Embedded Files
Software and languages
Software and languages
Programming languages Stata, R, SAS, Python, VBA, PowerShell, SQL
GIS software QGIS, ArcMap 10.x, ArcGIS Pro, ArcGIS Online, ArcGIS StoryMaps, ArcGIS Experience Builder, ModelBuilder, ArcPy
Project and code management Git, Github, Jira, Confluence, Trello, SharePoint, OneNote
Niche (enough) skills
Niche (enough) skills
I'm an expert at automating spatial analysis processes and data viz in R, Python, and ArcGIS and have been doing this since around 2014. Examples of geospatial work I've automated include:
Geocoding large datasets (upwards of 1M records) with the Google Maps API and ESRI StreetMap Premium;
Running large distance matrix calculations, for example computing the distance along the road network from each zip code to its nearest health facility;
Automating affine transformations to convert countless PDF maps into georeferenced shapefiles;
Converting line and polygon shapefiles into raster to create transit accessibility measures;
Running spatial autoregressive models; and
Making static maps with ggplot2 and tmap and interactive maps with R Leaflet.
I also have experience automating complex analysis and reporting products, mainly through developing data cleaning and analysis pipelines in PowerShell and using VBA to fill tables in Word documents with perfect formatting and footnotes.
Before Mathematica I worked at the Delaware Valley Regional Planning Commission, the metropolitan planning organization for Greater Philadelphia, so I'm familiar with transportation planning concepts and data.
Outside of work I am a skilled ceramicist and have been known to throw bowls on the potter’s wheel in under a minute. I like to throw sets of utilitarian objects (coffee mugs, bowls) and am constantly pursuing more technical mastery.
Things I am working on now
Things I am working on now
Measuring the added value of Texas higher education institutions through an analysis of 20 years of Texas students' demographic data, high school and postsecondary enrollment and performance, and wage data
Evaluating spatial distribution and accessibility of WIC retailers (and potential accessibility gaps) for all WIC agencies for the USDA's Food and Nutrition Service (FNS)
Creating a series of ArcGIS web maps showing all Head Start centers in the U.S., the difference between Head Start supply (number of slots) and estimated demand (number of income-eligible children) at the census tract level, and other tract-level demographic information
Evaluating the short- and long-term impacts of the RETAIN (Retaining Employment and Talent After Injury/Illness Network) Demonstration for the Social Security Administration and U.S. Department of Labor
Prior project examples
Prior project examples
Data visualization
Data visualization
School closures and the pre-pandemic digital divide
School closures and the pre-pandemic digital divide
Background
This app was created in R Shiny and enables the user to visualize county-level open-source data on school days impacted by school closures during the pandemic, students' access to technology, and student demographic characteristics. The map view, graphs, and infographics update automatically according to the combinations of variables selected and enable the user to visualize two variables simultaneously.
Link
Screenshot of the school closures app showing the intersection of school days impacted by school closure and broadband internet access for PA, NJ, and DE.
Community Connector
Community Connector
Background
This interactive web application was built in R Shiny and visualized social determinants of health data and health outcomes for Colorado Counties. It won the grand prize in the Agency for Healthcare Research and Quality's Visualization of Community-Level Social Determinants of Health Challenge.
Links
Screenshot of the Community Connector app.
FY 2023 Title X Service Grant Awards Grantee Profiles
FY 2023 Title X Service Grant Awards Grantee Profiles
Background
During FY 2023 the Office of Population Affairs awarded Title X Family Planning funding to 86 grantees nationwide. These brief Word documents summarize each grantee's service setting, clinic locations, and activities over the course of the year.
While the grantee profile documents are brief, they synthesize information from several sources, including data from a grantee survey and the American Community Survey; interviews of grant recipients; and addresses scraped, geocoded, and mapped from PDFs such as this one.
Contributions
Wrote code in Stata to clean a survey of grantees, merge on data from the Family Planning Annual Report, and produce grantee-level summary statistics on their activities during FY 2023
Wrote code in R to deduplicate and geocode grantees' clinic locations, download cartographic boundary and population data using packages such as tigris and tidycensus, and create static maps using ggplot2
Wrote code in VBA to automate the insertion of statistics; infographic elements with screen-reader friendly alternative text; maps; and document titles, footers, and metadata
Link
Screenshot of an example grantee profile.
