%	High Confidence	Safe for algorithmic decisions
Armstrong	$61,011	3.61%	High Confidence	Safe for algorithmic decisions
Beaver	$67,194	2.28%	High Confidence	Safe for algorithmic decisions
Bedford	$58,337	4.47%	High Confidence	Safe for algorithmic decisions
Berks	$74,617	1.60%	High Confidence	Safe for algorithmic decisions
Blair	$59,386	3.47%	High Confidence	Safe for algorithmic decisions
Bradford	$60,650	3.57%	High Confidence	Safe for algorithmic decisions
Bucks	$107,826	1.41%	High Confidence	Safe for algorithmic decisions
Butler	$82,932	2.61%	High Confidence	Safe for algorithmic decisions
Cambria	$54,221	3.34%	High Confidence	Safe for algorithmic decisions
Cameron	$46,186	5.64%	Moderate Confidence	Use with caution — monitor outcomes
Carbon	$64,538	5.31%	Moderate Confidence	Use with caution — monitor outcomes
Centre	$70,087	2.77%	High Confidence	Safe for algorithmic decisions
Chester	$118,574	1.70%	High Confidence	Safe for algorithmic decisions
Clarion	$58,690	4.37%	High Confidence	Safe for algorithmic decisions
Clearfield	$56,982	2.79%	High Confidence	Safe for algorithmic decisions
Clinton	$59,011	3.86%	High Confidence	Safe for algorithmic decisions
Columbia	$59,457	3.76%	High Confidence	Safe for algorithmic decisions
Crawford	$58,734	3.91%	High Confidence	Safe for algorithmic decisions
Cumberland	$82,849	2.20%	High Confidence	Safe for algorithmic decisions
Dauphin	$71,046	2.27%	High Confidence	Safe for algorithmic decisions
Delaware	$86,390	1.53%	High Confidence	Safe for algorithmic decisions
Elk	$61,672	6.63%	Moderate Confidence	Use with caution — monitor outcomes
Erie	$59,396	2.55%	High Confidence	Safe for algorithmic decisions
Fayette	$55,579	4.16%	High Confidence	Safe for algorithmic decisions
Forest	$46,188	9.99%	Moderate Confidence	Use with caution — monitor outcomes
Franklin	$71,808	3.00%	High Confidence	Safe for algorithmic decisions
Fulton	$63,153	3.65%	High Confidence	Safe for algorithmic decisions
Greene	$66,283	6.41%	Moderate Confidence	Use with caution — monitor outcomes
Huntingdon	$61,300	4.72%	High Confidence	Safe for algorithmic decisions
Indiana	$57,170	4.65%	High Confidence	Safe for algorithmic decisions
Jefferson	$56,607	3.41%	High Confidence	Safe for algorithmic decisions
Juniata	$61,915	4.79%	High Confidence	Safe for algorithmic decisions
Lackawanna	$63,739	2.58%	High Confidence	Safe for algorithmic decisions
Lancaster	$81,458	1.79%	High Confidence	Safe for algorithmic decisions
Lawrence	$57,585	3.07%	High Confidence	Safe for algorithmic decisions
Lebanon	$72,532	2.69%	High Confidence	Safe for algorithmic decisions
Lehigh	$74,973	2.00%	High Confidence	Safe for algorithmic decisions
Luzerne	$60,836	2.35%	High Confidence	Safe for algorithmic decisions
Lycoming	$63,437	4.39%	High Confidence	Safe for algorithmic decisions
McKean	$57,861	4.75%	High Confidence	Safe for algorithmic decisions
Mercer	$57,353	3.63%	High Confidence	Safe for algorithmic decisions
Mifflin	$58,012	3.43%	High Confidence	Safe for algorithmic decisions
Monroe	$80,656	3.17%	High Confidence	Safe for algorithmic decisions
Montgomery	$107,441	1.27%	High Confidence	Safe for algorithmic decisions
Montour	$72,626	7.09%	Moderate Confidence	Use with caution — monitor outcomes
Northampton	$82,201	1.93%	High Confidence	Safe for algorithmic decisions
Northumberland	$55,952	2.67%	High Confidence	Safe for algorithmic decisions
Perry	$76,103	3.17%	High Confidence	Safe for algorithmic decisions
Philadelphia	$57,537	1.38%	High Confidence	Safe for algorithmic decisions
Pike	$76,416	4.90%	High Confidence	Safe for algorithmic decisions
Potter	$56,491	4.42%	High Confidence	Safe for algorithmic decisions
Schuylkill	$63,574	2.40%	High Confidence	Safe for algorithmic decisions
Snyder	$65,914	5.56%	Moderate Confidence	Use with caution — monitor outcomes
Somerset	$57,357	2.78%	High Confidence	Safe for algorithmic decisions
Sullivan	$62,910	9.25%	Moderate Confidence	Use with caution — monitor outcomes
Susquehanna	$63,968	3.14%	High Confidence	Safe for algorithmic decisions
Tioga	$59,707	3.23%	High Confidence	Safe for algorithmic decisions
Union	$64,914	7.32%	Moderate Confidence	Use with caution — monitor outcomes
Venango	$59,278	3.45%	High Confidence	Safe for algorithmic decisions
Warren	$57,925	5.19%	Moderate Confidence	Use with caution — monitor outcomes
Washington	$74,403	2.38%	High Confidence	Safe for algorithmic decisions
Wayne	$59,240	4.79%	High Confidence	Safe for algorithmic decisions
Westmoreland	$69,454	1.99%	High Confidence	Safe for algorithmic decisions
Wyoming	$67,968	3.85%	High Confidence	Safe for algorithmic decisions
York	$79,183	1.79%	High Confidence	Safe for algorithmic decisions

