Methods & Limitations
This survey was commissioned by Welborn, conducted by Diehl Consulting Group, reported and published in partnership with Liz Tharp Consulting.
Survey & Analysis
Welborn’s 2026 edition of the Greater Evansville Health Survey (GEHS) report provides the best source for county-level health data specific to the region. The GEHS survey questions, the implementation process, weighting and analysis procedures are modeled after the Centers for Disease Control’s Behavioral Risk Factor Surveillance System (BRFSS), though local interests resulted in deviations from BRFSS methods in some cases. The BRFSS survey focuses on overall health, prevalence of health conditions, and related behaviors that influence health.
This edition of the survey was conducted in 2026. Similar studies were conducted in 2008, 2015, and 2021, though methodological differences prevent comparisons prior to 2021.
The survey was developed by Welborn in consultation with Diehl Consulting Group, Evansville, IN. Diehl Consulting Group also managed the data collection process. Similar to prior survey administrations, a child module was included in this edition of the survey to collect health information about a sample of the children in the respondents’ homes. This child module adds significant value to the survey and to the collective body of knowledge about children’s health in the region. All adults, aged 18 and older in Vanderburgh, Warrick, Gibson, and Posey Counties in Indiana and Henderson County in Kentucky were eligible to participate in the survey. These five survey counties were chosen to be representative of the area’s population centers, including urban, suburban, and rural populations.
A stratified random sample of households in Gibson, Henderson, Vanderburgh, and Warrick counties, and all households within Posey County, were invited to participate in the survey. The sample was drawn from all valid household addresses in the counties using the most current listing of occupied housing units as provided by DataMail (January 2025). Addresses included single-family and multi-family dwellings, but not PO boxes or businesses. Two versions of the survey were created. Adult survey items were the same for both versions, but the child instructions varied. One survey version asked respondents to base their responses on the oldest and the other on the youngest child. Households were randomly assigned to one of the two survey versions by county. The counterbalancing of child items was intended to create a greater representation of child ages. Each mailing included an outgoing envelope, the cover letter (with a perforated incentive contact form on the bottom), survey, and return envelope. The outgoing envelope included Welborn’s return address, and the return envelope was addressed directly to Diehl Consulting Group. While surveys were mailed, respondents did have the option to complete the survey electronically. A survey link was included in the invitation letter for those choosing the electronic option. Participants also had an option to view and complete the survey in Spanish, Marshallese, or Haitian-Creole. The first survey administration occurred in January 2025. To increase sample sizes and achieve desired margins of error, a second mailing was conducted in February 2025. Most data collection was completed by March 1, 2025.
Extra steps were taken to ensure representation of population sub-groups in the Greater Evansville region. A total of 1,977 adults completed the survey. In addition, information was collected from adults on a total of 320 children under 18 years old. For adult survey returns, the margin of error (with 95% confidence) was 2% for the overall region, 5% for Gibson, 5% for Henderson, 5% for Posey, 4% for Vanderburgh, and 5% for Warrick. For adult survey returns with the child module completed, the margin of error (with 95% confidence) was 5% for the overall region. For each county, data were weighted by race, ethnicity, sex, and age to ensure that the sample more accurately reflected the characteristics of the population from which it was drawn. Population characteristics (i.e., control variables) were derived from the 2024 American Community Survey (ACS; 5-year estimates). The survey sample was weighted to the population controls through an iterative raking process. Survey weighting, descriptive analyses, and inferential analyses were performed by Diehl Consulting Group. Where appropriate, highlighted findings involving direct comparisons between two or more groups were based on statistically significant differences. Descriptive findings are also presented to highlight areas determined to be of practical importance. Statistical testing included parametric and non-parametric methods.
Low-Income Definition
“Low-income” data throughout the report is defined as less than 80% of the median family income for the Evansville MSA in 2025, accounting for the number of individuals in the household.
Measuring Progress & Comparing Data Sources
Because of the consistency between the 2021 and 2026 surveys and methods, comparisons to the prior administration are included for key data points throughout the report. However, methodological differences should be considered when making comparisons with other data sources. While state and national data (as well as regional data from other sources) may be valuable in some cases to better understand regional findings, studies using different methodologies and analytic strategies do not allow for direct comparisons. Further, it is important to consider the timing of available data, as the most recent data presented in a given source may lag the data presented in this report and/or reflect a rolling average across years. Recognizing these considerations, this report provides an overall look at the top concerns, controllable risk factors, overall health status, and promising practices in 2026. This overall view of health in the region highlights the fact that, over survey years, some of the same issues keep rising to the top.
Limitations
A few limitations are important to note. First, the sample of adult respondents was adequate to ensure that the overall margin of error did not exceed 5% (with 95% confidence) for any individual county after rounding. Further, data were weighted prior to analyses to account for any demographic differences between the responding sample and the population from which it was drawn. However, disaggregation of results by geographic factors (e.g., county), demographic factors (e.g., race, income), and health factors (e.g., obesity) necessarily increases the margin of error around survey findings. It is not possible to have the same statistical confidence in some of these breakdowns as it is in the aggregated, regional findings.
Similarly, data collection efforts targeted a representative sample of households in each county. It was not possible, however, to control for the presence or absence of a child in the household. As a result, while the sample of adult respondents was robust, many were not able to provide information on children’s health. This relative lack of information around children’s health warranted a cautious approach to presenting child data, such as focusing on regional rather than county-specific rates.
Finally, self-report studies provide helpful information about the prevalence of disease and controllable risk factors. However, self-report does not provide a complete and thorough understanding of health and quality of life in a community. For example, self-report only provides information from those who have survived diseases and conditions, e.g., those who indicated that they have ever had heart disease. In addition, results for children are based on self-reports from parents. Self-reports can be subject to respondent bias.
This study is about disease prevalence and health behaviors, but does not include every possible disease, condition, or risk behavior, nor does it point to causes. To achieve a more comprehensive understanding of a community’s health, this study should be supplemented by epidemiology data, e.g., mortality rates for disease states such as heart disease, and other data sources, e.g., health department data. Other data sources may show different rates for the various health outcomes in this survey. The methodology of external data sources will point to possible reasons for different rates, e.g., different methods of collecting or analyzing data, different years, populations, or geographies.
Project Team
Welborn
Andrea Hays, Chief Program & Impact Officer
Mardi File, Healthy Communities Program Officer
Jeff Seymore, Chief Communications Officer
Sam Voss, Marketing & Communications Specialist
Survey Administration, Analysis, Report Writing, & Consultation
Doug Berry – Diehl Consulting Group, Evansville, IN
Liz Tharp – Liz Tharp Consulting
The 2026 edition of the Greater Evansville Health Survey is a product of Welborn
in partnership with Diehl Consulting Group and Liz Tharp Consulting. View the survey methodology.