Social Explorer provides access to many datasets beyond the American Community Survey. These non-ACS datasets cover topics that ACS does not measure, including crime, religion, economic activity, environmental conditions, housing programs, and international censuses. When a user submits a question that falls outside the scope of ACS, the AI Assistant selects the most appropriate non-ACS dataset based on topic, geographic coverage, and data availability.
When the AI Uses Non-ACS Datasets
The AI selects a non-ACS dataset when the question involves subjects that ACS does not collect. ACS focuses primarily on demographic, social, economic, and housing characteristics. It does not cover crime, religious affiliation, detailed environmental risks, election results or administrative program data. In these cases, the AI automatically retrieves results from specialized datasets within the Social Explorer Data Library.
Major Non-ACS Dataset Groups
Social Explorer includes a wide range of curated datasets from official and authoritative sources. The main categories include:
Decennial Census
Full population counts and basic demographics collected once every ten years. These datasets provide historical population trends from 1790 to 2020 and serve as the foundation for redistricting and many longitudinal analyses.
EASI Estimates and Projections
Modeled demographic, economic and market projections covering past, current and future years. These datasets are useful for planning, forecasting and market analysis. Typically available from 2010 through 2030.
Economic Data
Datasets covering employment, business patterns, industry activity and economic indicators. These sources help users analyze economic growth, local business environments and labor market trends. Most datasets range from the 1990s through 2024.
Health and Environmental Data
Datasets from agencies such as CDC and EPA that include information on cancer, health factors, air quality, environmental justice, food access and other health-related topics. Coverage generally spans the 2000s through 2024.
Housing Data
Administrative and survey-based datasets covering housing costs, housing permits, affordability and fair market rents. These datasets allow detailed analysis of housing markets and policy outcomes. Typically available from the early 2000s through 2024.
Religion Data
Surveys and administrative counts on religious affiliation and congregational membership. These datasets offer insight into religious composition and community structures. Common years range from 1980 through 2020.
Election and Voting Data
Datasets containing presidential and congressional election results, turnout statistics, and voting-age population characteristics. Coverage commonly spans elections from 2000 through 2024.
Crime Data
Crime statistics and law enforcement data from sources such as the FBI and UCR. These datasets include crime rates, counts, and agency-level records. Years generally range from the 1960s through the early 2020s.
International Census Data
Population, demographic, and economic data from Canada, the UK, Europe and other countries for users who need international comparisons. Typically available from the 1980s through the 2010s.
Specialized Indices
Datasets that combine multiple indicators into a single index, including social vulnerability, environmental justice, and food access scores. Most indices cover the 2000s through the 2020s.
How the AI Selects a Non-ACS Dataset
When the requested topic is not available in ACS, the AI evaluates the dataset groups that match the subject. It verifies that the dataset contains the variable needed to answer the question and checks whether the dataset supports the requested geographic level. If multiple datasets relate to the topic, the AI selects the most complete and recent version.
If a user requests information for a specific year, the AI chooses the version of the dataset that best fits the requested time period. When a dataset offers both historical and recent versions, the AI ensures that the selected dataset aligns with the user’s context.
Examples of Non-ACS Dataset Selection
- If a user asks for crime rates in a county, the AI selects crime datasets because the ACS does not contain crime statistics.
- If a user asks how many religious adherents live in a particular region, the AI uses religion datasets, since the ACS does not collect information on religion.
- If a user requests business or employment counts for a ZIP code, the AI selects economic datasets that include business and labor statistics.
- If a user asks for environmental risk indicators, the AI may use datasets such as environmental justice indices or air-quality data.
Fallback Logic for Non-ACS Data
When the ideal dataset cannot provide the required variables, year, or geography, the AI applies fallback rules. It may choose another available year, switch to a broader geograph,y or select a related dataset if one exists. If no dataset covers the topic, the AI clearly informs the user that official data sources do not capture the information.
Why Non-ACS Datasets Matter
Non-ACS datasets make it possible to analyze subjects that are not part of the American Community Survey. They extend the range of research questions users can ask, enabling analysis of crime, religion, environment, elections, business activity, health, and international populations. By automatically selecting the correct dataset, the AI ensures that users receive accurate, relevant, and properly sourced data across a wide range of topics.