Perceived risk of disease plays a key role in health behaviors, making it an important issue for cancer-preventive behavior research. This study describes studies using complimentary methodologies to investigate the determinants of perceived risk of developing colorectal cancer. In study 1, questionnaires were mailed to a community sample of 1,056 adults ages 45 to 65 years. They assessed risk factors for colorectal cancer and perceived risk of colorectal cancer and asked respondents to explain why they had rated their risk as they had. Consistent with previous studies, risk factors for colorectal cancer were significantly correlated with perceived risk, although associations were relatively weak. The most frequently cited reasons for risk judgments were diet, family history, and symptoms/general health. Not surprisingly, correlations between each risk factor and perceived risk were stronger among participants who had attributed their risk to that factor. Study 2 used semi-structured interviews to explore individuals' explanations for their perceived risk of colorectal cancer. Adults (n = 18) ages 60 to 63 years volunteered a variety of explanations, but their experiences of cancer seemed to be most salient to their risk estimates. These studies suggest that attributions people make for their risk judgments are important in understanding how they conceptualize risk, but appreciation of experiential and affective influences may be useful in fully understanding perceptions of risk. (Cancer Epidemiol Biomarkers Prev 2007;16(4):694–702)

Perceptions of risk have been identified as part of the “motivational engine” behind many health-protective actions. Individuals who feel at lower-than-average risk of a disease are less likely to engage in preventive and detection behaviors (1-6). In the light of evidence that many adults identify themselves as at lower-than-average risk for cancers and other serious but preventable diseases, it is important to understand the determinants of risk perceptions (7, 8).

The present studies took perceptions of risk for colorectal cancer as the focus. Colorectal cancer is a serious health threat and the second most common cause of cancer death in the United Kingdom (9). A recent survey in the United States found that as many as 48% of men and women reported their risk of developing colorectal cancer as lower than average compared with only 14% who perceived it as higher than average (10). Even in Britain, which has a less optimistic culture, 17% of older adults estimated their risk of developing colorectal cancer as lower than average compared with 9% who rated it as above average (11). Attempts to modify unrealistically positive perceptions of risk for colorectal cancer have had limited success, with risk factor feedback failing to consistently influence risk judgments (12, 13). Thus, there have been calls for further work to examine the processes involved in the formation of risk perceptions and risk attributions (e.g., refs. 10, 13). Lipkus et al. (ref. 13, p. 564) argued that it is premature to try to develop interventions to influence risk perceptions via risk attributions without a full understanding of the processes involved, and Hay et al. (10) called for research to elucidate basic risk perception processes. The aim of the present work was to use mixed methods (quantitative and qualitative) to explore these processes.

Previous studies have examined demographic and psychosocial correlates of colorectal cancer risk (10, 14-17), with our study (11) being the first to examine correlates of perceived risk for colorectal cancer in a population-based sample. We found that higher perceived risk was associated with a family history of colorectal cancer, poorer subjective health, more bowel symptoms, and higher anxiety. Smoking and not exercising were also related to higher perceptions of risk, whereas being male and older were associated with lower perceived risk. A limitation of that analysis was that we did not have data on diet (fruit, vegetables, and red meat), alcohol consumption, or body weight, which are known risk factors for colorectal cancer (18) and may also be important in understanding perceptions of risk. A recent study assessed a range of predictors of perceived risk for colorectal cancer in a national probability sample of U.S. adults (10). Overall, the results replicated our findings, although gender and smoking were not significantly related to perceived risk. Fruit and vegetable consumption and body weight were unrelated to perceived risk of colorectal cancer, but associations with red meat or alcohol were not examined. Given these inconsistencies and gaps in the existing data on correlates of perceived risk for colorectal cancer risk, further exploration is warranted to confirm and expand this evidence and obtain a more complete picture of the factors influencing conceptualizations of vulnerability to colorectal cancer.

In the present quantitative study, we not only included a broad range of factors hypothesized to influence risk perception, but we also asked about the reasons for risk judgments. Many studies of attributions of colorectal cancer risk (13, 15, 17) have focused on the five groups of determinants first described by Weinstein (19): actions and behavior patterns (e.g., diet and smoking), heredity (family history), physiology or physical attributes (e.g., stomach problems), psychological attributes (e.g., “I feel that I could get it”), and environmental factors (e.g., “things they put in food”). Lipkus et al. (15) concluded that the majority of their predominantly African-American sample attributed their risk judgment, whether high or low, to “psychological” factors such as “just feeling like you could get it” (35%). This was followed by heredity (20%), personal actions (17%), and physiologic factors (12%), with few citing environmental (0.005%) reasons. Blalock et al. (17) found that among a sample of adults with a first-degree relative with colorectal cancer, physiology was mentioned most frequently (27%), closely followed by heredity (25%) and personal actions (16%) as determinants of risk, whereas first-degree relatives of surgical patients cited personal actions and physiology with equal frequency (27%) and heredity less often (10%). In contrast to Lipkus et al.'s (15) results, psychological factors were not mentioned with sufficient frequency for statistical analysis. The differences could be related to ethnicity; Lipkus et al.'s participants were predominantly African-American and lower socioeconomic status, whereas Blalock et al.'s were predominantly White, but differences in family history could play a part. The two studies were conducted on specific population subgroups: siblings of colorectal cancer and surgical patients (17) and African Americans (15). No previous study has examined attributions of perceived risk in a population-based sample.

A further aim of the quantitative study was to explore associations between each individual's standing on a given risk factor and perceived risk among those who identified that risk factor as personally relevant. This is important because communications aimed at changing perceptions of personal vulnerability often depend on informing people of the “known” risk factors for the disease (20, 21). If risk perceptions are determined by influences other than these “known” risk factors, interventions may fail to effectively modify risk perceptions.

Quantitative research using a questionnaire format may limit respondents' ability to express their beliefs about their risk. A qualitative methodology provides a richer and fuller account of people's beliefs about their risk, which may shed more light on the determinants of perceived risk. Two qualitative studies (22, 23) examined impediments to colorectal cancer screening and touched on perceptions of risk. However, to the best of our knowledge, no previous study has examined attributions of perceived risk for colorectal cancer through in-depth interviews. A qualitative approach can discover whether the same attributions are found as in quantitative research and also identify additional important influences on perceived risk.

The present studies build on previous research (10, 11, 14-17) and are the first to take a mixed methods approach to better capture how people conceptualize their perceptions of vulnerability to colorectal cancer. The unique aims are (a) to examine a range of correlates of perceived risk for colorectal cancer in a community sample, (b) to describe the reasons people give for their risk judgments in a large community sample, (c) to examine the relationship between people's standing on a given risk factor and perceived risk among those attributing their risk to that factor, and (d) to explore through in-depth interviews the reasons people give for their risk attributions.

