Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Nov 1.
Published in final edited form as: Psychol Health. 2019 Apr 23;34(11):1294–1313. doi: 10.1080/08870446.2019.1604954

Physical activity: The relative associations with cognitive and affective risk beliefs

Eva Janssen a, Erika A Waters b
PMCID: PMC6801035  NIHMSID: NIHMS1527629  PMID: 31012749

Abstract

Objective:

The role of affectively-oriented risk beliefs in explaining health behaviors has not been examined in the context of physical activity or in diverse study populations. We evaluated whether affective risk beliefs account for unique variance in physical activity intentions and behavior above and beyond that accounted for by cognitive risk beliefs.

Design:

A cross-sectional survey of socio-demographically diverse U.S. residents (N=835; 46.4% no college training; 46.7% minority racial/ethnic ancestry; 42.6% men).

Main outcome measures:

Physical activity intentions and behavior.

Results:

Hierarchical linear regressions showed that affective risk beliefs accounted for variance in physical activity intentions beyond that predicted by socio-demographics and cognitive risk beliefs (F-change ps<.001). Specifically, intentions were higher among people with higher anticipated regret (ps<.001) and with higher absolute feelings of risk (ps<.05) or worry (ps<.05). There was an indirect relationship between perceived absolute likelihood and intentions through anticipated regret and feelings of risk. Neither cognitive nor affective risk beliefs accounted for variance in physical activity behavior (F-change ps>.05), but unsurprisingly, behavior was positively associated with physical activity intentions (p<.001).

Conclusion:

Future interventions could target affective risk beliefs – particularly anticipated regret – to increase intentions, and then add other intervention components to bridge the intention-behavior gap.

Keywords: affect, beliefs, risk perception, theory, physical activity

Introduction

A key precursor of volitional health behavior change is considering oneself at risk of experiencing a negative health outcome (Edwards, 1954; Janz & Becker, 1984; Rogers, 1975). Behavior change theories that extend from subjective expected value theory (Savage, 1954) often regard risk perceptions as largely cognitive judgments of the likelihood of a particular health problem, independent from affect (e.g., Ajzen, 1985; Becker, 1974; Fishbein & Ajzen, 1975; Schwarzer & Fuchs, 1995). However, foundational risk perception research (Paul Slovic, Fischhoff, & Lichtenstein, 1980) and modern perspectives (Cameron, 2003; Ferrer, Klein, Persoskie, Avishai-Yitshak, & Sheeran, 2016; Finucane, Alhakami, Slovic, & Johnson, 2000; Loewenstein, Weber, Hsee, & Welch, 2001; Sheeran, 2014) assert that risk perceptions are multifaceted constructs that are comprised of both cognitions and affect.

Cognitive Risk Beliefs

Cognitively-oriented aspects of perceived risk include different types of likelihood judgments. For example, perceived absolute likelihood is an individual’s judgment about the chances that he or she will develop a particular disease. In contrast, perceived comparative likelihood is an individual’s judgment about the changes that he or she will develop a disease relative to another person, group of people, or “the average person.” Both types of judgments are often viewed as cognitive in nature (e.g., Chapman & Coups, 2006; Ferrer et al., 2016) and so for clarity, we refer to them as cognitive risk beliefs when we refer to them collectively in this paper. Although people with higher absolute likelihood perceptions may show higher engagement in some preventive health behaviors (e.g., Weinstein et al., 2007), absolute likelihood perceptions can be less predictive than comparative likelihood perceptions when they are examined in the same statistical model (Dillard et al., 2011).

Affective Risk Beliefs

Conceptual and empirical work suggests that risk perceptions are formed, in part, by affect associated with one’s past experiences (Bechara, Damasio, Tranel, & Damasio, 1997; Damasio, 1994; Finucane et al., 2000; Loewenstein et al., 2001; Paul Slovic et al., 1980). The result is a feeling state that occurs automatically and rapidly in response to a stimulus. This feeling state provides an intuitive sense of the overall “goodness” or “badness” of a stimulus. Damasio (1984) refers to this intuitive state as “feelings.” Such feelings motivate behaviors that reduce bad feelings and encourage good feelings. For example, an individual’s real or virtual past negative experiences with cancer may produce automatic negative feelings about cancer when that individual is asked to appraise their cancer risk. These “feelings of risk,” which arise automatically and without conscious thought or effort, may or may not be consistent with cognitive perceptions of likelihood that are based on a systematic evaluation of risk factors.

A small amount of research has begun using survey methods to evaluate the extent to which feelings of risk motivate health behaviors. To encourage participants to access their intuitive feelings, researchers use the word “feel” in the question stem and/or instructions. For example, one study reported that higher vaccination behavior was associated with higher agreement with questions like, “With no flu shot, I would feel very vulnerable to the flu” (Weinstein et al., 2007). Other studies also reported that higher feelings of risk were associated with higher health behavior intentions and engagement; in some cases feelings of risk were more strongly associated with intentions or behavior than cognitive risk beliefs (Dillard, Ferrer, Ubel, & Fagerlin, 2012; Janssen, van Osch, Lechner, Candel, & de Vries, 2012; Janssen, Waters, van Osch, Lechner, & de Vries, 2014; Weinstein et al., 2007).

Worry and anticipated regret are also affectively-oriented beliefs that shape the way people think about and respond to health risks, but they do so in different ways. Whereas worry is an immediate reaction to the possibility of experiencing harm, anticipated regret is related to the expected consequences of one’s behavior (Loewenstein et al., 2001). Thus, at a given moment an individual may not be worried about developing cancer, but when they imagine developing cancer in the future, they may think that they will feel regretful about not taking preventive action. Worry and anticipated regret are individually associated with higher intentions and engagement in a variety of behaviors including physical activity, influenza vaccination, mammography screening, smoking cessation, and sun protection (Abraham & Sheeran, 2004; Brewer, DeFrank, & Gilkey, 2016; Chapman & Coups, 2006; Dijkstra & Brosschot, 2003; J. Hay, McCaul, & Mangan, 2006; Lechner, de Vries, & Offermans, 1997; Magnan, Köblitz, Zielke, & McCaul, 2009; Sheeran, 2014; Weinstein et al., 2007). Some research suggests that anticipated regret is a stronger predictor of intentions and behavior than worry (Janssen et al., 2012; Janssen et al., 2014; Weinstein et al., 2007).

In sum, feelings of risk, worry, and anticipated regret (hereafter collectively referred to as affective risk beliefs) are often associated with motivation to change behavior and behavior change itself for a variety of disease prevention and detection behaviors. Furthermore, although affective risk beliefs are correlated with cognitive risk beliefs, they load on different factors and make independent contributions to health behavior (Ferrer et al., 2016; Janssen et al., 2012; Moser, McCaul, Peters, Nelson, & Marcus, 2007; Schmiege, Bryan, & Klein, 2009). In this context, it is unsurprising that experts have argued that the potential of disease prevention and control practices can only be realized by accounting for the motivational power of affect (Brewer et al., 2016; Ferrer, McDonald, & Barrett, 2015; Klein & Stefanek, 2007; Lundgren & McMakin, 2013; P. Slovic, Peters, Finucane, & MacGregor, 2005).

