How to use margins of error in a needs statement
A grant needs statement should make a clear case without making survey data sound exact. County Profiles supplies estimates and margins of error together so you can report both. Use this guide alongside the community needs statement guide when choosing evidence for your application.
Read the estimate and margin together
An estimate is the survey’s reported value for a defined population, or universe. The margin of error (MOE), shown after ±, describes sampling uncertainty around that estimate. The published American Community Survey MOEs here use a 90% confidence level. Over repeated samples using the same method, about 90% of intervals constructed this way would contain the population value. A particular interval is not a guaranteed set of bounds.
These are ACS 2020–2024 five-year Data Profiles, using the 2024 endpoint. They describe a period, not today’s conditions or a single-year count. Keep the release, universe and units attached to any number you copy. MOEs describe sampling uncertainty; they do not make a county estimate a count of your program’s clients.
Percentage example: poverty in Autauga County
The Autauga County, Alabama profile (GEOID 01001) reports a poverty estimate of 11.3% and a 90% MOE of 2.1 percentage points. The estimate variable is DP03_0128PE; its MOE is DP03_0128PM. The universe is people for whom poverty status is determined, rather than an assumed count of every resident.
Subtracting and adding the MOE gives 11.3 − 2.1 = 9.2% and 11.3 + 2.1 = 13.4%. This illustrates the estimate’s sampling uncertainty, not a promise that the population value falls inside that interval. Percentage points are absolute differences on the percentage scale: the MOE is not 2.1% of 11.3, which would be only 0.2373 percentage points.
Dollar example: median household income
The same county row reports median household income of $72,481, with a 90% MOE of $5,640. Trace these to DP03_0062E and DP03_0062M. The universe is households, and the unit is 2024 inflation-adjusted dollars. This is a household median, not income per person or an average household income.
The arithmetic is $72,481 − $5,640 = $66,841 and $72,481 + $5,640 = $78,121. Those endpoints illustrate uncertainty around the median estimate; they do not describe the range of household incomes. Both examples come from the Census Bureau’s 2024 ACS five-year Data Profiles source, table DP03.
A sentence you can copy
In Autauga County, Alabama, an estimated 11.3% of people for whom poverty status was determined lived below the poverty level (90% margin of error: ±2.1 percentage points; U.S. Census Bureau, ACS 2020–2024 five-year Data Profiles, DP03_0128PE/DP03_0128PM).
Handle gaps and comparisons honestly
Unavailable, suppressed or not-computable values are not zero. Leave them unavailable and retain the status flag. A usable estimate can have an unavailable MOE; report that limitation rather than inventing a margin. Controlled estimates without a published MOE also must not receive a zero MOE. Source sentinel codes represent statuses, not negative measurements.
County and state estimates can differ without establishing a statistically significant difference. This site computes no difference MOE or significance test; comparing interval overlap alone does not provide that test. Describe comparisons as descriptive and retain each geography’s uncertainty. For repeated applications, the full county spreadsheet pack keeps estimates, MOEs, status flags and benchmarks together.