Study 1

Questionnaires (n = 1,056) were mailed to adults ages 45 to 67 years from the patient lists of two General Practices (Primary Care) in South-West England as part of a study of public awareness of colorectal cancer risk factors. At the time of the study, colorectal cancer screening was not part of the U.K. National Screening Programme; therefore, none of the participants would have previously been screened.

Measures

Perceived Risk. There is no “gold standard” for assessing perceived risk (24). Numerical absolute perceived risk measures have the disadvantage that there is poor numeracy in the general population (e.g., ref. 25). Verbal absolute measures avoid numeracy problems, but words such as “likely” can be interpreted in a variety of ways (26, 27). A comparative perceived risk question was used in the current studies because it tends to be more strongly correlated with actual risk factors (28). The question was “Compared to others of the same sex and age, my chances of getting bowel cancer1

1

In the United Kingdom, bowel cancer is a more widely used term than colorectal cancer.

are: much below average; below average; average; above average; much above average; have had bowel cancer.” The same item was used in many of Weinstein's studies (e.g., ref. 8). This was followed with “Why have you rated your chance of getting bowel cancer in this way,” based on the question used by Blalock et al. (17). The open-ended question was coded into the categories devised by Weinstein (19) and also in terms of whether each factor was presented by the respondent as risk increasing or risk decreasing.

Demographic Characteristics. Age and gender were known from the lists provided by the Practices. Simple items were used to assess ethnicity and marital status. Socioeconomic deprivation of the area of residence was established from post code (zip code) data for each participant's area of residence (Townsend Material Deprivation Index; ref. 29).

Health Behaviors. Health behaviors were assessed using items derived from the European Health Survey (30). Smoking was assessed with the following question: “Please tick the box that best describes your smoking habits: never-smoker/non-smoker; ex-smoker; smoker.” Diet was assessed with three items: fruit, vegetable, and red meat intake [“On a typical day how many servings of the following would you eat: fruit (fresh, frozen or canned); vegetables (including salad, but excluding potatoes”); red meat (including beef, pork, lamb)]. There was one item on alcohol consumption: “In a typical week how many units of alcohol would you consume.” Physical activity was assessed with the following item: “During the past 7 days, on how many days did you: engage in vigorous activity that caused you to breathe much harder than normal and sweat (e.g. swimming, jogging, aerobics, football); engage in moderate activity that caused you to breathe somewhat harder than normal (e.g. cycling, gardening, dancing, brisk walking).” In addition, respondents also recorded the number of minutes per day. For statistical analysis, smoking status was categorized as smoker versus nonsmoker/never smoker/ex smoker, and physical activity was categorized as meeting U.K. activity recommendations (3 days of at least 20 min of vigorous activity, or 5 days of at least 30 min of moderate activity; ref. 31) or not. Diet and alcohol were treated as continuous variables.

Family History. Family history of which the respondent was aware was assessed with the following question: “Have any members of your family (BLOOD relatives, not relatives by marriage) had bowel cancer.”

Bowel Symptoms, Subjective Health, and Body Mass Index. Bowel symptoms over the past 3 months were assessed with a list of seven symptoms (constipation, hemorrhoids, diarrhea, wind, pain in abdomen, incontinence, and blood in stools). The subjective health question asked the following: “Would you say that for someone of your age your own health in general is: excellent; good; fair; poor” (32). Body mass index was calculated from self-reported height and weight.

Psychological Factors. Perceived control over bowel cancer was measured with the following item: “There are things I can do to control whether I get bowel cancer or not,” with the following response options: “strongly disagree, disagree, not sure, agree, strongly agree.” State anxiety was assessed using the shortened, six-item version of the Spielberger State Trait Anxiety Inventory (33, 34). Internal reliability of the STAI was high (Cronbach's α = 0.84).

Beliefs about Environmental Factors. Assessment of the “environment” domain included two attitudinal statements: “Environmental pollution may increase the risk of bowel cancer” and “The things that are added to food (e.g. additives and preservatives) may increase the risk of bowel cancer,” with the following response options: “strongly disagree, disagree, not sure, agree, strongly agree.”

Analysis

Results were analyzed using SPSS (version 10.1) and Stata. One-sample t tests were used to detect a significant deviation from zero for Townsend scores and comparative perceived risk. Spearman's ρ correlations were obtained between each individual's standing on a given risk factor and its perceived risk. The correlations based on the whole sample were compared with the correlations based on those who attributed their risk to that factor. The statistical significance of the differences between the correlations was obtained using bootstrap sampling with replacement using Stata statistical software.

Study 2

Study 2 used a qualitative method to examine in-depth people's accounts of their risk judgments. Individuals aged between 60 and 63 years were recruited from two urban General Practices that were due to be taking part in a trial of population-based, nurse-led flexible sigmoidoscopy screening for colorectal cancer. Individuals contacted for the present study were later invited to attend flexible sigmoidoscopy screening, although they were not aware that they would be invited at the time of the interview. None had been screened before. Potential participants (n = 60) were contacted by letter, giving them brief information and inviting them to return an opt-out form in a prepaid envelope if they did not wish to be contacted about the study. Six returned opt-out forms, and attempts were made to contact the remainder. Thirteen did not have a valid telephone number or were not contactable after five attempts; six no longer resided at the given addresses; two had inadequate English; and three, respectively, had died, were too ill, and had a learning disability. Thirty-six were therefore eligible for interview, of whom 10 declined when contacted, leaving 20 of 36 eligible participants in the sample. Ethical approval was granted by the Harrow Research Ethics Committee as part of the larger study of nurse-led flexible sigmoidoscopy screening.

Interviews took place from October to December 2003. They were semi-structured, used a topic guide, and took ∼40 min. The topic guide covered background demographic information, perceptions of risk for colorectal cancer, and explanations for risk judgments. At the start of the interview, participants were asked to estimate their comparative perceived risk for colorectal cancer on the standard scale: “Compared to others of the same sex and age, my chances of getting bowel cancer are: much below average; below average; average; above average; much above average” (8). They were then invited to explain how they had made their judgment. General prompts such as “Why is that” and “Can you tell me a little more about that” were used to get more detail, but potential reasons (e.g., family history and health behaviors) were not given as prompts.

Analysis

All participants agreed to the interviews being tape-recorded and transcribed. Content analysis was used to identify and quantify the main reasons given for risk judgments. Five randomly selected transcripts were studied in detail to identify key themes, and a coding sheet was developed. A detailed coding manual was written following the recommendations of Krippendorff (35) to make explicit how transcripts were to be coded. The first two transcripts were coded independently by two researchers, and 92% inter-rater agreement on coding categories was obtained. The κ inter-rater reliability statistic (36) was slightly lower at 0.72 but represented an acceptable level of agreement (37). In both cases, inter-rater reliability was calculated before differences were discussed. The remaining transcripts were coded by one researcher (K.R.). The second method (thematic analysis) was used to provide a more in-depth account of the themes and to explore differences between participants who viewed their risk as below average, average, or above average. Content analysis and thematic analysis complement one another because content analysis is more connected to the quantitative tradition with a strict coding scheme and tests of inter-rater reliability, whereas thematic analysis places more emphasis on qualitative analysis of these themes in context (38).

