Data Types - Screening and Risk Factors Data Types
Direct Estimates are the usual estimates reported from a survey-directly estimated from the survey data using appropriate weighting. Most of these direct estimates are based on data from the Behavioral Risk Factor Surveillance System (BRFSS), a large annual telephone survey of U.S. households designed to measure at the state level.
State-Level or County-Level Modeled Estimates combine results from BRFSS and the National Health Interview Survey (NHIS), a smaller, in-person survey, using a statistical model that takes advantage of the strengths of each survey. These estimates:
- use the strength of the NHIS (better response rate and reaches houses without land-line telephones) to adjust BRFSS estimates.
- smooth the results by applying data from demographically similar areas across the country.
- enable estimates for counties with few or no BRFSS respondents by using data from similar areas, albeit at the loss of specificity to that particular small area.
- pool data from multiple years to obtain more responses for small areas.
- can only be used when both surveys ask identical questions, so the number of statistics that can be modeled this way is limited.
For each outcome, the state level modeled estimates are controlled to the direct regional estimates and the county level modeled estimates are controlled to the controlled state level modeled estimates so the aggregated county level estimates equal to the corresponding state level estimates and the aggregated state level estimates equal to the corresponding regional level direct estimates.
See http://sae.cancer.gov for more details.
Summary of Available Statistics by Data Type
|Direct Estimates||Bias-Adjusted Modeled Estimates|
|Time Period||Most current single year||Pooled years|
|Survey(s)||BRFSS (or CPS or ANR*)||BRFSS and NHIS combined|
|Topics Available||Clinical Breast Exam
Home Blood Stool Test
Overweight or Obese
Signmoidoscopy or Colonoscopy
Smoking Not Allowed and Smoking Laws (not based on BRFSS*)
|Limitations||Telephone-only survey method introduces bias and inaccuracies (greater for some statistics than for others)||Depending on the local sample rate, estimates for small areas may not reflect local factors (such as a media campaign to promote mammograms)|
* Smoking Not Allowed measures are direct estimates and are from the Current Population Survey (CPS) Tobacco Use Supplement. The data for Smoking Laws come from Americans for Nonsmokers' Rights (ANR) and are based on analysis of laws across the country.
- State-Level and County-Level Modeled Estimates (Small Area Estimates), National Cancer Institute
- Behavioral Risk Factor Surveillance System, Centers for Disease Control and Prevention
- National Health Interview Survey, Centers for Disease Control and Prevention
- Current Population Survey Tobacco Use Supplement, National Cancer Institute and U.S. Census Bureau
- U.S. Tobacco Control Laws Database©, Americans for Nonsmokers' Rights Foundation