Public policy analysis and impact evaluation
Public policy analysis and impact evaluation
Promoting Readiness of Minors in Supplemental Security Income (PROMISE)
Promoting Readiness of Minors in Supplemental Security Income (PROMISE)
Background
This project analyzed the impact of a program implemented across 6 state agencies to support youth with disabilities receiving Supplemental Security Income. The analysis included an evaluation of impacts five years after random assignment on youth and parent outcomes, a cost-benefit analysis, and a series of special topic reports. We worked with both survey and administrative data, such as surveys of youth and their parents five years after random assignment, earnings data from the Social Security Administration's Master Earnings File, and data on Medicaid and Medicare expenditure and use from the Centers for Medicare & Medicaid Services.
Contributions
Assisted with the analysis programming, which mainly consisted of multiple regression with survey-weighted and multiply-imputed data in Stata. The challenge was to apply consistent analysis methods across 7 different samples (6 state agencies and 1 sample pooling all records) that each had distinct sets of control variables and their own quirks.
Conducted sequence and cluster analyses in R for a special report on youth's pathways to education and employment.
Filled all the tables in the technical appendix. This included creating programs in Stata to correctly format estimates depending on the units of the results (e.g. dollar, percentage) and in Python and VBA to correctly transfer the results into the tables where they belong.
Links
Special report on youth's pathways to education and employment
Screenshot of the most common pathways for PROMISE youth.
Assessing the benefits of the Success Sequence for economic self sufficiency and family stability
Assessing the benefits of the Success Sequence for economic self sufficiency and family stability
Background
The "Success Sequence" posits that achieving the life milestones of education, employment, marriage, and childbearing, in this order, are associated with better economic outcomes in later life. This study tested the Success Sequence hypothesis by analyzing youth's milestone achievements and economic outcomes using two nationally representative longitudinal datasets: the National Longitudinal Survey of Youth and the National Longitudinal Study of Adolescent to Adult Health.
Contributions
Analyzed permutations of milestone achievement (summarizing the youth's milestones achieved and the order they were achieved in).
Ran LASSO regressions to identify demographic and background characteristics most closely related to economic outcomes so that these characteristics could be included in regression analysis as control variables.
Ran several multiple and logit regressions comparing milestone achievement to economic outcomes such as household income and income above the Federal Poverty Level.
Link
Screenshot of milestone completion and order for NLSY youth.
Large-scale data collection, cleaning, and summary
Large-scale data collection, cleaning, and summary
Family Planning Annual Report (FPAR) 2.0
Family Planning Annual Report (FPAR) 2.0
Background
Grantees receiving Title X federal funding for family planning are required to submit an annual report (that is, the Family Planning Annual Report) of their activities to the Office of Population Affairs (OPA) to monitor program performance. Mathematica's work on FPAR 2.0 includes creating and running an online portal to gather grantee data submissions; issuing a summary of grantees' activities, overall and by region; and conducting additional analyses as requested by OPA.
Contributions
Wrote code in R to create the formatted exhibits for the FPAR 2022 National Summary.
Double-programmed the calculations made in the FPAR 2.0 online data portal for quality assurance purposes.
Link
State Child Abuse & Neglect (SCAN) Policies Database
State Child Abuse & Neglect (SCAN) Policies Database
Background
The SCAN Policies Database is a longitudinal quantitative dataset of policies related to child maltreatment for the 50 U.S. states, DC, and Puerto Rico. These policies differ by state and can change from year to year; this database enables researchers to quickly compare across states and over time. Without the database, the researcher would need to track down each state's statutes documents online, download them, and read them before being able to make a comparison. I was part of the team for the first two rounds of data collection.
Contributions
Wrote code in Stata to clean the data, produce a data file for public use, and compare responses over time.
Created a process in Stata and VBA to programmatically update the data across 52 "State Profiles," or documents summarizing each state’s child maltreatment policies, with the push of a button.
Wrote code in VBA to compile state-specific notes and citations into a single Word document and apply styles and formatting to match the final deliverable.
Link
Screenshot of the SCAN Data Explorer.
Individuals with Disabilities Education Act (IDEA) 2019 state and local implementation study
Individuals with Disabilities Education Act (IDEA) 2019 state and local implementation study
Background
This study evaluated the ways states, districts, and schools were supporting children with disabilities 15 years after the Individuals with Disability Education Act (IDEA) was last updated and compared these results to similar surveys conducted in 2009. To get a full picture of IDEA implementation, the study issued 6 surveys, including 3 surveys of states, 2 surveys of a nationally representative sample of school districts, and one survey of a nationally representative sample of schools.