The Present Study

This study adds to the literature examining the role of affective risk beliefs in health behavior in four ways. First, it examines whether prior research that demonstrates the importance of affective risk beliefs even after controlling for cognitive risk beliefs (Dillard et al., 2012; Ferrer et al., 2016; Janssen et al., 2012; Janssen et al., 2014; Magnan et al., 2009; Weinstein et al., 2007) generalizes to a new health behavior context: physical activity. Prior research examining the roles of cognitive and affective physical activity beliefs has focused primarily on how participants’ attitudes about possible outcomes of physical activity (e.g., enjoyable feelings, health benefits) are associated with physical activity intentions and/or behavior (Conner, Rhodes, Morris, McEachan, & Lawton, 2011; Lawton, Conner, & McEachan, 2009; Morris, Lawton, McEachan, Hurling, & Conner, 2016); it has not examined cognitive and affective beliefs about health risks resulting from insufficient physical activity. Based on previous research (Dillard et al., 2012; Janssen et al., 2012; Janssen et al., 2014; Magnan et al., 2009; Weinstein et al., 2007), we hypothesized that affective risk beliefs will account for additional variance in intentions and behavior over and above the variance acccounted for by cognitive risk beliefs.

Second, this study includes multiple types of affective risk beliefs in the same model. A recent meta-analysis demonstrates that interventions that increase both cognitive and affective risk beliefs can increase physical activity behavior (Sheeran, 2014), but few studies have examined cognitive risk beliefs and multiple affective risk beliefs simultaneously (but see Janssen et al., 2014 for a non-physical activity example). This distinction is important because identifying which cognitive and/or affective risk beliefs are associated with intentions and behavior even after controlling for the other beliefs will guide the development of future physical activity interventions. Based on prior research (Janssen et al., 2014; Weinstein et al., 2007), we predicted that feelings of risk and anticipated regret – but not worry or cognitive risk beliefs – will predict physical activity intentions and behavior.

Third, we assessed both absolute and comparative versions of cognitive risk beliefs and feelings of risk because understanding how these four concepts collectively relate to physical activity may advance understanding of the nature of perceived risk. Scant research has examined the origins of absolute and comparative feelings of risk, but cognitively-based absolute and comparative perceptions are thought to originate from different sources (i.e. attention to personal risk-related behaviors vs. type of comparison target, respectively) (French & Marteau, 2008; Helweg-Larsen & Shepperd, 2001). In addition, comparative perceptions may have a different relationship with affect than absolute perceptions (French & Hevey, 2008; French & Marteau, 2008; Helweg-Larsen & Shepperd, 2001). It has been suggested that social comparison information might be more affective than objective information and that comparative risk may tap more into feelings of risk (Fagerlin, Ubel, Smith, & Zikmund-Fisher, 2007; Janssen, Verduyn, & Waters, 2018; Klein, 2003; Schwartz, 2009. However, there is also evidence showing absolute risk may be more strongly associated with affect compared to comparative risk (French & Marteau, 2008). Due to conflicting data about whether comparative risk or absolute risk is more strongly associated with affect (Fagerlin, Ubel, Smith, & Zikmund-Fisher, 2007; Janssen, Verduyn, & Waters, 2018; French & Marteau, 2008; Klein, 2003; Schwartz, 2009), we examined absolute and comparative perceived likelihood and feelings of risk in an exploratory way.

Fourth, this study’s sample more accurately reflects the socio-demographic composition of the U.S. than previous work, which often recruited convenience samples of university students, faculty, and staff (Chapman & Coups, 2006; Weinstein et al., 2007) and people who were white and/or had high levels of formal education (Dillard et al., 2012; Ferrer et al., 2016; Janssen et al., 2014). Since people with less formal education and who have a minority racial/ethnic background are less likely to be physically active and to be aware of the link between physical activity and health (Centers for Disease Control & Prevention, 2013; Coups, Hay, & Ford, 2008; Cullen & Buzek, 2009; Oh et al., 2010; Ramirez, Finney Rutten, Vanderpool, Moser, & Hesse, 2013), they are a key target group for physical activity interventions and therefore must be adequately represented in physical activity-related research.

Methods

We conducted secondary analysis of data from a cross-sectional survey that was completed during an experiment that used risk ladders to communicate hypothetical risk estimates of four diseases: heart disease, stroke, diabetes, and colon cancer. Full details about the methodology for the parent experiment are reported in (Janssen, Ruiter, & Waters, 2017) and are described in brief below. All study materials can be obtained from the corresponding author upon request. Ethical approval was obtained from the Human Research Protection Office of the Washington University School of Medicine (IRB approval number 201501028). All participants provided electronic written informed consent.

Participants

As described in (Janssen et al., 2017), data were collected from November 11, 2015 to December 7, 2015 from participants recruited from the GfK KnowledgePanel® Internet survey panel. Individuals were eligible for the parent experiment if they were 30 to 65 years old. Because the parent study sought to intervene on individuals who did not meet U.S. national physical activity guidelines (U.S. Department of Health and Human Services, 2008), eligible participants were required to obtain fewer than 150 minutes per week of moderate intensity aerobic physical activity (U.S. Department of Health and Human Services, 2008). Recruitment was stratified so at least 50% of the sample reported having a minority racial/ethnic background and at least 50% of the sample reported having no more than vocational-technical training.

Procedure and Measures

After consenting, participants indicated how many minutes of physical activity they obtained each day. Then, participants were randomly assigned to view one of twelve experimental conditions that varied the strategies used to communicate hypothetical risk estimates of heart disease, stroke, diabetes, and colon cancer and, for half of the participants, how the risk would be reduced if they engaged in the recommended amount of physical activity (Janssen et al., 2017). Next, participants completed a questionnaire assessing key outcome variables. Where possible, existing items from the literature or national surveys were used. Cognitive testing (Willis, 2004) was performed on the survey with individuals with limited formal education (data not reported). Adaptations to survey wording were made, as needed, to reduce the literacy requirements and to clarify the meaning of response options.

Physical activity behavior was assessed by multiplying two items together (National Cancer Institute, 2009): “In the last 7 days, how many days did you do any physical activity of at least moderate intensity?”; “On the days that you did any physical activity of at least moderate intensity, for how long did you do these activities?” Intentions to obtain physical activity was assessed with three items for which a mean score was calculated and used for analyses (α=0.93) (Conner & Sparks, 1995): “I [intend/want/am likely] to get regular physical activity in the next 3 months.

Risk beliefs were assessed with two sets of four items each: perceived absolute and comparative perceived likelihood (National Cancer Institute, 2009) and absolute and comparative feelings of risk (adapted from Janssen, van Osch, de Vries, & Lechner, 2010; Weinstein et al., 2007). To reduce burden associated with completing the four items for each of the four diseases shown in the parent risk communication experiment, these items were asked only twice: for colon cancer and for “any exercise-related diseases” (including colon cancer). The cognitive beliefs were prefaced with the following sentence: The first two questions are about your THOUGHTS about [colon cancer/diseases shown in the picture] (emphasis in original). Perceived absolute likelihood was assessed with the item, “How likely do you think it is that you will get [colon cancer/sick from any of the diseases shown in the picture] in the next 10 years, if you do not get regular physical activity? [1=not likely, 4=very likely].” Perceived comparative likelihood was assessed with the item, “Compared to other people your age and sex, how likely do you think it is that you will get [colon cancer/sick from any of the diseases shown in the picture] in the next 10 years, if you do not get regular physical activity? [1=much less likely, 5=much more likely].” The feelings questions were prefaced with the following sentence: The next questions are about your FEELINGS about colon cancer. Absolute feelings of risk: “How much do you agree with the following statement: “I feel like I could easily get [colon cancer/sick from any of the diseases shown in the picture] in the next 10 years if I do not get regular physical activity [1=do not agree, 4=strongly agree].” Comparative feelings of risk: “Compared to other people your age and sex, how easily do you feel you could get [colon cancer/sick from any of the diseases shown in the picture] in the next 10 years if you do not get regular physical activity? [1=much less easily, 5=much more easily].” Each item included an “I don’t know” response option placed at the end of the scale.