Study 1

Six hundred forty-eight people (61%) returned a completed questionnaire. The mean age of the sample was 55 years. More women (53%) than men (47%) returned the questionnaire. Ninety-eight percent were White, and 68% were married. Respondents were significantly more affluent than the national average, with the mean Townsend (deprivation) score (M = −1.44, SD = 2.56) deviating significantly below zero, which is the average score for England and Wales [t(632) = −14.19, P < 0.001].

Overall, 24.8% regarded their risk of developing bowel cancer as below average; 66% saw it as average; and 9% as above average, showing the usual optimistic bias [M = −0.16, SD = 0.56, t(635) = −7.06, P < 0.001]. Correlations between individuals' standing on the self-reported risk factors and perceived risk are presented in the first column of Table 1. Healthier behaviors were negatively correlated with perceived risk, with the exception of alcohol consumption. Bowel symptoms, poorer subjective health, and body mass index were positively associated with risk. Anxiety was associated with higher perceived risk, and perceived control was associated with lower risk. In terms of environmental factors, the belief that food additives increased risk was not significantly associated with perceived risk, but people who believed that pollution played a role were more comparatively optimistic. Among the demographic factors, only ethnicity was significantly related to perceived risk, with non-White respondents being more optimistic.

Table 1.

Study 1 relationship between perceived risk and demographic and psychosocial characteristics (higher scores = higher perceived risk)

Correlation between perceived risk and standing on risk factor (Spearman's ρ)Correlation between perceived risk and standing on risk factor among those attributing their risk to that factor
Family history, First Degree Relative only (0, no family history; 1, family history) 0.232* (n = 636) 0.551* (n = 128) 
Smoking (0, smoker; 1, ex smoker; 2, never smoked) −0.116* (n = 630) 0.532* (n = 39) 
Meeting physical activity recommendations (0, not meeting recommendations; 1, meeting recommendations) −0.088 (n = 636) −0.027 (n = 48) 
Red meat consumption (continuous variable) 0.182* (n = 617) 0.132 (n = 171) 
Vegetable consumption (continuous variable) −0.118* (n = 630) −0.204 (n = 171) 
Fruit consumption (continuous variable) −0.110* (n = 626) −0.116 (n = 170) 
Alcohol consumption (continuous variable) −0.019 (n = 623) 0.168 (n = 16) 
Bowel symptoms (0, 0-1 symptom; 1, 2-3 symptoms; 2, ≥4 symptoms) 0.171* (n = 636) 0.375* (n = 94) 
Subjective health (0, excellent; 1, good; 2, fair; 3, poor) 0.238* (n = 633) 0.442* (n = 94) 
Body mass index (continuous variable) 0.116* (n = 623) 0.121 (n = 9) 
Anxiety (continuous variable) 0.169* (n = 634) — 
Perceived control (0, strongly disagree to 4, strongly agree) −0.166* (n = 629) — 
Beliefs about environmental pollution (0, strongly disagree to 4, strongly agree) −0.082 (n = 631) 0.376 (n = 7) 
Beliefs about things added to food (0, strongly disagree to strongly agree) −0.003 (n = 632) 0.725 (n = 7) 
Demographic factors   
    Age (continuous variable) −0.043 (n = 636) 0.287 (n = 17) 
    Gender (0, female; 1, male) 0.062 (n = 636) — 
    Marital status (0, married; 1, not married) 0.032 (n = 631) — 
    Ethnicity (0, White; 1, non-White) −0.085 (n = 631) — 
    Socioecnonomic deprivation: Townsend score (continuous variable) −0.078 (n = 622) — 
Correlation between perceived risk and standing on risk factor (Spearman's ρ)Correlation between perceived risk and standing on risk factor among those attributing their risk to that factor
Family history, First Degree Relative only (0, no family history; 1, family history) 0.232* (n = 636) 0.551* (n = 128) 
Smoking (0, smoker; 1, ex smoker; 2, never smoked) −0.116* (n = 630) 0.532* (n = 39) 
Meeting physical activity recommendations (0, not meeting recommendations; 1, meeting recommendations) −0.088 (n = 636) −0.027 (n = 48) 
Red meat consumption (continuous variable) 0.182* (n = 617) 0.132 (n = 171) 
Vegetable consumption (continuous variable) −0.118* (n = 630) −0.204 (n = 171) 
Fruit consumption (continuous variable) −0.110* (n = 626) −0.116 (n = 170) 
Alcohol consumption (continuous variable) −0.019 (n = 623) 0.168 (n = 16) 
Bowel symptoms (0, 0-1 symptom; 1, 2-3 symptoms; 2, ≥4 symptoms) 0.171* (n = 636) 0.375* (n = 94) 
Subjective health (0, excellent; 1, good; 2, fair; 3, poor) 0.238* (n = 633) 0.442* (n = 94) 
Body mass index (continuous variable) 0.116* (n = 623) 0.121 (n = 9) 
Anxiety (continuous variable) 0.169* (n = 634) — 
Perceived control (0, strongly disagree to 4, strongly agree) −0.166* (n = 629) — 
Beliefs about environmental pollution (0, strongly disagree to 4, strongly agree) −0.082 (n = 631) 0.376 (n = 7) 
Beliefs about things added to food (0, strongly disagree to strongly agree) −0.003 (n = 632) 0.725 (n = 7) 
Demographic factors   
    Age (continuous variable) −0.043 (n = 636) 0.287 (n = 17) 
    Gender (0, female; 1, male) 0.062 (n = 636) — 
    Marital status (0, married; 1, not married) 0.032 (n = 631) — 
    Ethnicity (0, White; 1, non-White) −0.085 (n = 631) — 
    Socioecnonomic deprivation: Townsend score (continuous variable) −0.078 (n = 622) — 
*

Significant at P < 0.01.

The difference between the correlation for the whole sample and the correlation among those attributing their risk to that factor is significant at P < 0.05.

Significant at P < 0.05.

Table 2 lists the attributions reported in the open-ended question following the perceived risk judgment, which were coded into five factors, and identified as being reported as either increasing or decreasing personal risk. Many more factors were mentioned as risk decreasing (n = 413) than risk increasing (n = 118). Diet was the most frequently mentioned risk-decreasing factor followed by family history and symptoms/general health. In terms of risk-increasing factors, symptoms and general health were mentioned most frequently followed by family history, smoking, and diet.

Table 2.

