br General health literacy br The Brief Health
2.2.3. General health literacy
The Brief Health Literacy Screening Tool used by Haun and colleagues  was adopted to measure levels of health literacy with three items. One of items includes: “How difficult is it 1905453-18-0 for you to understand information that doctors, nurses and other health professionals tell you?” Each item was scored on a five-point Likert scale, ranging from 0 (‘never’) to 4 (‘always’). A sum score of the three items was used in the analyses as a continuous variable. In the models, general health literacy was also used as a mediating variable, and the Cronbach’s a of the three items was 0.679.
2.2.4. Information overload
Information overload was a mediating variable in the models. To assess participants’ past health information seeking experience (i.e., perceived online information processing), one item was adopted from the Health Information National Trends Survey questionnaire : “Based on the results of your past 12-month search for information about health or medical topics, how much do you agree or disagree with the following statements?—You felt frustrated during your search for the information.” The item was scored on a five-point Likert scale, ranging from 0 (‘never’) to 4 (‘absolutely’).
Eleven items of Internet use for seeking health-related information from the Health Information National Trends Survey questionnaire  were used to assess OHIS behaviors within the past 12 months. The items include: “I used e-mail or the Internet to communicate with a doctor or doctor’s office,” “I visited a ‘social networking’ site to read and share about medical topics,” and “I looked for health or medical information for someone else.” A ‘yes’ response was denoted as 1 and ‘no’ was 2. All ‘yes’ (=1) responses were summed up. The summed scores in the analyses indicate the number of behaviors regarding seeking health information through the Internet for different purposes. The total health-related Internet use score demonstrated moderate internal consistency with an alpha coefficient of 0.741.
A history of a cancer for self and family, respectively, was used as a control variable for the analyses. Both variables were measured with an item, respectively: “Have you ever had a cancer?” and “Has any of your family members (e.g., parents, grandparents, siblings, and kin members) ever had a cancer?” A ‘yes’ was denoted as 1 and a ‘no’ was 2.
Descriptive analysis was conducted to describe the character-istics of the sample. Independent t-tests and chi-square tests were implemented to assess bivariate associations. Path analysis was performed to test the hypotheses about relationships between OHIS behaviors and CRC screening outcomes. Because the outcome variable was dichotomous and the mediator variables had non-normal distribution, a robust weighted least-squared estimator was employed  to allow for flexibility in analyzing ordinal data and accounts for violations of normality. Just-identified models were tested and were constructed with a probit distribution, with the probit function transforming outcome probabilities to z-scores from a standard normal distribution by the weighted least-squared estimator. Descriptive and bivariate analyses were performed using SPSS 22.0; path analysis was conducted using Mplus 7.4.
3.1. Descriptive statistics
Table 1 reports the descriptive statistics of the variables. The mean age of the participants was about 58 (SD = 6.72) years. The majority (60.8%) of the participants were female. In the study sample, about 14% reported a previous cancer diagnosis; about 55% reported a family member’s previous cancer diagnosis. For CRC screening, almost half (49.2%) of the participants reported having had colonoscopy, while 7.5% and 21.7% and reported having had FOBT and sigmoidoscopy, respectively. Among those having had CRC screening, about 38% (n = 68) reported any of CRC screening tests within the past year. Participants reported on-average three types of OHIS behaviors (SD = 2.49). Participants also reported moderate levels of information overload (M = 2.1, SD = 1.42) and health literacy (M = 2.1, SD = 1.01). The decisional balance scores of the participants were 0.72, .70, and 0.74 for FOBT, sigmoidoscopy, and colonoscopy, respectively.
3.2. The path model of CRC screening
The path analyses show that H1, H1a, H1b, and H2b in bronchitis study were supported for all CRC screening tests. By comparison, the analyses show that H2a was supported for sigmoidoscopy and colonoscopy, while H2 was not supported for any of the tests. r> Table 1