Worry (National Cancer Institute, 2009) had two disease-specific versions: “How worried are you about getting [colon cancer/sick from any of the diseases shown in the picture]? [1=not worried; 4=very worried].” Anticipated regret (adapted from Weinstein et al., 2007) also had two disease-specific versions, each created from two items for which a mean score was calculated and used for analyses (α=0.93 for colon cancer and α=0.95 for other diseases): “I would be mad at myself if I got [colon cancer/sick from the diseases shown in the picture] because I did not get regular physical activity” and “I would regret it if I got sick from the diseases shown in the picture because I did not get regular physical activity.” Response scales for both items were [1=do not agree; 4=strongly agree].

Participants then indicated their sex, age, education, and race/ethnicity (National Cancer Institute, 2009), and completed two numeracy items (Lipkus, Samsa, & Rimer, 2001) and four graph literacy items (Galesic & Okan, 2013). Exact item wording for these and other variables that were assessed but are outside the scope of this paper can be obtained from the corresponding author.

Statistical Analyses

The analytic dataset included people who had no missing data on any items needed for the main analyses and whose questionnaire completion time was between the 3rd and 97th percentiles (MedianCompletionTime=21 minutes). Thus, 835 of the original 1161 respondents were included in the analyses.

Analyses were performed separately for colon cancer and “any diseases.” To examine the additional contribution of all the affective risk beliefs over and above all the cognitive risk beliefs, hierarchical linear regressions were performed. For physical activity intentions and behavior separately, the socio-demographics were entered in the first block, the cognitive risk beliefs in the second block, and the affective risk beliefs in the third block. Intentions were included in the fourth block for the analyses predicting physical activity behavior.

The parent study provided people with hypothetical risk information, but it had overwhelmingly null effects on health cognitions and outcomes (Janssen et al., 2017). In the few situations that yielded significant results, the effect sizes were so small as to be negligible. Nevertheless, we repeated the hierarchical regressions described above with the three experimental variables from the parent study included in the first block. None of the results changed in terms of direction of the point estimate or its significance. Therefore, in the interest of model parsimony, the analyses reported here do not control for the three experimental variables reported in (Janssen et al., 2017).

Results

As reported in (Janssen et al., 2017), the average age in the analytic sample was 48 years old (SD=10.22), most (57%) participants were female, and almost half of participants had minority racial/ethnic ancestry (47%) or had no college training (46%). The average numeracy score was 1.3 (SD=0.76; range=0–2) and the average graph literacy score was 2.2 (SD=1.10; range=0–4). Participants reported moderately high physical activity intentions (M=2.83, SD=0.77, Range=1–4) and engaged in an average of 52.5 (SD=44.98) minutes of moderate intensity physical activity per week. Intercorrelations among the risk belief items related to colon cancer ranged from r=.18 (for perceived comparative likelihood and anticipated regret) to r=.74 (for perceived comparative likelihood and comparative feelings of risk). Intercorrelations among the risk belief items related to “any of the diseases” were similar in direction and magnitude. Means, standard deviations, and intercorrelations among all the cognitive and affective risk belief items can be found in Table 1.

Table 1.

Means, SDs, and intercorrelations among cognitive and affective risk beliefs

Colon cancer Any of the diseases
1 2 3 4 5 6 7 8 9 10 11 12
Colon cancer 1. Perceived absolute likelihooda 1.00
2. Perceived comparative likelihoodb 0.52 1.00
3. Absolute feelings of riska 0.65 0.53 1.00
4. Comparative feelings of riskb 0.45 0.74 0.61 1.00
5. Worrya 0.49 0.40 0.57 0.47 1.00
6. Anticipated regreta 0.16 0.15 0.29 0.18 0.20 0.20
Any of the diseases 7. Perceived absolute likelihooda 0.51 0.42 0.49 0.43 0.40 0.21 1.00
8. Perceived comparative likelihoodb 0.37 0.58 0.42 0.65 0.34 0.17 0.63 1.00
9. Absolute feelings of riska 0.47 0.45 0.58 0.52 0.41 0.32 0.72 0.63 1.00
10. Comparative feelings of riskb 0.35 0.57 0.46 0.65 0.32 0.22 0.58 0.82 0.68 1.00
11. Worrya 0.38 0.33 0.44 0.41 0.63 0.28 0.54 0.48 0.57 0.48 1.00
12. Anticipated regreta 0.22 0.16 0.31 0.18 0.21 0.83 0.28 0.20 0.36 0.25 0.33 1.00
Mean
(Standard deviation)		 2.07 (0.94) 3.15 (1.01) 2.10 (0.91) 3.12 (0.96) 1.78 (0.87) 2.82 (0.95) 2.54 (0.93 3.28 (0.96) 2.49 (0.89) 3.26 (0.94) 2.15 (0.90) 2.86 (0.92)
Open in a new tab

Note: All correlations significant at p<01.

a

Response scale: 1–4; 4 represents the highest levels of the construct.

b

Response scale: 1–5; 5 represents the highest level of the construct.

Physical Activity Intentions

Full details of the hierarchical regression analyses are in Tables 2 and 3. As demonstrated by the F-change statistics in the tables, each step of the analysis accounted for a significant proportion of variance in intentions to engage in physical activity. For Step 1, physical activity intentions were higher among people who reported having minority (vs. non-minority) racial ancestry (βColonCancer=0.11, p<.01 and βAnyDiseases=0.14, p<.01) and people who had more than (vs. less than) vocational/technical training (βColonCancer=0.15, p<.01 and βAnyDiseases=0.14, p<.01). Step 2 revealed that people who reported higher perceived absolute likelihood of developing colon cancer or “any of the diseases” shown in the risk assessment tool had higher intentions (βColonCancer=0.11, p<.05 and βAnyDiseases=0.19, p<.001), but there was no effect for perceived comparative likelihood (βColonCancer=0.07, p>.05 and βAnyDiseases<0.00, p>.05). Adding all four of the affective risk belief items simultaneously to Step 3 reduced the effect of perceived absolute likelihood to non-significance (βColonCancer=0.02, p>.05 and βAnyDiseases=0.02, p>.05). However, the only individual affective risk belief items that showed significant relationships with intentions were absolute feelings of risk of colon cancer or “any of the diseases” (βColonCancer=0.14, p<.05 and βAnyDiseases=0.16, p<.01) and anticipated regret about getting colon cancer or “any of the diseases” (βColonCancer=0.31, p<.001 and βAnyDiseases=0.29, p<.001). There were no significant associations between intentions and either comparative feelings of risk (βColonCancer=0.04, p>.05 and βAnyDiseases=0.06, p>.05) or worry (βColonCancer=0.01, p>.05 and βAnyDiseases=0.04, p>.05).

Table 2.

