Data Limitations and Risks
Learn about the limitations and risks in interpreting and comparing data, particularly across data sources and citations.
Estimates posted on this website may differ from those posted on other CDC and external sites. Differences between estimates can occur for several reasons:
- Differences in sample exclusion criteria or analytic methodologies.
- Differences in data sources, which will yield different results for the same indicator due to survey and measurement methodology. For example, hypertension in one survey may be assessed by direct measurement of blood pressure. In another survey, the participant may be asked if they had ever been told by a clinician that they had high blood pressure. These survey instruments would yield very different estimates of hypertension prevalence in the population.
Data Trends and Maps statistics used samples of specific sex, age, and race groups and excluded participants missing information for demographic variables and the indicator of interest. Therefore, the estimates presented may differ from other published sources, depending on sex, age, and race inclusion, standardization or adjustment factors, and the definition of the indicator. Please refer to footnotes found on the specific indicator pages for detailed information about definitions and inclusion/exclusion criteria.
Data Trends and Maps estimates are age-standardized, which produces an artificial estimate that minimizes the effects of different age distributions and allows comparisons between different populations. It represents what the crude percentage would have been in the study population if that population had the same age distribution as a standard population (a population in which the age composition is known precisely as a result of a census). The data presented on this website are directly standardized to the age distribution of the 2010 U.S. Census population, using the following age groups: 18–39, 40–59, 60+; or 20–39, 40–59, 60+; or 21–39, 40–59, 60+.
Data Trends and Maps estimates should not be used to assess statistical significance in trends over time. More sophisticated statistical techniques are required to test for significant linear trends in the estimates presented here. Users are also cautioned that estimates from different data sources should not be compared for differences or trends.