Study 1 attributions given to explain personal perceived risk estimates

Total (n = 636)Below average (n = 158)Perceived risk average (n = 420)Above average (n = 58)
Risk decreasing     
    Actions and behavior patterns     
        Diet 152 100 51 
        Smoking 26 20 
        Exercise 45 31 14 
        Alcohol 14 
        Weight 
        Lifestyle 23 11 12 
    Heredity 90 52 38 
    Physical/physiology     
        Age 
        Symptoms/general health 47 26 19 
    Psychological factors 
    Environment 
Total risk decreasing 413 262 145 
Risk increasing     
    Actions and behavior patterns     
        Diet 
        Smoking 12 
        Exercise 
        Alcohol 
        Weight 
        Lifestyle 
    Heredity 34 10 24 
    Physical/physiology     
        Age 
        Symptoms/general health 38 19 18 
    Psychological factors 
    Environment 
Total risk increasing 118 53 62 
Total (n = 636)Below average (n = 158)Perceived risk average (n = 420)Above average (n = 58)
Risk decreasing     
    Actions and behavior patterns     
        Diet 152 100 51 
        Smoking 26 20 
        Exercise 45 31 14 
        Alcohol 14 
        Weight 
        Lifestyle 23 11 12 
    Heredity 90 52 38 
    Physical/physiology     
        Age 
        Symptoms/general health 47 26 19 
    Psychological factors 
    Environment 
Total risk decreasing 413 262 145 
Risk increasing     
    Actions and behavior patterns     
        Diet 
        Smoking 12 
        Exercise 
        Alcohol 
        Weight 
        Lifestyle 
    Heredity 34 10 24 
    Physical/physiology     
        Age 
        Symptoms/general health 38 19 18 
    Psychological factors 
    Environment 
Total risk increasing 118 53 62 

The second column of Table 1 presents correlations between specific risk factors and perceived risk among individuals who mentioned that particular risk factor in explaining their perceived risk judgment. These correlations were significantly larger than in the whole sample for family history, smoking, subjective health, and symptoms.

Study 2

Sixteen interviews took place in participants' homes, three at University College London, and one over the telephone because that was the individual's preference. Two interviews were excluded from the analyses (one White man aged 60 years and one Indian woman aged 61 years) because the tape recorder failed; thus, analyses were carried out on 18 interview transcripts.

The interviews for analysis were with 12 women and 6 men aged between 60 and 63 years, with a mean age of 62 years (SD = 1.27). The majority were White (n = 14) and married (n = 13). They had a range of educational qualifications from none at all (n = 2) to university degree level (n = 2). Seven were working full time; one cared for her terminally ill husband; seven were either semi-retired (n = 3) or retired (n = 4); and three described themselves as not working at present.

Comparative Perceived Risk and Explanations

Five participants rated themselves at lower-than-average risk (28%); 9 believed they were average risk (50%); and 4 thought they were at higher-than-average risk (22%) In response to being asked how they made their risk judgments, people mentioned a range of factors. Figure 1 shows the reasons given and indicates whether they were described as risk increasing or risk decreasing. Seven people initially said they “didn't know,” but five of them went on to give reasons for their estimates. The remaining two who responded with “don't knows” had both rated their risk as being “average” and were unable to give more detail: “Bowel cancer, I just don't know” (003). Overall, 47 attributions were mentioned: 32 risk decreasing and 15 risk increasing.

Figure 1.

Study 2 reasons given for comparative colorectal cancer risk judgments.

Figure 1.

Study 2 reasons given for comparative colorectal cancer risk judgments.

Close modal

Actions and Behavior Patterns. Health behaviors were the most frequently mentioned category in explaining risk judgments (see Fig. 1), among which, by far, the most cited factor was diet, with nine participants giving this as the first explanation. Diet was mentioned exclusively as a risk-decreasing factor: “Food-wise we eat quite healthy food…I've got a feeling, you know, that so far there's no problem so it's got to be below average” (016). Eating plenty of fruits, vegetables, fiber, or “roughage” and limiting intake of red meat and fatty foods were all mentioned as reducing risk. No one cited diet as a cause of higher risk.

Exercise was mentioned by two people as reducing their chances of getting colorectal cancer. One said: “I'm very fit and I do conservation work at the weekends and so my lifestyle is better than most people who are sedentary these days” (005). This quote also shows the respondent making a downward social comparison by comparing himself to a stereotyped view of others who are inactive. No one cited inactive lifestyle as a risk increasing factor.

Alcohol was mentioned by two respondents. One believed that because he drank in moderation, this would decrease his risk: “I do drink but not to excess; I probably drink weekends but I don't drink mid-week” (009). One believed that her drinking (and smoking) would increase her risk; she was the only respondent to mention smoking: “I would say with my lifestyle I would be a person at risk; definitely, without a doubt. Because I smoke, I drink. Classic!” (012). People were not asked directly about smoking, and although 10 people mentioned being nonsmokers, they did not refer to this when explaining their risk estimate.

Sexual behavior was given as an explanation by one person. She believed because she had had only one sexual partner for 30 years, this reduced her risk of developing colorectal cancer: “…it might sound silly but if you go around with a lot of like men and so on, ever since 30 years or whatever I've been with my husband, I don't stray with you know Tom, Dick or Harry, I just stay put” (001).

Heredity. Heredity or family history was given as an explanation for risk judgments by seven people and mentioned as both risk increasing (n = 3) and risk decreasing (n = 4). Among those citing it as risk increasing, two had a family history of colorectal cancer: “It's been in our family” (015). A third person who mentioned family history as increasing her chances did not have a family history of colorectal cancer specifically but had a history of other cancers and felt at risk of all types of cancer. She also had a theory that she was particularly at risk: “Because my mum is alive and her brother is alive and they have nothing diagnosed. Cancer has skipped a generation so the next generation will be me and my children…I think we're a risk, very high, with any form” (014). People referring to family history as a risk-decreasing factor spoke of the absence of colorectal cancer within their family: “It doesn't run in the family” (017).

Physiology or Physical Attributes. Symptoms and general health were frequently given as reasons for risk judgments (n = 10). Two respondents had been through a previous colorectal examination for symptoms in which they felt they had been given the “all clear.” An absence of symptoms and “feeling well” were commonly given reasons for risk judgments and were mentioned by eight participants as reducing their risk: “I've never had any problems with that area of myself, ever” (004); “I've got very good bowel movements, no stomach pains” (016). Only two people mentioned symptoms as increasing their risk. One had a history of diverticular disease, and the other said, “Yes I have trouble. Mostly with the kidneys, but I've also had trouble with the bowels a bit.” (015).

Age was mentioned by four people. In three of the four cases, it was regarded as risk increasing: “if you live a bit longer, your risk of getting it is probably higher” (013). The one person who saw age as decreasing her risk gave a rather confused explanation: “So the older you get, we're hoping the chances are going to get less. We know that's not true but, that's my form of thinking” (017).