Hierarchical regression predicting physical activity intentions and behavior from cognitive and affective colon cancer risk beliefs

Intentions (N=549a) Behavior (N=549a)
Predictorb β R2 (Fchange)c β R2 (Fchange)c
Step 1 0.04 (F=4.78***) 0.02 (F=2.84*)
  Sex 0.08 −0.02
  Age 0.01 0.00
  Education 0.15** 0.18***
  Race 0.11** 0.01
  Graph literacy 0.04 −0.00
  Numeracy 0.04 −0.05
Step 2 0.06 (F=6.77**) 0.03 (F=2.62)
  Sex 0.09 −0.02
  Age 0.00 −0.01
  Education 0.16*** 0.18***
  Race 0.10* −0.00
  Graph literacy 0.06 0.01
  Numeracy 0.05 −0.05
  Perceived absolute likelihood 0.11* 0.12*
  Perceived comparative likelihood 0.07 −0.06
Step 3 0.17 (F=19.54***) 0.03 (F=1.19)
  Sex 0.08 −0.01
  Age 0.02 −0.00
  Education 0.15*** 0.19***
  Race 0.08 −0.01
  Graph literacy 0.05 0.02
  Numeracy 0.06 −0.04
  Perceived absolute likelihood 0.02 0.06
  Perceived comparative likelihood −0.04 −0.07
  Absolute feelings of risk 0.14* 0.09
  Comparative feelings of risk 0.04 −0.06
  Worry 0.01 0.05
  Anticipated regret 0.31*** 0.03
Step 4 0.07 (F=25.54***)
  Sex −0.03
  Age −0.01
  Education 0.15**
  Race −0.03
  Graph literacy 0.01
  Numeracy −0.05
  Perceived absolute likelihood 0.06
  Perceived comparative likelihood −0.06
  Absolute feelings of risk 0.06
  Comparative feelings of risk −0.07
  Worry 0.04
  Anticipated regret −0.04
  Intention 0.23***
Open in a new tab
a

Of the 835 participants in Study 1, 549 provided a valid response to all four of the colon cancer perceived risk questions; 286 reported “don’t know” to at least one of the questions.

b

For sex, 1=male, 2=female. For education, 1=vocational-technical training or less, 2=more than vocational-technical training. For race, 1=white and 2=non-white. Graph literacy range: 0–4. Numeracy range: 0–2. Absolute likelihood judgments, worry, regret, and intentions ranged from 1 (lowest) to 4 (highest); Scores for comparative likelihood judgments ranged from 1–5.

c

Fchange: The significance of adding additional variables in each step.

*

p < 0.05

**

p < 0.01

***

p < 0.001.

Table 3.

Hierarchical regression predicting physical activity intentions and behavior from cognitive and affective risk beliefs about “any of the diseases” shown in the risk assessment tool

Intentions (N=670a ) Behavior (N=670a)
Predictorb β R2 (Fchange)c β R2 (Fchange)c
Step 1 0.04 (F=5.89***) 0.02 (F=3.10**)
  Sex 0.08* −0.03
  Age 0.01 0.01
  Education 0.14** 0.17***
  Race 0.14** 0.01
  Graph literacy 0.04 0.01
  Numeracy 0.03 −0.04
Step 2 0.07 (F=12.29***) 0.02 (F=1.79)
  Sex 0.08 −0.04
  Age −0.01 0.00
  Education 0.14*** 0.17***
  Race 0.12** 0.01
  Graph literacy 0.05 0.01
  Numeracy 0.03 −0.03
  Perceived absolute likelihood 0.19*** 0.09
  Perceived comparative likelihood −0.00 −0.08
Step 3 0.18 (F=22.63***) 0.02 (F=1.55)
  Sex 0.04 −0.04
  Age 0.01 0.01
  Education 0.13** 0.16***
  Race 0.09* −0.00
  Graph literacy 0.04 0.01
  Numeracy 0.04 −0.03
  Perceived absolute likelihood 0.02 0.02
  Perceived comparative likelihood −0.12 −0.10
  Absolute feelings of risk 0.16** 0.10
  Comparative feelings of risk 0.06 −0.02
  Worry 0.04 0.03
  Anticipated regret 0.29*** 0.05
Step 4 0.07 (F=33.91***)
  Sex −0.05
  Age 0.00
  Education 0.13**
  Race −0.02
  Graph literacy 0.00
  Numeracy −0.04
  Perceived absolute likelihood 0.01
  Perceived comparative likelihood −0.08
  Absolute feelings of risk 0.06
  Comparative feelings of risk −0.04
  Worry 0.02
  Anticipated regret −0.02
  Intention 0.24***
Open in a new tab
a

Of the 835 participants in Study 1, 670 provided a valid response to the cognitive and affective absolute and comparative “any disease” risk questions; 165 reported “don’t know” to at least one item.

b

For sex, 1=male, 2=female. For education 1=vocational-technical training or less, 2=more than vocational-technical training. For race, 1=white and 2=non-white. Graph literacy range: 0–4. Numeracy range: 0–2. Absolute likelihood judgments, worry, regret, and intentions ranged from 1 (lowest) to 4 (highest); Scores for comparative likelihood judgments ranged from 1–5.

c

Fchange: The significance of adding additional variables in each step.

*

p < 0.05

**

p < 0.01

***

p < 0.001.

Physical Activity Behavior

As shown in Step 1 of Tables 2 and 3, people with higher (vs. lower) education reported higher engagement in physical activity behavior (βColonCancer=0.18, p<.001 and βAnyDiseases=0.17, p<.001). Because the F-change statistics for Steps 2 and 3 were not significant, we do not discuss the association between the individual cognitive or affective risk belief items and behavior for steps 2 and 3. Adding physical activity intentions to Step 4 accounted for a significant proportion of variance in behavior over and above the variance accounted for by Steps 1–3. People with higher intentions reported engaging in more activity (βColonCancer=0.23, p<.001 and βAnyDiseases=0.24, p<.001). In other words, for the colon cancer Step 4 analyses, people with more education engaged in an average of 13.8 (SE=4.1) more minutes of physical activity than people with less education, and each 1-unit increase in intentions amounted to an average of 13.8 (SE=2.7) more minutes of activity. Results were similar for the “any of the diseases” analyses; more education was associated with 12.0 (SE=3.7) more minutes than less education, and each 1-unit increase in intentions equated to 14.7 (SE=2.5) more minutes of physical activity

Exploratory Indirect Effect Analyses

For both disease targets (colon cancer and “any of the diseases”), a significant Step 2 relationship between perceived absolute likelihood and intentions became nonsignificant in Step 3 after adding the block of four affective risk belief variables. This may indicate an indirect effect of perceived absolute likelihood on intentions through one or more of the affective risk belief variables. We investigated this possibility using the %INDIRECT macro (Hayes, 2018). Specifically, we estimated two multiple mediator models (one for each disease target separately) in which the indirect effect of perceived absolute likelihood on intentions occurred through all four of the affective variables at once while controlling for the covariates. Then, we used the macro to calculate the bias-corrected bootstrapped 95% confidence intervals (Hayes & Scharkow, 2013) of the indirect effect to evaluate the statistical significance of the effect.

Both multiple mediation models indicated that, even controlling for the covariates, there was a significant indirect effect of perceived absolute likelihood on intentions through absolute feelings of risk (βColonCancer = 0.06, 95% CIColonCancer 0.01 – 0.13; βAnyDiseases = 0.09, 95% CIAnyDiseases 0.02 – 0.16) and anticipated regret (βColonCancer = 0.04, 95% CIColonCancer 0.03 – 0.070; βAnyDiseases = 0.06, 95% CIAnyDiseases 0.04 – 0.10). However, there were no indirect effects through comparative feelings of risk (βColonCancer = 0.01, 95% CIColonCancer: −0.03 – 0.04; βAnyDiseases = 0.01, 95% CIAnyDiseases −0.06 – 0.04) or worry (βColonCancer < 0.00; 95% CIColonCancer: −0.03 – 0.04; βAnyDiseases = 0.01, 95% CIAnyDiseases −0.03 – 0.05).

