Methodology

This survey is based on a similar survey conducted among North Carolina AARP members in 2002 and on an earlier survey developed by the Missoula Demonstration Project (now the Life’s End Institute (LEI)) and conducted in Missoula as part of LEI’s community profile of end-of-life experience and care (www.lifes-end.org). The Nebraska End of Life Survey is the first such survey conducted with a random sample of adults across an entire state. Responses to each question asked are included in bar graph form for reference as an appendix in this report.

A random sample of 3,002 Nebraska households in which at least one adult was twenty-five years or older was obtained from ACUDATA America, a company that has access to all of the most credible lists and data for mail, telephone sales, permission-based email and Web marketing campaigns. Rural, urban, and Hall County, households were sampled separately to obtain enough participants in each geographic location to conduct sub-group analyses. From the original sample (3,002), 2,402 contacts were randomly selected and mailed surveys. Extra contacts were obtained for the purpose of further mailings if necessary. Three-hundred-four residents responded. However, four surveys were incomplete and not analyzed, resulting in an N of 300, and a response rate of 13 percent.

The survey sampling error is plus or minus 6 percent. This means that in 95 out of 100 samples of this size, the results obtained in the sample would fall in a range of + 6 percentage points of what would have been obtained if every Nebraska household had been surveyed.

All contacts were considered to be the head of the household. Race, age ranges of household members, gender, income, and phone numbers of those who had them were included in the sample data. This data was used to compare the respondents to non-respondents on these characteristics for the purpose of examining potential biases due to response rate.

In July 2003, all contacts received a survey with a return postage paid envelope and a letter asking them, or another adult in the household, to complete the survey and return it within two weeks. As an incentive, participants were offered the opportunity to take part in a raffle if they returned a completed survey. Four eligible participants received $50. Follow-up procedures consisted of one post-card reminder to non-respondents mailed in August.

To determine if the sample of respondents differed from the group that did not respond analyses were conducted to examine potential differences between the two groups on various demographic items. The significance level was set at p < .05. Using this criterion it was determined there were no differences in the rate of response between the rural communities, urban communities, and Hall County. A statistically equal percentage of individuals responded from all three locations, c ² (2) = 1.799, p = .407. Analysis of the responses themselves did not indicate any significant urban/rural differences. Additionally, there were no differences between respondents and non-respondents on three other variables: household income, c ² (8) = 11.98, p = .152; race, c ² (10) = 12.14, p = .276; and marital status, c ² (3) = 3.777, p = .287. Thus, we can conclude that the respondents were representative of the entire sample on these variables.

The gender of the respondents and non-respondents was also examined. The contact list was based on head of household. In most cases, a male was listed as head of household and the survey was addressed to him. Typically women are more likely to complete surveys than are men. That was the case with survey, as well (see table below). Additionally, a significantly higher percentage of female heads of household completed the survey compared to the female contacts who did not, (40% female respondents vs. 31% female non-respondents), c ² (1) = 9.155, p = .002. Except as noted in the report itself, response difference between woman and men were not great.

Individual ages were not available, but information on the age range, and gender of each household member was provided. Each classification grouping (e.g., females ages 25 to 34, males ages 25 to 34, etc.) was examined comparing respondents and non-respondents. Only one group was different, females age 35 to 44. There were fewer females in the household age 35 to 44 in the respondent group (15%) compared to the non-respondent group (23%), c ² (1) = 9.626, P = .002, indicating that this age group of women respondents were not representative of all the women sampled in this age group. One could interpret this difference in response as indicating that women in Nebraska falling in this age range are less inclined to complete a survey of end-of-life issues than other female age groups. It might be surmised that, in general, women in this age group may be focused on raising young children, may not yet face the decline of their parents, and/or may be too busy to complete a long survey.

The scant differences between the demographic variables of the respondents and non-respondents suggest that the Nebraska survey results are representative of the entire sample.