Environmental Factors. Environmental influences were cited four times. One respondent believed that because he spent a lot of time outdoors, his chances of developing colorectal cancer were reduced: “spending plenty of time outside which I think gives you immunity to most ills generally” (005). The other three respondents spoke about diet. One believed chemicals in food increased her risk: “I mean we don't know what manufacturers are putting into the food.” (014). Another believed that because she avoided eating tinned (canned) food, this reduced her risk: “There is things in tins that can really mess up your health really. And ever since I was growing back home in Jamaica, that's where I'm from, we don't eat a lot of tinned food. Anything we eat is sort of fresh” (001). A fourth person mentioned eating organic food: “I try to have all these organic foods nowadays” (009).

Psychological Factors. True “explanations” in the psychological domain were rare. However, three respondents stated that they “just hoped” they wouldn't get bowel cancer: “I hope I've been realistic, I would have hoped unlikely” (019); “I'm hoping below average” (009). These responses were coded as “psychological” because previous studies have categorized similar expressions (e.g., “trust that I won't get it”) this way (11). However, the impression was that those who “hoped” they wouldn't get cancer were either using a common expression with little weight behind it or making a statement of faith rather than believing that a positive attitude would keep them well.

Chance. Chance was given as an explanation in four instances, all for higher risk. In each case, it served as a counterpoint following on from discussion of risk-reducing factors: “I'm not saying it can't happen because I mean, anything can happen to everybody” (001).

Thematic Analysis

To understand better why respondents made the risk judgments they did, we looked beyond the factors mentioned explicitly and considered other potential influences such as closeness to family members who had suffered from cancer and experience of cancer through friends.

Family History. People who saw their risk as lower than average were less likely to report having a family history of cancer, and those two who had a family history distanced themselves from the risk. In talking about her mother having stomach cancer: 017 said, “Well she, she had MS. I mean she died in…….about 30, I'm trying to think back, it's so many years ago, about 38 years ago……but she had MS as well so whether, you know, I often wondered was, you know does a sedentary life, she never walked for 26, 28 years, is it a sedentary life kicks off the other things, you know, if you're not as healthy as you could be” (017). This woman acknowledges few similarities between her own and her mother's health and regards her mother's sedentary life and compromised health as the main cause of her stomach cancer. The second described how his father died 42 years ago: “He died of cancer of the liver, and well his body was racked with cancer, but he smoked very heavy. But at the same time he was one of the misfortunate ones that was in the First World War and he had tear (gas), he had mustard gas and he was a very sick man” (009). Neither respondent considered genetic susceptibility apparently because they had clear explanations why their relative developed cancer and why these explanations do not apply to them personally. In addition, both relatives died many years ago.

Two respondents had a family history of bowel cancer, and both reported their risk as being higher than average. Among the other two participants reporting higher-than-average risk, one had a history of Hodgkin's disease, and the other had a strong family history of cancer. Both respondents with a family history of bowel cancer mentioned this as a risk-increasing factor and mentioned genes: “My father died of bowel cancer so going on the hereditary factor, I suppose I'm probably a bit more at risk” (020). The other two did not cite family history directly in explaining their perceived risk but mentioned it at a later point in the interview. As one woman described: “My father died of cancer….the genes of my father go through the females in our family” (012).

Only one person who rated her risk as “average” mentioned family history, but in the interview, she described her family history of cancer as increasing her chances of developing bowel cancer, which suggests she had either neglected it in her original risk judgment or gave the wrong judgment. Five others who rated their risk as “average” and had a family history of cancer did not mention this as a factor in explaining their risk. One reason why a family history may not influence perceived vulnerability is the distance (temporal or geographic) from the affected individual. In one case, the respondent's relatives lived in Ireland, so she had little experience of their illness, whereas another described his uncle dying between 30 and 40 years ago. Another who seemed to have been close to the relative with cancer did not feel at risk for bowel cancer because breast and cervical cancer were the ones she felt most vulnerable too, despite none of her relatives having had these. One person had a striking family history, with her sister dying of cervical cancer in her 50s, her brother dying of lung cancer, and her mother having cancer of the kidney. Nevertheless, this woman said, “from a genetic point of view, I don't think I'm overly prone” (010). The interviewer asked her directly if she felt at risk of cervical cancer because of her sister's history, and she replied, “No, I don't think so. I mean, her lifestyle was different from mine anyway.” She identified the cause of her brother's lung cancer as smoking, which she had given up many years ago: “He was a very heavy smoker so you know, that's the main reason”. In both instances, the respondent sees herself as dissimilar to her affected siblings. In describing her mother she says, “Oh my mother actually had cancer when she was in her late seventies. One of her kidneys was affected and she had it removed. But as I said, she was in her late seventies and, she survived for another 6 years after that and she didn't die of cancer”. Again, it seems that the mother's age and the fact that cancer did not kill her protected the respondent from feeling vulnerable.

Experience. A second influence that seemed pertinent to perceived risk was experience with cancer. “Experience” was conceived as capturing not only contact with people with cancer but also the nature of the experience. On the whole, comparative optimists had little direct experience of cancer. One said she did not know anyone who had had cancer, and her only experience was through what she saw on television and read in the newspaper (surprising in a woman aged 60 years). Two mentioned knowing people who had bowel cancer. In one case, it was a friend who the respondent described as eating a diet of “rubbish,” and in the other, it was a work colleague who was described as coping very well with the illness and having been in remission for over 3 years. In the first case, the respondent contrasted her own good diet and her friend's poor diet to explain why she was not personally vulnerable. In the second case, the respondent's work colleague presented a relatively encouraging image of bowel cancer that may have led him to feel less prone.

People who regarded their risk as higher than average had very different experiences. One man described the death of his uncle from bowel cancer as being “particularly tragic, it upset everybody” (015). He also described his mother-in-law's death from bowel cancer: “Poor women died in absolute agony, it was terrible. Went on months and months” (015). He presented two very vivid and traumatic images of people he had known who had died of bowel cancer, which seemed to be influencing how he feels about his vulnerability to bowel cancer. Another woman who had lost two friends to bowel cancer explained the following: “I've lost dear friends……when I look at their lifestyle and my lifestyle and everything, I can definitely see a connection” (012). It seemed that the similarity in lifestyle between herself and her friends led this respondent to feel particularly vulnerable.

Study 1

Overall, associations between individual's standing on a range of factors and their perceived risk for colorectal cancer replicated previous work (10, 11). In this study, we were particularly interested in examining relationships between perceived risk and diet, alcohol, and body weight because these factors were not included in our previous analyses (11). Unlike Hay et al., we found that lower fruit and vegetable consumption and higher body weight were (correctly) associated with higher perceived risk for colorectal cancer. Higher consumption of red meat was also (correctly) positively correlated with perceiving greater risk, whereas alcohol consumption was unrelated to perceived colorectal cancer risk.