Supplemental Exploratory Analyses

A growing literature demonstrates that “do not know” (DK) responses to risk perception questions is a meaningful indicator of a unique population of people who may need additional health education and behavior change interventions (J. L. Hay, Orom, Kiviniemi, & Waters, 2015; Kiviniemi, Orom, Waters, McKillip, & Hay, 2018; Erika A. Waters, Hay, Orom, Kiviniemi, & Drake, 2013; E. A. Waters, Kiviniemi, Orom, & Hay, 2016). DK responding also occurs more frequently for cognitive than affective risk perception items (Janssen, Verduyn, & Waters, 2018). For these reasons, we repeated the hierarchical analyses described above after replacing the absolute and comparative perceived likelihood predictors and the absolute and comparative feelings of risk predictors with dichotomous predictors indicating whether people did or did not mark the DK option for that item.

The full results of these analyses are shown in Tables 4 and 5. For intentions, the F-change statistics for Steps 1, 2, and 3 were all statistically significant for both of the disease targets. Furthermore, the specific relationships between the individual items assessing DK responding and intentions were similar for both disease targets. In Step 2, DK responses for perceived comparative likelihood items were statistically significantly associated with lower intentions to engage in physical activity (β=−0.15, p<.001 for colon cancer and any diseases, see Tables 4 and 5 respectively). Adding all four of the affective risk beliefs items in Step 3 reduced those effects to non-significance (βColonCancer=−0.01, p>.05 and βAnyDiseases=−0.07, p>.05). Intentions were significantly higher among participants with higher worry (βColonCancer=0.09, p<.01 and βAnyDiseases=0.07, p<.05) and anticipated regret (β=0.33, p<.001 for colon cancer and any diseases). Intentions were not related to DK responding to the absolute feelings of risk item (βColonCancer=0.02, p>.05 and βAnyDiseases=−0.03, p>.05) or DK responding to the comparative feelings of risk item (βColonCancer=−0.07, p>.05 and βAnyDiseases=0.00, p>.05).

Table 4.

Hierarchical regression exploring the effect “don’t know” responses to items assessing perceived risk of colon cancer

Intentions (N=835) Behavior (N=835)
Predictora β R2 (Fchange)b β R2 (Fchange)b
Step 1 0.06 (F=9.57***) 0.02 (F=4.08***)
  Sex 0.10** −0.04
  Age 0.01 −0.00
  Education 0.16*** 0.17***
  Race 0.13*** 0.02
  Graph literacy 0.08 0.01
  Numeracy 0.04 −0.02
Step 2 0.08 (F=9.42***) 0.03 (F=3.63*)
  Sex 0.09** −0.04
  Age 0.01 −0.00
  Education 0.15*** 0.16***
  Race 0.14*** 0.02
  Graph literacy 0.07 0.00
  Numeracy 0.02 −0.03
  DKR perceived absolute likelihood 0.01 0.03
  DKR perceived comparative likelihood −0.15*** −0.11*
Step 3 0.20 (F=32.85***) 0.03 (F=2.09)
  Sex 0.08* −0.04
  Age 0.02 −0.00
  Education 0.15*** 0.17***
  Race 0.11** 0.01
  Graph literacy 0.06 0.00
  Numeracy 0.04 −0.02
  DKR perceived absolute likelihood −0.01 0.03
  DKR perceived comparative likelihood −0.04 −0.04
  DKR absolute feelings of risk 0.02 0.09
  DKR comparative feelings of risk −0.07 −0.09
  Worry 0.09** 0.06
  Anticipated regret 0.33*** 0.02
Step 4 0.09 (F=55.27***)
  Sex −0.06
  Age −0.01
  Education 0.13**
  Race −0.02
  Graph literacy −0.01
  Numeracy −0.03
  DKR perceived absolute likelihood 0.03
  DKR perceived comparative likelihood −0.03
  DKR absolute feelings of risk −0.00
  DKR comparative feelings of risk −0.07
  Worry 0.04
  Anticipated regret −0.08*
  Intention 0.28***
Open in a new tab
a

For sex, 1=male, 2=female. For education 1=vocational-technical training or less, 2=more than vocational-technical training. For race, 1=white and 2=non-white. Graph literacy range: 0–4. Numeracy range: 0–2. For DKR, 1=answer other than “don’t know” and 2=answered “don’t know”. Absolute likelihood judgments, worry, regret, and intentions ranged from 1 (lowest) to 4 (highest); Scores for comparative likelihood judgments ranged from 1–5.

b

Fchange: The significance of adding additional variables in each step.

*

p < 0.05

**

p < 0.01

***

p < 0.001.

Table 5.

Hierarchical regression exploring the effect “don’t know” responses to items assessing perceived risk of “any of the diseases” shown in the risk assessment tool

Intentions (N=835) Behavior (N=835)
Predictora β R2 (Fchange)b β R2 (Fchange)b
Step 1
  Sex 0.10** 0.06 (F=9.57***) −0.04 0.02 (F=4.08***)
  Age 0.01 −0.00
  Education 0.16*** 0.17***
  Race 0.13*** 0.02
  Graph literacy 0.08 0.01
  Numeracy 0.04 −0.02
Step 2
  Sex 0.09** 0.09 (F=17.32***) −0.04 0.03 (F=4.62*)
  Age 0.01 −0.01
  Education 0.15*** 0.16***
  Race 0.13*** 0.02
  Graph literacy 0.06 −0.00
  Numeracy 0.02 −0.03
  DKR perceived absolute likelihood −0.07 0.02
  DKR perceived comparative likelihood −0.15** −0.12**
Step 3
  Sex 0.06 0.21 (F=31.94***) −0.05 0.03 (F=1.75)
  Age 0.02 −0.00
  Education 0.14*** 0.16***
  Race 0.10** 0.01
  Graph literacy 0.05 −0.00
  Numeracy 0.02 −0.02
  DKR perceived absolute likelihood −0.02 0.03
  DKR perceived comparative likelihood −0.07 −0.10
  DKR absolute feelings of risk −0.03 0.05
  DKR comparative feelings of risk −0.00 −0.05
  Worry 0.07* 0.06
  Anticipated regret 0.33*** 0.05
Step 4
  Sex −0.06 0.09 (F=50.30***)
  Age −0.01
  Education 0.12**
  Race −0.02
  Graph literacy −0.02
  Numeracy −0.03
  DKR perceived absolute likelihood 0.03
  DKR perceived comparative likelihood −0.08
  DKR absolute feelings of risk 0.06
  DKR comparative feelings of risk −0.05
  Worry 0.04
  Anticipated regret −0.04
  Intention 0.27***
Open in a new tab
a

For sex, 1=male, 2=female. For education 1=vocational-technical training or less, 2=more than vocational-technical training. For race, 1=white and 2=non-white. Graph literacy range: 0–4. Numeracy range: 0–2. For DKR, 1=answer other than “don’t know” and 2=answered “don’t know”. Absolute likelihood judgments, worry, regret, and intentions ranged from 1 (lowest) to 4 (highest); Scores for comparative likelihood judgments ranged from 1–5.

b

Fchange: The significance of adding additional variables in each step.

*

p < 0.05

**

p < 0.01

***

p < 0.001.

For behavior, the F-change statistics for Steps 1, 2, and 4 were statistically significant for both disease targets (Tables 4 and 5). In Step 2, DK responding for comparative likelihood items was associated with lower physical activity behavior (βColonCancer=−0.11, p<.05, 12.1 [SE=4.7] fewer minutes and βAnyDiseases=−0.12, p<.05, 16.1 [SE=6.1] fewer minutes). Adding the four affective risk beliefs items in Step 3 did not improve prediction and so the Step 3 findings will not be discussed. Adding intentions in Step 4 accounted for a significant amount of variance. Each unit increase in intentions was associated with 16.0 (SE=2.6] and 15.5 [SE=2.2] more minutes of physical activity behavior for the colon cancer and “any disease” models, respectively. For colon cancer only, Step 4 also yielded a significant negative association between anticipated regret and physical activity behavior (βColonCancer=−0.08, p<.05, 3.5 [SE=1.8] fewer minutes of activity).

