Data on this site come from multiple sources including the US Census Bureau, the RI Department of Health, RI State Police and Broadband RI. Detailed public school indicators are expected to be added in the near future from the RI Department of Education.

We are actively pursuing new data partnerships and will add additional datasets as they become available.

According to the U.S. Constitution, the Census Bureau is mandated to collect population data every 10 years. This decennial census includes detailed data on age, sex, households, families, and race. It is released as Census Summary File 1 (100% data).

In addition to basic demographic information collected during the decennial census, the Census Bureau also collects detailed information on a sample of the population. This data was collected prior to the 2010 Census decennially through the long form Census survey Summary File 3 (sample data), but is now collected annually through the American Community Survey (ACS).

The ACS is an ongoing survey of the population conducted continuously throughout the year and includes additional questions about socio-economic and housing characteristics including family and relationships, income, health insurance, education, employment, and disabilities. Many of the questions asked by SF3 are now available on the ACS. Additionally, due to a smaller sample size, ACS data are unavailable or unreliable at smaller levels of geography (Census blocks or block groups).

The ACS is available for 3 different time periods which are to be selected carefully based on precision and population:
* 1-year (12 months of collected data) - the smallest sample size, which is less reliable than 3-year or 5-year samples, but contains the most current data. Contains data for areas with populations of 65,000+;
* 3-year (36 months of collected data) - larger sample size than 1-year data and is more reliable than 1-year, but less reliable than 5-year data. Contains data for areas with populations of 20,000+;
* 5-year (60 months of collected data) - the largest sample size and the most reliable data, but the least current. Contains data for all areas (including tract and block group).

When analyzing large populations and currency is more important that precision, 1-year estimates will be appropriate. When analyzing small populations where precision is more important than currency, 5-year estimates are best. For the RI Community Profiles we have chosen to display the most recent 5-year data since estimates are more reliable and available down to the block group level.

It is appropriate to compare SF3 and ACS 5-year data for many but not all variables. The 5-year ACS estimates are the data that best replicate former decennial SF3 tables since census tract and block group level data are available. However, Margin of Error (MOE) are generally wide at lower levels of geography and caution should be used when reporting. See the Margin of Error section below for a description of MOE and sampling error for the ACS.

Census geographies are updated every decade to account for population shifts and growth. Governmental boundaries such as state, county, and municipality tend to remain unchanged between censuses. The geographies which tend to change are the smaller, non-governmental boundaries such as census tracts and block groups.

State and county boundaries are always census tract boundaries in the standard census geographic hierarchy. Block groups nest in tracts; tracts nest in municipalities (county subdivisions); municipalities nest in counties; and counties nest in states.

Census tracts are small statistical subdivisions of a county delineated by a local committee of census data users for the purpose of presenting data. Tracts nest within counties, and their boundaries normally follow visible features, but may follow legal geography boundaries and other non-visible features in some instances. Census tracts ideally contain about 4,000 people and 1,600 housing units, but can have a population size between 1,200 and 8,000 people. Tract boundaries are delineated with the intention of being maintained over time so that statistical comparisons can be made from census to census. Tracts are occasionally split due to population growth or merged as a result of substantial population decline.

Block Groups (BGs) are statistical divisions of census tracts, generally defined to contain between 600 and 3,000 people and 240 and 1,200 housing units. BGs are used to present data and control block numbering. A BG consists of clusters of blocks within the same census tract. BGs are the smallest geographic unit for which the Census Bureau tabulates sample data (SF3 and ACS).

Documentation of all changes is available on the Census TIGER website. http://www.census.gov/geo/www/tiger/

The American Community Survey (ACS) is a continuous sampling of population and households throughout the year. Since the ACS data are based on survey responses, they are subject to error. This error is expressed using the Margin of Error (MOE) at a 90 percent confidence interval. A 90 percent confidence interval can be interpreted roughly as providing 90 percent certainty that the true number falls between the upper and lower bounds. The ACS is generally the only data set published in RI Community Profiles that contains MOE. The MOE is a measure of how close an estimate is to the actual figure. This MOE is a value that is added to or subtracted from the estimate to represent a range within which the actual value occurs. For example, an estimate of 640 with a MOE of 115 means that one can say with 90% certainty that the actual value falls between 525 and 755.

You can see the MOEs on RI Community Profiles in:
* Tables by hovering over the "+/-" to the right of the multi-year estimate.
* The large maps by hovering over each geography - a box in the upper right corner updates with the geography name, the estimated value, and the MOE following "+/-".
* Data table downloads - a MOE will accompany every ACS estimate. RI Community Profiles provides MOEs for each ACS estimate across geography levels and should always be referenced. Some estimates may have a MOE too large to make any meaningful conclusions. The smaller the MOE, the better the estimate. An estimate with a MOE which is 50% or more of that estimate should be considered unreliable.

Another way to determine if an estimate has too much error to be reliable is to calculate the "relative MOE". The relative MOE is the MOE divided by the estimate and multiplied by 100. There is not a standard for the ranges of acceptable relative MOEs, but the California State Data Center recommends the following reliability scale for relative MOEs:
* < 33 % is good
* 33-49 % is use with caution
* 50 % + is unreliable

RI Community Profiles deals with MOE in several ways. First, we recalculate MOE for any indicator that was aggregated or changed from the original census published format to provide the truest MOE available. Secondly, we supply MOE in all of our ACS tabular output allowing a user to easily obtain the MOE, compare with an indicator value, or calculate the relative MOE. Lastly, we are in the process of establishing a system and visual format to publish MOE reliability estimates on the site in the future.

To overcome large MOEs users have a few options. They can sum to create new groups (ex: age 0-5 instead of age 0,1,2,3,4,5 alone), sum geography levels to create a new geography level (ex: add several tracts together to get a neighborhood), or do a combination of the two. However, users will have to create new MOEs to accompany their new estimates, which may be a difficult process for those not familiar with advanced mathematics. Documentation of necessary formulas is provided by the U.S. Census in Appendix 3 of the ACS Handbook.

For more details see ACS 2008-2012 Technical Documentation

Not on current iteration of site (for future reference): Lastly, we provide error bars on our Map & Ranking graphics. Although we do provide interactive mapping without indication of MOE, our hope is that users will get more mileage out of the Map & Ranking because it is a more powerful data display for analysis.