In terms of people's reasons for their risk judgments, the results indicated that attributions for colorectal cancer risk in this community-based sample fell into the five-factor framework first described by Weinstein (19). Overall, respondents cited many more risk-decreasing than risk-increasing attributions, and this was particularly true for “actions and behavior patterns,” as previous studies have also found (17, 19). Not surprisingly, individuals who viewed their risk as “below-average” or “average” cited many more risk-decreasing reasons than risk-increasing reasons. The most frequently mentioned factors were diet, family history, and symptoms, making the results more consistent with Blalock et al.'s (17) pattern of responses than with Lipkus et al.'s (15), where the most frequent response was in the “psychological” category. This may be due to Lipkus et al.'s sample comprising predominantly low-income, African-American participants, although whether ethnicity or lower socioeconomic status give rise to this differential response is a subject for future work.

To the best of our knowledge, the present study is the first to look also at associations between risk factors and perceived risk specifically among people who have attributed their risk judgment to that factor. Taking this approach significantly strengthened associations between perceived risk and family history, smoking, bowel symptoms, and subjective health. Differences between the correlation for the whole sample and those attributing their risk to that factor were not significant for consumption of red meat, vegetables, fruit, and alcohol or exercise. This may be because people tend to be biased in their assessment of behavioral factors (other than whether or not they smoke) and often perceive themselves to be in a healthier behavioral category than they really are.

Study 2

This study showed that the main reasons for people's risk judgments were similar to those found in study 1, with the three most commonly cited reasons being diet, family history, and bowel symptoms. Again, respondents mentioned many more risk-decreasing than risk-increasing factors, providing further evidence of attribution bias. Other work has noted that the public tend to believe that “others” eat a poor diet, whereas they themselves eat pretty well (39). Such biases have important implications for work on health behavior change because unless people recognize limitations in their own health behaviors, they are unlikely to take steps to improve them. Future work might consider how to help people recognize that behaviors such as diet or activity may be suboptimal.

Our interpretation was influenced by the theoretical framework developed by Walter et al. (40), which describes the way people with a family history create and manage their personal sense of vulnerability. The core constructs are salience, personalizing process, and personal vulnerability. Our results suggested that having a positive family history or knowing someone with cancer were particularly salient factors in determining perceived risk, especially if the experience was traumatic, although in some cases, contrasts in lifestyle were used to create distance. Those without a salient example in their own experience were more likely to consider personal health behaviors in drawing conclusions about vulnerability, and as other studies have shown, these were almost always in a self-serving direction (1, 17, 19).

General Discussion

The purpose of this study was to examine the determinants of perceived risk for colorectal cancer. A mixed-methods approach was taken to better elucidate the processes involved. In study 1, a broad range of correlates of perceived risk for colorectal cancer were examined in a community sample of older adults. The results suggested that, in general, people were appropriately taking their standing on risk factors into account when judging their personal risk (e.g., poorer health behaviors were associated with higher perceived risk). However, consistent with other studies (e.g., refs. 10, 14), the associations were relatively weak probably because people weigh the reasons for their risk estimates in an idiosyncratic fashion; therefore, the correlation with perceived risk for the group is weak. Furthermore, any one risk factor may only be salient for a few people (14).

The second goal of study 1 was to describe the attributions people made for their risk judgments, and this was the first study to explore this in a population-based sample. Most participants offered rational explanations for their risk judgments, but they showed a strong positive bias, with 78% of attributions being risk decreasing. Another unique aspect of study 1 was that we were able to assess the relationship between risk factors and perceived risk specifically among people who had attributed their risk to that factor. This generated stronger associations, but there was still a good deal of unexplained variance, suggesting that people attribute their risk to several factors, or that other (unmeasured) factors are important. Study 2 confirmed that the same types of attributions were being made as seen in study 1, but an in-depth analysis pointed to the importance of the nature of people's experience of cancer in understanding their sense of risk from colorectal cancer.

In considering these results, it is useful to draw on Slovic et al.'s (41) work, which developed Epstein's (42) distinction between rational and experiential thought systems, with the rational or “analytic” system basing decisions on a conscious appraisal of the situation, whereas the experiential/emotional system uses associative connections that are more likely to be subconscious and automatic. Loewenstein et al. (43) have similarly hypothesized that “feelings” may dominate during decision making. If the experiential system is important in risk perception, an appreciation of both the analytic/rational and the experiential/affective factors will be required to understand how people estimate personal risk. Attributions that people make following their risk judgment may predominantly access the analytic system. Study 1 showed that anxiety was significantly related to perceived risk, and other studies have reported similar results (10, 11), indicating that emotional state plays a role in risk perception. However, this influence was not mentioned by a single participant in the quantitative or qualitative studies most likely because they were unaware of its effect. We had not necessarily expected people to be able to report that anxiety led them to feel at increased risk, but the observation highlights the limitations of relying on individual, self-reported attributions to understand the determinants of risk perception.

From our analyses, we propose that when people are asked to provide explanations for their risk judgments, their responses primarily reflect the analytic thought system. Analytic explanations certainly account for some of the variance in perceived risk (as shown in study 1), but the results from the qualitative analysis show that the experiential/emotional system also plays a role. Thus, when asked to explain their risk, people offer a range of analytic/rational explanations, but these may not wholly capture the processes involved in reaching their decision; they reveal only part of the story. Because the emotional thought system is automatic and probably subconscious, people are unable to report such influences in explaining their risk judgments. Indeed, it was only through conducting in-depth analysis of what people said during an extended interview that it became clear that personal experiences of cancer were influencing their risk judgments.

Figure 2 incorporates elements of previous work (19, 40-42) to illustrate how people may simultaneously draw on experiential/emotional and analytic thought systems in judging their personal risk for colorectal cancer. Zajonc (44) proposed that affective reactions may occur first, and we suggest that the subsequent analytic process of weighing up risk factors will be perceived through this initial experiential or emotional filter, although the two systems will continually interact. We propose that the experiential/emotional system will be activated by the factors listed in Fig. 2 (cancer salience, experience of cancer, and temperamental characteristics) as an initial assessment. If a person has a strong response from the experiential/emotional system due to their experience of cancer, a temperamental characteristic (e.g., anxiety), or high cancer salience, this system may dominate their risk judgment rather than the analytic system. If an individual has little experience of cancer and is low on cancer salience, anxiety, or cancer fear, they will progress to the analytic system to consider their standing on the factors listed in Fig. 2 (heredity, physical/physiology, actions, and behavior patterns). Our results suggest that in the absence of a family history or relevant symptoms, people turn to consideration of their behavior patterns, and this will most likely be in a self-serving manner. We would predict greater interaction between the two systems for people who either have a family history of colorectal cancer or symptoms because these more “analytic” factors will likely influence cancer salience. Temperamental characteristics may also moderate emotional/experiential mechanisms, so that a person with an optimistic disposition will regard their family history of cancer less severely than a pessimistic individual. These suggestions are speculative, and further careful analyses and refinements are necessary before a model of sources of risk perceptions for colorectal cancer can be established.