Key Recommendations:

Your Task: Use your analysis results to provide specific guidance to the department.

Counties suitable for immediate algorithmic implementation: These are counties with high confidence data. Their margins of error are below 5 percent, which means the ACS estimates are stable and can be used in automated ranking and prioritization. The department can proceed with algorithmic allocation in these places with standard outcome monitoring to ensure fairness. Counties: Adams, Allegheny, Armstrong, Beaver, Bedford, Berks, Blair, Bradford, Bucks, Butler, Cambria, Centre, Chester, Clarion, Clearfield, Clinton, Columbia, Crawford, Cumberland, Dauphin, Delaware, Erie, Fayette, Franklin, Fulton, Huntingdon, Indiana, Jefferson, Juniata, Lackawanna, Lancaster, Lawrence, Lebanon, Lehigh, Luzerne, Lycoming, McKean, Mercer, Mifflin,
Monroe, Montgomery, Northampton, Northumberland, Perry, Philadelphia, Pike, Potter, Schuylkill, Somerset, Susquehanna, Tioga, Venango, Washington, Wayne, Westmoreland, Wyoming, and York.
Counties requiring additional oversight: These are counties with moderate confidence data, where margins of error fall between 5 and 10 percent. Estimates here are less precise, so algorithms should be used with caution. The department should adjust scoring to account for reliability, add human review for borderline cases, and monitor allocations to verify that communities are neither under- nor over-prioritized due to sampling error. Counties: Cameron, Carbon, Elk, Forest, Greene, Montour, Snyder, Sullivan, Union, and Warren.
Counties needing alternative approaches: These are counties or tracts where margins of error exceed acceptable thresholds. At the county level this group may be empty, but at the tract level many places fall here. In such cases, alternative strategies are needed: aggregating small geographies, combining ACS with administrative data, and requiring manual validation before final decisions. This ensures communities are not misclassified due to unstable estimates.

Questions for Further Investigation

Are high MOE tracts geographically clustered and do these clusters overlap with areas of higher service need?
How stable are county and tract level reliability measures across ACS releases?
What other datasets could be combined with ACS to reduce uncertainty and strengthen targeting?

Technical Notes

Data Sources: - U.S. Census Bureau, American Community Survey 2018-2022 5-Year Estimates - Retrieved via tidycensus R package on 09/23/2025.

Reproducibility: - All analysis conducted in R version 4.4.1 - Census API key required for replication - Complete code and documentation available at: https://musa-5080-fall-2025.github.io/portfolio-setup-kavanaraju/assignments/assignment_1/assignment1_template.html

Methodology Notes: Reliability categories were defined using relative margins of error. Counties were selected to represent different reliability levels. Tracts were flagged as high MOE if any race or ethnicity variable exceeded a 15 percent margin of error.

Limitations: Margins of error are especially large at small geographic scales and for subgroup estimates, making tract level analysis more uncertain. This work uses one ACS release and does not test temporal consistency. Only a subset of variables was analyzed, and nonresponse patterns in ACS may add further bias not captured by MOE.

Submission Checklist

Before submitting your portfolio link on Canvas:

All code chunks run without errors
All “[Fill this in]” prompts have been completed
Tables are properly formatted and readable
Executive summary addresses all four required components
Portfolio navigation includes this assignment
Census API key is properly set
Document renders correctly to HTML

Remember: Submit your portfolio URL on Canvas, not the file itself. Your assignment should be accessible at your-portfolio-url/assignments/assignment_1/your_file_name.html