Contributions
Helped program two Stata ado-files to compute survey-weighted summary statistics of survey questions and export them into Excel tables of any layout. These ado-files were used to programmatically compute summaries of all survey questions for inclusion in the Compendium of Survey Results (linked below) and to produce smaller custom sets of tables for summary reports.
Wrote a suite of ado-files in Stata to fill in skip logic and special missing codes across six surveys based on survey metadata.
Provided programming support for other survey data-cleaning tasks, including incorporating back coding and constructing analysis variables.
Created codebooks for inclusion in the disclosure analysis plan.
Link
Head Start Family and Child Experiences Survey (FACES)
Head Start Family and Child Experiences Survey (FACES)
Background
FACES is a series of nationally representative surveys of Head Start programs, centers, classrooms, and the children and families who participate. Surveys are conducted twice a year and are regularly updated to address current events or policy questions. These surveys help programs and practitioners understand the characteristics of children who participate in Head Start, their families, and their teachers. I have been a member of the FACES team across four rounds of data collection with six surveys per round.
Contributions
Write code in Stata to clean the survey data as new responses are available.
Conduct analysis of datasets with complex survey designs in Stata, SAS, and R, including survey-weighted summary statistics, t-tests and chi-square analysis, linear regression, multiple imputation, and multilevel modeling.
Automate the table filling process in Word documents using VBA.
Links
American Indian and Alaska Native (AIAN) FACES Report Spring 2020
Research brief on performance of new cognitive assessments with Head Start children
Research brief on Head Start families' and program selection experiences
Geospatial and planning
Geospatial and planning
Regional Transit Screening Platform (RTSP)
Regional Transit Screening Platform (RTSP)
Background
This interactive geospatial web application provides information for transportation planners in Greater Philadelphia to consider public transit needs and opportunities, including a transit network gap analyzer, data on surface transit reliability and ridership, and other measures.
Contributions
Created the segment-level surface transit reliability measure for the RTSP using QGIS and R.
Links
Screenshot of surface transit reliability measure from the Regional Transit Screening Platform.
Equity Analysis for the Greater Philadelphia Region
Equity Analysis for the Greater Philadelphia Region
Background
The Delaware Valley Regional Planning Commission (DVRPC) creates a regularly-updated tract-level measure to evaluate the equity impacts of planning projects in Greater Philadelphia. This measure synthesizes American Community Survey (ACS) data on population age, race, ethnicity, disability status, income, and other variables into a single aggregate measure.
Contributions
Wrote a program in R to update the equity analysis measure (which is still in use after I wrote the code in 2018!)
Links
Screenshot of composite equity measure from the Equity Analysis for Greater Philadelphia web map.
Pedestrian Cyclical Count Program
Pedestrian Cyclical Count Program
Background
The Southeastern Pennsylvania Pedestrian Cyclical Count Program conducts counts of pedestrians at a selection of representative locations throughout the Greater Philadelphia region over time.
Contributions
Developed a stratified random sampling scheme to select a series of segments along the road network to conduct pedestrian counts. These strata accounted for population density, demographic characteristics, proximity to schools and transit, and road functional class.
Links
Screenshot of pedestrian count locations in Philadelphia.
Prior presentations and research
Prior presentations and research
Conference presentations
Conference presentations
Larson, A. (2019, May). Evaluating the reliability of ACS data for transportation planning. Paper presented at the American Community Survey Data Users Conference, Washington, DC.
Moran, S. & Larson, A. (2019, April). Data-driven transit planning tools. Project presented at the New Jersey TransAction Conference, Atlantic City, NJ.
Larson, A. (2018, April). Urban monocentricity and the journey to work: Commutes of low-income workers in Dallas, TX and Washington, DC. Paper presented at the Annual Meeting of the American Association of Geographers, New Orleans, LA.
University research
University research
Larson, A. (2018). Urban structure, residential choice, and proximity to work for low-income residents (Master's project). The University of Texas at Dallas, Richardson, TX.
Larson, A. (2017). Emergency call box visibility on the UT Dallas campus (Safety study for Student Government). The University of Texas at Dallas, Richardson, TX.
Larson, A. (2016). Hotbeds of hate: Analyzing spatial and temporal disparities in hate crime across U.S. cities (Undergraduate honors thesis). The University of Texas at Dallas, Richardson, TX.
Google Sites
Report abuse