Discussion

Adding a block of affective risk beliefs to a model comprised of socio-demographic and cognitive risk belief variables accounted for a significant amount of additional variance in physical activity intentions regardless of whether participants were asked about one specific disease or for a “composite” evaluation of four diseases. This finding is similar to research in other health domains, including smoking cessation intentions, fruit intake, and sun protection (Janssen et al., 2012; Janssen et al., 2014).

Examination of the effects of individual affective beliefs variables indicated that only anticipated regret was associated with higher intentions to engage in physical activity in both the main analytic models (Tables 2–3) and in the exploratory models that included people who indicated they did not know their risk (Tables 4–5). Other research has also found that higher anticipated regret can be associated with higher intentions and actual engagement in a variety of health behaviors (Brewer et al., 2016; Chapman & Coups, 2006; Janssen et al., 2014; Sheeran, 2014; Weinstein et al., 2007).

The relationships between intentions and the other three affective risk belief variables were mixed. As in prior research, comparative feelings of risk were not associated with intentions in any of the models (Janssen et al., 2010). In the models that excluded “don’t know” (DK) responders, absolute feelings of risk were associated with intentions but worry was not. This is consistent with Study 2 of (Janssen et al., 2014) and (Weinstein et al., 2007). Future research is needed to determine the mechanisms driving the effect.

In contrast, the models that included the dichotomous variables indicating DK responses to absolute and comparative perceived likelihood and feelings of risk indicated that worry – but not DK responding to absolute or comparative feelings of risk items – was associated with intentions. With a few exceptions (Dillard et al., 2012; Ferrer et al., 2016), much prior research showing that higher worry is associated with higher intentions and behavior has not included feelings of risk (Dijkstra & Brosschot, 2003; J. Hay et al., 2006; Schmiege et al., 2009), and none compared individuals who did and did not respond DK to risk perception items. Therefore, it is not possible to directly compare the results. Nevertheless, it could be that worry is only influential for some diseases, for some behaviors, or when the sample is comprised of both individuals who are and who are not able and willing to indicate their perceived risk.

The data also showed that, whereas the sample that excludes don’t know responders indicates that perceived absolute likelihood and absolute feelings of risk are significantly related to intentions, the sample that includes DK responders yields significant effects for DK responses to perceived comparative likelihood and worry. Previous research that systematically examined differences in DK responding by item format found that DK responding was higher for cognitively-oriented perceived risk items than feelings of risk items, but there were no differences according to whether the items related to absolute or comparative risk (Janssen et al., 2018). However, that research did not examine these issues in the context of behavior or intentions, and research reporting that DK responding may be associated with lower physical activity behavior only examined comparative likelihood perceptions (Waters et al., 2016). More work is needed to determine if the differences reported here are the result of fundamental differences in the way people conceptualize absolute and comparative risk (e.g., French & Marteau, 2008), the result of differential affective intensity of absolute and comparative risk (French & Hevey, 2008), or some other mechanism.

The indirect effects analyses indicated that anticipated regret and absolute feelings of risk may mediate the link between perceived absolute likelihood and intentions. This finding, in addition to the accumulated evidence from this study and prior research, suggests that future interventions may be more effective if they focus on attempting to increase anticipated regret and absolute feelings of risk than worry and comparative feelings of risk. In addition, anticipated regret was significantly related to intentions in samples that included and excluded DK responders. Thus, health behavior interventions that target anticipated regret might more easily reach those participants who are likely to respond DK and therefore may be in most need of intervention (i.e., those who are the most socio-demographically disadvantaged and have the least health knowledge J. L. Hay et al., 2015; Kiviniemi et al., 2018; Erika A. Waters et al., 2013).

The general lack of associations between the affective risk beliefs variables and behavior (except for the exploratory DK analyses related to colon cancer) contrasts with several studies that found that higher feelings of risk, worry, and anticipated regret were associated with more engagement in healthy behaviors (Brewer et al., 2016; Chapman & Coups, 2006; Dijkstra & Brosschot, 2003; J. Hay et al., 2006; Janssen et al., 2012; Janssen et al., 2014; Lechner et al., 1997; Sheeran, 2014; Weinstein et al., 2007). This difference could be because our studies included a different combination of affective risk beliefs than most of the other studies. Our results could also be due to the nature of physical activity, which is a repeated and difficult behavior. Support for this is seen in Study 1 of (Janssen et al., 2014), which found null effects of affective risk belief items for smoking cessation, another difficult behavior to change. Researchers (Sheeran, 2014) have noted that people engage in physical activity for a variety of reasons unrelated to disease risk reduction, and that focusing on factors such as emotional, social, and cosmetic benefits will likely be useful in encouraging change. Detailed information about behavior-specific differences in the magnitude of the perceived risk-behavior relatonship can be found in (Sheeran, 2014). Nevertheless, future research should examine the conditions under which anticipated regret and other affective risk beliefs are and are not related to behavior.

Strengths, Limitations, and Future Directions

This study was the first to examine the added contribution of affective risk beliefs over and above cognitive risk beliefs in predicting physical activity intentions and behavior. It was also one of the few studies that recruited a large, highly socio-demographically diverse sample of U.S. residents. In addition, this sample included participants with somewhat lower graph literacy and numeracy compared to the general U.S. population (i.e., 55% of graph literacy items correct compared to 67% correct in (Galesic & Okan, 2013) and 65% of numeracy items correct compared to 77% in (Nelson, Moser, & Han, 2013)).

Nevertheless, these results should be viewed in light of the following limitations. First, we used a cross-sectional observational design. This precludes drawing conclusions about causality or directionality, including whether or not the indirect effects we report can be considered causal mediational pathways. Future research should use longitudinal designs and include designs in which affective and cognitive beliefs are experimentally induced (see examples in Portnoy, Ferrer, Bergman, & Klein, 2014). Second, we primarily used single-item measures. Using multi-item measures may improve reliability. Third, by examining only one behavioral context it is not possible to understand how affective risk beliefs might interact with different characteristics of different behaviors. Research that examines a wider variety of behavioral situations is needed. Future studies could also intervene on anticipated regret and absolute feelings of risk to increase intentions, and then add other intervention components (e.g., improve self-regulatory skills) to bridge the intention-behavior gap. Fourth, the parent experiment targeted only individuals who were not meeting U.S. national physical activity guidelines; it is possible that the resulting restricted range precluded finding a significant association between risk beliefs and actual physical activity behavior. Fifth, it is challenging to assess an intuitive construct with explicit survey measures. Future research should explore the utility of using more implicit measures instead.

Conclusions

Using a large, socio-demographically diverse sample, we discovered that a group of four affective risk beliefs account for unique variance in physical activity intentions above and beyond that accounted for by traditional socio-demographic variables and cognitive risk beliefs. Anticipated regret was the only affective risk belief that was associated with intentions in all the models examined, making it a likely candidate for intervention. Although neither cognitive nor affective risk beliefs were associated with behavior, this study provides insight into potential intervention targets for future research and practice.

Acknowledgments

Funding

This research was supported by the U.S. National Cancer Institute [R01CA190391] and the National Center for Advancing Translational Sciences [UL1TR000448].

Footnotes

Declaration of Conflicting Interests

The authors declare that there is no conflict of interest.

Data Availability Statement

All study materials, including the dataset and statistical code, can be obtained from the corresponding author.