Figure 2.

Model of sources for perceived risk of colorectal cancer.

Figure 2.

Model of sources for perceived risk of colorectal cancer.

Close modal

Attempts to change perceptions of risk have been largely unsuccessful (45),2

2

K.A. Robb et al. Impact of risk information on perceived colorectal cancer risk: a randomized trial. Submitted for publication.

even when people are provided with tailored information about their own colorectal cancer risk (12, 13). The current findings provide some insights into this by indicating that attributions only reveal one component of people's conceptualization of risk. That is not to say that risk attributions and known risk factors should be ignored in attempts to modify risk perceptions because we have shown in study 1 that they correlate to an extent with risk perceptions. Indeed, a recent study that presented participants with social comparison risk factor feedback (telling participants that their total number of colorectal cancer risk factors was greater than the average or was not greater than average) found that optimistic comparative risk perceptions could be reduced (46), which is a promising finding that warrants further testing. However, giving analytic, risk factor–type information alone may not effectively influence perceptions of vulnerability, and engaging the emotional system may be a more powerful means of helping people to recognize their personal vulnerability and motivating them to adopt cancer-preventive behaviors.

Biases and heuristics have long been recognized as influential in judgment making (47), and in a recent review, Peters et al. (48) concluded that recognition of these heuristics could improve cancer communications. They discussed Katie Couric's televised colonoscopy following her husband's death from colorectal cancer as an example of how availability and effect heuristics can be successfully accessed to motivate cancer protective behavior (in this case, colonoscopy) following the publicity (49). Similarly, there was an increase in breast cancer screening in Australia after Kylie Minogue's breast cancer diagnosis (50). Observations such as these are testament to the importance of experiential or emotional systems in risk perception. Perhaps, there is something about the presentation of cancer stories in the media that effectively accesses the experiential system. Future work might consider how publicity about a disease motivates people to act, in order that the “active” ingredients can be more effectively harnessed for public good.

There are a number of limitations to these studies. In study 1, the risk factors were assessed by self-report, and the design was cross-sectional. However, given that colorectal screening was not available in the United Kingdom at the time of data collection, and the level of public awareness of risk factors is low (51), it seems unlikely that people's perceptions of risk for colorectal cancer could have influenced their risk factors. Several of the measures were based on single items (e.g., perceived control and beliefs about the environment); thus, the findings are less reliable than if validated scales had been used. However, we were reluctant to burden participants with lengthy scales. A further limitation was that participants could have read the questions on risk factors for bowel cancer that appeared in the questionnaire before answering the risk perception question, and this may have influenced their risk judgment. The questionnaire was structured so that the risk perception question and the reason for their risk judgment item were at the beginning to limit this, but we cannot be certain that this worked in all cases. However, the types of explanations were the same in the qualitative study where participants had no access to potential “cues,” which suggests that this was not a major problem. Study 2 was limited because respondents represented only those who agreed to be interviewed, and because the age group was restricted. However, qualitative research is not usually concerned with making generalizations about the sample population but, rather, with identifying important themes. Nonetheless, it is possible that respondents differed in important dimensions to those who declined to participate.

The current research has examined correlates of risk perception and the attributions people make about their perceived risk of colorectal cancer in a large community-based sample and in a series of in-depth interviews. The results suggest that whereas the attributions that people make for their risk judgments are important in understanding how they conceptualize their risk, greater appreciation of other, particularly affective, influences on perceived risk may prove to be useful in developing risk communications.

Grant support: Medical Research Council and Cancer Research UK.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Prof. John Campbell and Dr. Philip Evans of Peninsula Medical School for their assistance with study 1, Dr. Alice Simon for her help with study 2, and the two anonymous reviewers for their helpful comments.

