The Hoosier Life Survey (HLS) is the nation’s most comprehensive statewide public-opinion survey of environmental change to date. The survey offers state-specific insights on public attitudes toward environmental change, personal values, trust in news media, attitudes toward a variety of kinds of risk, and more.
Between August and December 2019, the HLS team reached out to 10,000 Hoosiers across Indiana. In total, more than 2,700 Hoosiers—representing 90 of the state’s 92 counties—responded.
Taken together, respondents’ answers show what Hoosiers think about environmental change—its origins, its extent, its impact on their families. The survey also indicates how Hoosiers learn about the issues vital to their future—who they trust, who they listen to, and who they’d like to hear more from.
The HLS highlights how much Indiana residents are already doing—or are prepared to do—to build resilience against environmental change. And it reveals the role of political and personal values—along with social, demographic, and economic differences—in shaping Hoosiers' approach to a global challenge.
The survey from which data for this map was drawn was sent out to 10,000 Hoosiers between August and December 2019. The survey focused on gathering a broad range of information related to Indiana residents’ views of their community, environmental changes and risk, climate change beliefs, the household- and community- level actions they were taking or supported being pursued, and their personal values. Surveys were sent to Indiana households using a spatially stratified sampling approach. To ensure adequate coverage of people across the entire state and for later geographically specific analysis, our team developed eight in-state regions, defined by clusters of counties. Each of Indiana’s 92 counties was included in a region. From each region, 1,250 home addresses were drawn at random from the United States Postal Service’s list (for a total of 10,000), which was purchased from a private address-based sampling vendor.
In mailing surveys to these households, a modified Dillman approach was used with a total of five mailing waves (Dillman, Smyth, & Christian, 2014). In an initial wave of mailings, households received a cover letter informing them about the survey, noting the confidentially of their responses and asking them to fill out the survey online. A link to the online survey and user ID number were provided in the cover letter. A second letter with the survey link and user ID was sent to those who had yet to respond two and a half weeks after a postcard reminder was sent to all respondents.
One month later, respondents who had yet to fill out the survey online were sent a paper booklet version of the survey and another cover letter requesting their participation. A final mailing wave, containing another booklet and cover letter, was sent to all remaining non-participants after two-months. Both the initial contact for the web- based survey and the mail-based version contained $1 pre-incentive payments. Upon completing the survey, respondents could request a $20 Amazon or Walmart gift card. In total, our weighted response rate was just over 29 percent. Case-wise deletion analysis was used to address missing data in this map, resulting in 1,630 cases being examined. The composite margin of error for this sample, accounting for the impact of design effects, is +/-3.7 at a 95% confidence level. Patterns of missing data were explored, as were relationships between missing responses and key demographics. No consistent patterns emerged, nor were strong relationships identified. In terms of differences between the full and complete samples, average age of respondents was the only significantly different demographic variable, with the complete-case sample being only very marginally younger. Future HLS data products may use data imputation methods depending on analysis type and the variables of interest.
To ensure accurate population estimates for this analysis, survey weights were used. Weighting incorporates: (1) a base weight adjustment for unequal probabilities of selection due to disproportionate stratified sampling by region and due to the number of adults in the household, (2) a differential nonresponse adjustment to correct for unequal response rates by region, and (3) a calibration adjustment to 2018 American Community Survey estimates on gender, age, education, race, and Hispanic origin in the Indiana adult population. Weights were trimmed and scaled to the unweighted number of respondents.
Metropolitan Mapping Data:
The data for the metropolitan areas is drawn from the same survey used to gather the state-level data. As the survey was not specifically designed to assess public opinion at the metropolitan level, results presented at this level should be interpreted cautiously.
To enable the most accurate estimates of metropolitan level public opinion, survey weights were used to ensure that sample and population characteristics (e.g. age, gender, etc.) were roughly equal. A weighting process similar to the process described for the state-level was used to prepare the metropolitan area weights. No adjustment for differential response rates by region was made since these were sub-state level weights.
Pairwise deletion was used to address missing values for the metro specific analyses. This enabled each analysis to draw on the greatest number of possible responses. The number of responses below represent the lowest number of cases from each metropolitan area included in the map (excluding questions for which certain respondents were asked to skip):
Bloomington, IN: 173 respondents
Average Margin of Error: 9.3
Evansville, IN-KY: 167 respondents
Average Margin of Error: 10.1
Fort Wayne, IN: 163 respondents
Average Margin of Error: 9.6
Indianapolis, IN: 442 respondents
Average Margin of Error: 6
New Albany-Jeffersonville, INL 150 respondents
Average Margin of Error: 10.1
Northwest Indiana: 202 respondents
Average Margin of Error: 8.0
South Bend-Mishawaka, IN: 166 respondents
Average Margin of Error: 8.6
Regional Mapping Data:
The data for the regions is drawn from the same survey used to gather the state-level data.
To enable the most accurate estimates of regional level public opinion, survey weights were used to ensure that sample and population characteristics (e.g. age, gender, etc.) were roughly equal. A weighting process similar to the process described for the state-level was used to prepare the regional weights.
Pairwise deletion was used to address missing values for the regional analyses. This enabled each analysis to draw on the greatest number of possible responses. The number of responses below represent the lowest number of cases from each region included in the map (excluding questions for which certain respondents were asked to skip):
Corn and Soy Belt: 518 respondents
Margin of Error: 6.3
Metropolitan Indianapolis: 362 respondents
Margin of Error: 6.8
Northern Industrial/Suburban: 317 respondents
Margin of Error: 8.01
Ohio River Valley: 331 respondents
Margin of Error: 7.5
Southeast: 383 respondents
Margin of Error: 7.2
Southwestern/Indiana Uplands: 392 respondents
Margin of Error: 6.6
Wabash River Valley: 313 respondents
Margin of Error: 7.4