References

  1. Abraham C, & Sheeran P (2004). Deciding to exercise: The role of anticipated regret. British journal of health psychology, 9(2), 269–278. [DOI] [PubMed] [Google Scholar]
  2. Ajzen I (1985). From intentions to action: A theory of planned behavior. In Kuhl J & Beckman J (Eds.), Action Control: From Cognitions to Behaviors (pp. 11–39). New York: Springer. [Google Scholar]
  3. Bechara A, Damasio H, Tranel D, & Damasio AR (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275(1293), 1293–1295. [DOI] [PubMed] [Google Scholar]
  4. Becker MH (1974). The health belief model and personal health behavior. Health Education Monographs, 2, 324–508. [Google Scholar]
  5. Brewer NT, DeFrank JT, & Gilkey MB (2016). Anticipated regret and health behavior: A meta-analysis. Health Psychol, 35(11), 1264–1275. 10.1037/hea0000294 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cameron LD (2003). Anxiety, cognition, and response to health threats. In Cameron LD & Leventhal H (Eds.), The self regulation of health and illness behavior (pp. 157–183). London: Routledge. [Google Scholar]
  7. Centers for Disease Control & Prevention. (2013). CDC Health Disparities & Inequalities Report – United States, 2013. Morbidity and Mortality Weekly Report, 62(Supplement 3), 1–187.23302815 [Google Scholar]
  8. Chapman GB, & Coups EJ (2006). Emotions and preventive health behavior: Worry, regret, and influenza vaccination. Health Psychology, 25(1), 82–90. [DOI] [PubMed] [Google Scholar]
  9. Conner M, Rhodes RE, Morris B, McEachan R, & Lawton R (2011). Changing exercise through targeting affective or cognitive attitudes. Psychology & Health, 26(2), 133–149. [DOI] [PubMed] [Google Scholar]
  10. Conner M, & Sparks P (1995). The theory of planned behaviour and health behaviours. In Conner M & Norman P (Eds.), Predicting Health Behaviour (pp. 121–162). Philadelphia, PA: Open University Press. [Google Scholar]
  11. Coups EJ, Hay J, & Ford JS (2008). Awareness of the role of physical activity in colon cancer prevention. Patient Educ Couns, 72(2), 246–251. 10.1016/j.pec.2008.03.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cullen KW, & Buzek BB (2009). Knowledge About Type 2 Diabetes Risk and Prevention of African-American and Hispanic Adults and Adolescents With Family History of Type 2 Diabetes. The Diabetes Educator, 35(5), 836–842. 10.1177/0145721709341851 [DOI] [PubMed] [Google Scholar]
  13. Damasio A (1994). Descartes’ error: Emotion, reason, and the human brain New York, NY: Quill. [Google Scholar]
  14. Dijkstra A, & Brosschot J (2003). Worry about health in smoking behaviour change. Behaviour research and therapy, 41(9), 1081–1092. 10.1016/s0005-7967(02)00244-9 [DOI] [PubMed] [Google Scholar]
  15. Dillard AJ, Ferrer RA, Ubel PA, & Fagerlin A (2012). Risk perception measures’ associations with behavior intentions, affect, and cognition following colon cancer screening messages. Health Psychology, 31(1), 106–113. 10.1037/a0024787 [DOI] [PubMed] [Google Scholar]
  16. Dillard AJ, Ubel PA, Smith DM, Zikmund-Fisher BJ, Nair V, Derry HA, … Fagerlin A (2011). The Distinct Role of Comparative Risk Perceptions in a Breast Cancer Prevention Program. Ann Behav Med, 42(2), 262–268. 10.1007/s12160-011-9287-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Edwards W (1954). The theory of decision making. Psychological bulletin, 51(4), 380. [DOI] [PubMed] [Google Scholar]
  18. Fagerlin A, Ubel PA, Smith DM, & Zikmund-Fisher BJ (2007). Making numbers matter: Present and future research in risk communication. American Journal of Health Behavior, 31(Suppl 1), S47–S56. [DOI] [PubMed] [Google Scholar]
  19. Ferrer RA, Klein WMP, Persoskie A, Avishai-Yitshak A, & Sheeran P (2016). The Tripartite Model of Risk Perception (TRIRISK): Distinguishing Deliberative, Affective, and Experiential Components of Perceived Risk. Annals of Behavioral Medicine, 50(5), 653–663. 10.1007/s12160-016-9790-z [DOI] [PubMed] [Google Scholar]
  20. Ferrer RA, McDonald PG, & Barrett LF (2015). Affective science perspectives on cancer control: Strategically crafting a mutually beneficial research agenda. Perspectives on psychological science : a journal of the Association for Psychological Science, 10(3), 328–345. 10.1177/1745691615576755 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Finucane ML, Alhakami A, Slovic P, & Johnson SM (2000). The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making, 13(1), 1–17. [Google Scholar]
  22. Fishbein M, & Ajzen I (1975). Belief, attitude, intention, and behavior: An introduction to theory and research Reading, MA: Addison-Wesley. [Google Scholar]
  23. French DP, & Hevey D (2008). What do people think about when answering questionnaires to assess unrealistic optimism about skin cancer? A think aloud study. Psychology, Health and Medicine, 13(1), 63–74. [DOI] [PubMed] [Google Scholar]
  24. French DP, & Marteau TM (2008). Different Measures of Risk Perceptions and Distress Yield Different Patterns of Results: The Importance of Specificity in Measurement and Theory. Psychology, Health and Medicine, 13(5), 591–596. [DOI] [PubMed] [Google Scholar]
  25. Galesic M, & Okan Y (2013). Short graph literacy scale. Paper presented at the 35th Annual Meeting of the Society for Medical Decision Making, Baltimore, MD, United States. [Google Scholar]
  26. Hay J, McCaul KD, & Mangan RE (2006). Does worry about breast cancer predict screening behaviors? A meta-analysis of the prospective evidence. Preventive Medicine: An International Journal Devoted to Practice and Theory, 42(6), 401–408. [DOI] [PubMed] [Google Scholar]
  27. Hay JL, Orom H, Kiviniemi MT, & Waters EA (2015). “I don’t know” my cancer risk: exploring deficits in cancer knowledge and information-seeking skills to explain an often-overlooked participant response. Medical Decision Making, 35(4), 436–445. 10.1177/0272989X15572827 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hayes AF (2018). SPSS, SAS, and Mplus macros and code Retrieved from http://www.afhayes.com/spss-sas-and-mplus-macros-and-code.html
  29. Hayes AF, & Scharkow M (2013). The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: does method really matter? Psychol Sci, 24(10), 1918–1927. 10.1177/0956797613480187 [DOI] [PubMed] [Google Scholar]
  30. Helweg-Larsen M, & Shepperd JA (2001). Do moderators of the optimistic bias affect personal or target risk estimates? A review of the literature. Personality and Social Psychology Review, 5(1), 74–95. [Google Scholar]
  31. Janssen E, Ruiter RA, & Waters EA (2017). Combining risk communication strategies to simultaneously convey the risks of four diseases associated with physical inactivity. J Behav Med 10.1007/s10865-017-9894-3. [DOI] [PMC free article] [PubMed]
  32. Janssen E, van Osch L, de Vries H, & Lechner L (2010). Measuring risk perceptions of skin cancer: Reliability and validity of different operationalizations. British journal of health psychology, 16, 92–112. 10.1348/135910710X514120 [DOI] [PubMed] [Google Scholar]
