Sampling

The sample for K-HANS was statewide with stratified over-sampling in areas of the state where African Americans and Hispanics/Latinos reside in high density to achieve a sample that will provide reliable results for the entire Kansas population as well as the two minority subgroups. U.S. Census and other sources of demographic information were used to identify blocks of telephone numbers with the highest proportions of targeted racial and ethnic minority communities across the state. A disproportionate stratified random digit dial (RDD) sample was purchased from a commercial vendor. This sampling method improves survey efficiency by sampling from listed telephone numbers at a higher rate than unlisted telephone numbers. The unequal sampling rates are accounted for by applying weights to the survey data. The sample vendor pre-screened the sample to eliminate businesses, institutions, and non-working numbers. Potential working telephone numbers were called during three separate calling occasions (daytime, evening, and weekends) for a total of 15 call attempts before being replaced. Upon reaching a valid residential number, one adult was randomly selected from each household to participate in the study. If the selected respondent was not available, an appointment was made to call at a later date. Information regarding Hispanic ethnicity and race were elicited from the selected adult respondent. All selected respondents who were Hispanic or black or African American continued with the interview. For all other respondents, a randomization process was utilized to interview a sub-sample. This process ensured that the resulting sample contained approximately the same number of respondents who are Hispanic, black or African American, and all other races and ethnicities. If the selected respondent could not be reached during the survey calling period or refuses to participate, then the telephone number was replaced with another randomly selected number. The goal was to have completed surveys from 2,400 Kansans including 800 interviews each from African Americans and Hispanic/Latino respondents.

If the household contained children under the age of 18, a child was selected at random, so that the respondent answered a set questions pertaining to the child’s dietary behaviors and physical activity.

Weighting Procedure

Data Collection

Data collection was conducted by the Health Risk Studies Program at KDHE. The Health Risk Studies Program utilizes WinCATI, a state-of-the-art fully automated and networked computer-assisted telephone interviewing system. WinCATI is loaded onto workstations throughout the Bureau of Health Promotion and is used by Health Risk Studies interviewers during the daytime, evening, and weekend hours to conduct interviews. Respondent data is entered into the WinCATI system by the interviewers as each interview is conducted and is stored on a secure server and backed up locally.

The survey was translated into Spanish, and interviewers fluent in the Spanish language were available to interview sampled individuals whose primary language was Spanish.

Questionnaire Design

The K-HANS consisted of 136 questions. Survey topics included: dietary knowledge, dietary intake, community environment, availability of food choices, obesity status (derived from reported height and weight), self-reported health status, physical activity, diagnosis of related chronic disease or risk factors, weight control efforts, preventive counseling related to overweight/obesity, and food security. The majority of these questions have been used on previous Behavioral Risk Factor Surveillance System (BRFSS) questionnaires. Other questions were taken from: North Carolina Six County Cardiovascular Health Survey, Nurses Health Study II, and various research studies. Some questions were developed by the principal investigator, Kansas Health Institute staff, and Bureau of Health Promotion staff. Kansas Health Institute and Bureau of Health Promotion will use data from the Kansas Health, Activity, and Nutrition Survey for program planning and policy-relevant research.

Response Rate

The sample for K-HANS included a total of 69,870 telephone numbers and 17,384 households. In total, 2,171 respondents particpated in this survey. Race and ethnicity breakdown includes: 909 Hispanic, 655 Black, Non Hispanics, and 607 Other, Non-Hispanics. For the purpose of this study, Other was defined as: White, Asian, Hawaiian or Other Pacific Islander, Alaskan Native or American Indian, and respondents who indicated their race as Other.

Limitations

Sampling: Sampling yields results which are an estimate of the true answer for the entire population. The more persons that are interviewed, the greater the precision of the estimate. When the data are subdivided to look at sub-populations (e.g., an age subgroup) these estimates will be less precise; if the number of persons interviewed was small because the subgroup represents a small fraction of the population, the estimate may become too uncertain to be of value.

Because the survey is conducted by telephone, persons without telephones could not be reached. Since phone ownership is highly correlated to income, persons without a phone are more likely to have low incomes than persons with a telephone. This will potentially affect questions with responses that are highly dependent on income (e.g., health insurance) more than other questions. However, because phone ownership is high in Kansas (greater than 95%), it is unlikely that failing to reach these persons will substantially alter results.

Questionnaire Design and Administration: How a question is written and which questions preceded it in the questionnaire can influence responses in unpredictable ways. Not all the questions used in the survey have been tested to ensure that all persons understand the intended meaning. Furthermore, not all questions are equally easy for respondents to answer. While it may be easy for a respondent to provide a personal opinion, it may be much harder to recall a past event (last blood cholesterol test) or provide factual information (household income).

Interviewers are trained and monitored (see Quality Control Page) to ensure that they administer the survey in a neutral voice and read the written question verbatim and without comment. Nonetheless, it is possible for the interviewer to bias the results through tone of voice or administration technique. Coding errors may also occur if the interviewer types in the wrong response to the question. In addition, the person being interviewed may alter his or her response to give the interviewer the most socially acceptable answer. This may be a problem especially for questions which may have a perceived stigma.

Confounding and Causation: Personal characteristics which are presented on this web site are univariate (i.e., examine each risk factor in relationship to only one characteristic at a time); however, the complexity of health associations are not fully represented by examining single relationships.This web site does not attempt to explain the causes of the health effects examined. Being a cross-sectional survey, the results from this survey provide a “snapshot” of disease, risk factors, and population characteristics for adult residents of Kansas at a point in time.