1
Aiken LS, Fenaughty AM, West SG, Johnson JJ, Luckett TL. Perceived determinants of risk for breast cancer and the relations among objective risk, perceived risk, and screening behavior over time.
Women Health
1995
;
1
:
27
–50.
2
Avis NE, Smith KW, McKinlay JB. Accuracy of perceptions of heart attack risk: what influences perceptions and can they be changed?
Am J Public Health
1989
;
79
:
1608
–12.
3
Cummings KM, Jette AM, Brock BM, Haefner DP. Psychosocial determinants of immunization behavior in a swine influenza campaign.
Med Care
1979
;
17
:
639
.
4
Kahn JA, Goodman E, Slap GB, Huang B, Emans SJ. Intention to return for Papanicolaou smears in adolescent girls and young women.
Pediatrics
2001
;
108
:
333
–41.
5
McCaul KD, Branstetter AD, Schroeder DM, Glasgow RE. What is the relationship between breast cancer risk and mammography screening? A meta-analytic review.
Health Psychol
1996
;
15
:
423
–9.
6
Weller DP, Owen N, Hiller JE, Willson K, Wilson D. Colorectal cancer and its prevention: prevalence of beliefs, attitudes, intentions and behaviour.
Aust J Public Health
1995
;
19
:
19
–23.
7
Weinstein ND. Unrealistic optimism about future life events.
J Pers Soc Psychol
1980
;
39
:
806
–20.
8
Weinstein ND. Unrealistic optimism about susceptibility to health problems: conclusions from a community-wide sample.
J Behav Med
1987
;
10
:
481
–500.
9
Quinn M, Babb P, Brock A, Kirby L, Jones J. Cancer trends in England and Wales 1950-1999. Studies on medical and population subjects. No. 66. London: Office for National Statistics; 2001; pp. 54–69.
10
Hay J, Coups E, Ford J. Predictors of perceived risk for colon cancer in a national probability sample in the United States.
J Health Commun
2006
;
11
Suppl 1:
71
–92.
11
Robb KA, Miles A, Wardle J. Demographic and psychosocial factors associated with perceived risk for colorectal cancer.
Cancer Epidemiol Biomarkers Prev
2004
;
13
:
366
–72.
12
Weinstein ND, Atwood K, Puleo E, Fletcher R, Colditz G, Emmons KM. Colon cancer: risk perceptions and risk communication.
J Health Commun
2004
;
9
:
53
–65.
13
Lipkus IM, Skinner CS, Green LS, Dement J, Samsa GP, Ransohoff D. Modifying attributions of colorectal cancer risk.
Cancer Epidemiol Biomarkers Prev
2004
;
13
:
560
–6.
14
Vernon SW, Myers RE, Tilley BC, Li S. Factors associated with perceived risk in automotive employees at increased risk of colorectal cancer.
Cancer Epidemiol Biomarkers Prev
2001
;
10
:
35
–43.
15
Lipkus IM, Rimer BK, Lyna PR, Pradhan AA, Conaway M, Woods-Powell CT. Colorectal screening patterns and perceptions of risk among African-American users of a community health center.
J Community Health
1996
;
21
:
409
–27.
16
Price JH. Perceptions of colorectal cancer in a socioeconomically disadvantaged population.
J Community Health
1993
;
18
:
347
–62.
17
Blalock SJ, DeVellis BM, Afifi RA, Sandler RS. Risk perceptions and participation in colorectal cancer screening.
Health Psychol
1990
;
9
:
792
–806.
18
Colditz GA, Atwood KA, Emmons K, et al. Risk Index Working Group, Harvard Center for Cancer Prevention. Harvard report on cancer prevention. Vol. 4: Harvard Cancer risk index.
Cancer Causes Control
2000
;
11
:
477
–88.
19
Weinstein ND. Why it won't happen to me: perceptions of risk factors and susceptibility.
Health Psychol
1984
;
3
:
431
–57.
20
Lipkus IM, Crawford Y, Fenn K, et al. Testing different formats for communicating colorectal cancer risk.
J Health Commun
1999
;
4
:
311
–24.
21
Lipkus IM, Green LG, Marcus A. Manipulating perceptions of colorectal cancer threat: implications for screening intentions and behaviors.
J Health Commun
2003
;
8
:
213
–28.
22
McCaffery K, Borril J, Williamson S, et al. Declining the offer of flexible sigmoidoscopy screening for bowel cancer: a qualitative investigation of the decision-making process.
Soc Sci Med
2001
;
53
:
679
–91.
23
Weitzman ER, Zapka J, Estabrook B, Goins KV. Risk and reluctance: understanding impediments to colorectal cancer screening.
Prev Med
2001
;
32
:
502
–13.
24
Diefenbach MA, Weinstein ND, O'Reilly J. Scales for assessing perceptions of health hazard susceptibility.
Health Educ Res
1993
;
8
:
181
–92.
25
Black WC, Nease RF, Jr., Tosteson AN. Perceptions of breast cancer risk and screening effectiveness in women younger than 50 years of age.
J Natl Cancer Inst
1995
;
87
:
720
–31.
26
Nakao MA, Axelrod S. Numbers are better than words. Verbal specifications of frequency have no place in medicine.
Am J Med
1983
;
74
:
1061
–5.
27
Budescu DV, Wallston TS. Consistency in interpretation of probabilistic phrases.
Organ Behav Hum Decis Process
1985
;
36
:
391
–405.
28
Woloshin S, Schwartz LM, Black WC, Welch HG. Women's perceptions of breast cancer risk: how you ask matters.
Med Decis Making
1999
;
19
:
221
–9.
29
Townsend P, Phillimore P, Beattie A, editors. Health and deprivation: inequality and the North. Kent: Croom Helm; 1988.
30
Wardle J, Steptoe A. The European Health and Behaviour Survey: rationale, methods and initial results from the United Kingdom.
Soc Sci Med
1991
;
33
:
925
–36.
31
Department of Health. At least five a week: evidence on the impact of physical activity and its relationship to health. London: Department of Health; 2004.
32
Prescott-Clarke P, Primatesta P. Health survey for England. London: The Stationery Office; 1996.
33
Marteau TM, Bekker H. The development of a six-item short-form of the state scale of the Spielberger State-Trait Anxiety Inventory (STAI).
Br J Clin Psychol
1992
;
31
:
301
–6.
34
Spielberger CD, editor. Manual for the State-Trait Anxiety Inventory STAI (Form Y). Palo Alto (CA): Consulting Psychologists Press; 1983.
35
Krippendorff K, editor. Content analysis: an introduction to its methodology. Newbury Park (CA): Sage; 1980.
36
Cohen J. A coefficient of agreement for nominal scales.
Educ Psychol Meas
1960
;
20
:
37
–46.
37
Landis JR, Koch GG. The measurement of observer agreement for categorical data.
Biometrics
1977
;
33
:
159
–74.
38
Joffe H, Yardley L. Content and thematic analysis. In: Marks D, Yardley L, editors. Research methods for clinical and health psychology. London: Sage; 2004. p. 56–68.
39
Sparks P, Shepherd R, Wieringa N, Zimmermanns N. Perceived behavioural control, unrealistic optimism and dietary change: an exploratory study.
Appetite
1995
;
24
:
243
–55.
40
Walter FM, Emery J, Braithwaite D, Marteau TM. Lay understanding of familial risk of common chronic diseases: a systematic review and synthesis of qualitative research.
Ann Fam Med
2004
;
2
:
583
–94.
41
Slovic P, Finucane ML, Peters E, MacGregor G. Risk as analysis and risk as feelings: some thoughts about affect, reason, risk and rationality.
Risk Anal
2004
;
24
:
311
–22.
42
Epstein S. Integration of the cognitive and the psychodynamic unconscious.
Am Psychol
1994
;
49
:
709
–24.
43
Loewenstein GF, Weber EU, Hsee CK, Welch N. Risk as feelings.
Psychol Bull
2001
;
127
:
267
–86.
44
Zajonc RB. Feeling and thinking: preferences need no inferences.
Am Psychol
1980
;
35
:
151
–75.
45
Weinstein ND, Klein WM. Resistance of personal risk perceptions to debiasing interventions.
Health Psychol
1995
;
14
:
132
–40.
46
Lipkus IM, Klein WMP. Effects of communicating social comparison information on risk perceptions for colorectal cancer.
J Health Commun
2006
;
11
:
391
–407.
47
Tversky A, Kahneman D. Judgment under uncertainty: heuristics and biases.
Science
1974
;
185
:
1124
–31.
48
Peters E, McCaul KD, Stefanek M, Nelson W. A heuristics approach to understanding cancer risk perception: contributions from judgment and decision-making research.
Ann Behav Med
2006
;
31
:
45
–52.
49
Cram P, Fendrick AM, Inadomi J, Cowen ME, Carpenter D, Vijan S. The impact of a celebrity promotional campaign on the use of colon cancer screening: the Katie Couric effect.
Arch Intern Med
2003
;
163
:
1601
–5.
50
Chapman S, McLeod K, Wakefield M, Holding S. Impact of news of celebrity illness on breast cancer screening: Kylie Minogue's breast cancer diagnosis.
Med J Aust
2005
;
183
:
247
–50.
51
McCaffery K, Wardle J, Waller J. Knowledge, attitudes, and behavioral intentions in relation to the early detection of colorectal cancer in the United Kingdom.
Prev Med
2003
;
36
:
525
–35.