  33. Janssen E, van Osch L, Lechner L, Candel M, & de Vries H (2012). Thinking versus feeling: Differentiating between cognitive and affective components of perceived cancer risk. Psychology & Health, iFirst, 1–17. [DOI] [PubMed]
  34. Janssen E, Verduyn P, & Waters EA (2018). Don’t know responses to cognitive and affective risk perception measures: Exploring prevalence and socio-demographic moderators. British journal of health psychology 10.1111/bjhp.12296 [DOI] [PMC free article] [PubMed]
  35. Janssen E, Waters EA, van Osch L, Lechner L, & de Vries H (2014). The importance of affectively-laden beliefs about health risks: the case of tobacco use and sun protection. J Behav Med 10.1007/s10865-012-9462-9 [DOI] [PMC free article] [PubMed]
  36. Janz N, & Becker MH (1984). The health belief model: A decade later. Health Education Quarterly, 11(1), 1–47. [DOI] [PubMed] [Google Scholar]
  37. Kiviniemi MT, Orom H, Waters EA, McKillip M, & Hay JL (2018). Education-based disparities in knowledge of novel health risks: The case of knowledge gaps in HIV risk perceptions. British Journal of Health Psychology 10.1111/bjhp.12297 [DOI] [PMC free article] [PubMed]
  38. Klein WMP (2003). Self-Prescriptive, Perceived, and Actual Attention to Comparative Risk Information. Psychology & Health, 18(5), 625–643. 10.1080/0887044031000118465 [DOI] [Google Scholar]
  39. Klein WMP, & Stefanek M (2007). Cancer risk elicitation and communication: lessons from the psychology of risk perception. CA A Cancer Journal for Clinicians, 57, 147–167. [DOI] [PubMed] [Google Scholar]
  40. Lawton R, Conner M, & McEachan R (2009). Desire or Reason: Predicting Health Behaviors From Affective and Cognitive Attitudes. Health Psychology, 28(1), 56–65. [DOI] [PubMed] [Google Scholar]
  41. Lechner L, de Vries H, & Offermans N (1997). Participation in a breast cancer screening program: Influence of past behavior and determinants on future screening participation. Preventive Medicine, 26(4), 473–482. 10.1006/pmed.1997.0161 [DOI] [PubMed] [Google Scholar]
  42. Lipkus IM, Samsa G, & Rimer BK (2001). General performance on a numeracy scale among highly educated samples. Medical Decision Making, 21(1), 37–44. [DOI] [PubMed] [Google Scholar]
  43. Loewenstein GF, Weber EU, Hsee CK, & Welch N (2001). Risk as feelings. Psychological bulletin, 127(2), 267–286. [DOI] [PubMed] [Google Scholar]
  44. Lundgren RE, & McMakin AH (2013). Risk communication: A handbook for communicating environmental, safety, and health risks New Jersey: John Wiley & Sons. [Google Scholar]
  45. Magnan RE, Köblitz AR, Zielke DJ, & McCaul KD (2009). The effects of warning smokers on perceived risk, worry, and motivation to quit. Annals of Behavioral Medicine, 37(1), 46–57. 10.1007/s12160-009-9085-8 [DOI] [PubMed] [Google Scholar]
  46. Morris B, Lawton R, McEachan R, Hurling R, & Conner M (2016). Changing self-reported physical activity using different types of affectively and cognitively framed health messages, in a student population. Psychology, health & medicine, 21(2), 198–207. [DOI] [PubMed] [Google Scholar]
  47. Moser RP, McCaul K, Peters E, Nelson W, & Marcus SE (2007). Associations of perceived risk and worry with cancer health-protective actions. Journal of Health Psychology, 12(1), 53–65. 10.1177/1359105307071735 [DOI] [PubMed] [Google Scholar]
  48. National Cancer Institute. (2009). Health Information National Trends Survey (HINTS) items for years 2003–2007 Retrieved from https://hints.cancer.gov/view-questions-topics/all-hints-questions.aspx
  49. Nelson WL, Moser RP, & Han PK (2013). Exploring objective and subjective numeracy at a population level: findings from the 2007 Health Information National Trends Survey (HINTS). J Health Commun, 18(2), 192–205. 10.1080/10810730.2012.688450 [DOI] [PubMed] [Google Scholar]
  50. Oh A, Shaikh A, Waters EA, Atienza A, Moser RP, & Perna F (2010). Health disparities in awareness of physical activity and cancer prevention: findings from the National Cancer Institute’s 2007 Health Information National Trends Survey (HINTS). J Health Commun, 15 Suppl 3, 60–77. 10.1080/10810730.2010.522694 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Portnoy DB, Ferrer RA, Bergman HE, & Klein WM (2014). Changing deliberative and affective responses to health risk: a meta-analysis. Health Psychology Review, 8(3), 296–318. 10.1080/17437199.2013.798829 [DOI] [PubMed] [Google Scholar]
  52. Ramirez AS, Finney Rutten LJ, Vanderpool RC, Moser RP, & Hesse BW (2013). Correlates and geographic patterns of knowledge that physical activity decreases cancer risk. J Prim Prev, 34(1–2), 31–39. 10.1007/s10935-012-0289-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Rogers RW (1975). A protection motivation theory of fear appeals and attitude change. The Journal of Psychology, 91, 93–114. [DOI] [PubMed] [Google Scholar]
  54. Savage LJ (1954). The foundations of statistics New York: Wiley. [Google Scholar]
  55. Schmiege SJ, Bryan A, & Klein WMP (2009). Distinctions between worry and perceived risk in the context of the theory of planned behavior. Journal of Applied Social Psychology, 39(1), 95–119. 10.1111/j.1559-1816.2008.00431.x [DOI] [Google Scholar]
  56. Schwartz PH (2009). Disclosure and rationality: comparative risk information and decision-making about prevention. Theor Med Bioeth, 30(3), 199–213. 10.1007/s11017-009-9111-7 [DOI] [PubMed] [Google Scholar]
  57. Schwarzer R, & Fuchs R (1995). Self-efficacy and health behaviours. In Conner M & Norman P (Eds.), Predicting Health Behaviour (pp. 169–196). Philadelphia, PA: Open University Press. [Google Scholar]
  58. Sheeran P (2014). Does heightening risk appraisals change peoples intentions and behavior? A meta-analysis of experimental studies. Psychological bulletin, 140(2), 511. [DOI] [PubMed] [Google Scholar]
  59. Slovic P, Fischhoff B, & Lichtenstein S (1980). Facts and fears: Understanding perceived risk. Societal risk assessment: How safe is safe enough, 4, 181–214. [Google Scholar]
  60. Slovic P, Peters E, Finucane ML, & MacGregor DG (2005). Affect, risk, and decision making. Health Psychology, 24(4), S35–S40. [DOI] [PubMed] [Google Scholar]
  61. U.S. Department of Health and Human Services. (2008). 2008 Physical Activity Guidelines for Americans Retrieved from
  62. Waters EA, Hay JL, Orom H, Kiviniemi MT, & Drake BF (2013). “Don’t Know” Responses to Risk Perception Measures: Implications for Underserved Populations. Medical Decision Making, 33(2), 271–281. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Waters EA, Kiviniemi MT, Orom H, & Hay JL (2016). “I don’t know” My Cancer Risk: Implications for Health Behavior Engagement. Annals of Behavioral Medicine 10.1007/s12160-016-9789-5 [DOI] [PMC free article] [PubMed]
  64. Weinstein ND, Kwitel A, McCaul KD, Magnan RE, Gerrard M, & Gibbons FX (2007). Risk perceptions: assessment and relationship to influenza vaccination. Health Psychology, 26(2), 146–151. [DOI] [PubMed] [Google Scholar]
  65. Willis GB (2004). Cognitive Interviewing: A Tool for Improving Questionnaire Design Thousand Oaks, CA: Sage Publications. [Google Scholar]

RESOURCES