Journal of Behavioral Medicine

, Volume 30, Issue 5, pp 403–409

Masculinity and Intentions to Perform Health Behaviors: The Effectiveness of Fear Control Arguments

Authors

    • Department of PsychologyUniversity of Nevada
  • Jeremy Ashton Houska
    • Department of PsychologyUniversity of Nevada
Article

DOI: 10.1007/s10865-007-9113-8

Cite this article as:
Millar, M.G. & Houska, J.A. J Behav Med (2007) 30: 403. doi:10.1007/s10865-007-9113-8

Abstract

This study examined the impact of fear control and danger control messages on intentions to perform health behaviors. It was hypothesized that persons high in masculinity would be more influenced by messages that address fear than messages that exclusively focus on the efficacy of the behavior (danger control). To test this proposition, 172 participants were classified into high and low masculinity groups according to their scores on the Bem Sex Role Inventory, and randomly assigned fear reducing or danger control messages either encouraging detection behavior (skin cancer self-examination) or promotion behavior (sunscreen usage). Highly masculine participants demonstrated greater behavioral intentions and more positive attitudes about the behavior when exposed to a fear-reducing message. The results suggest that it is possible to motivate adaptive health behavior even when a person is engaging in fear control.

Keywords

Extended parallel process modelPersonality characteristicsMasculinityDetection behaviorsPrevention behaviorsSkin cancer

Introduction

Health messages often deal with threatening subject matter. Calls to examine oneself for skin cancer or have one’s blood pressure checked by their very nature remind us of our vulnerability and mortality. Fittingly, the extant literature has examined how persons respond to threatening health information (see Ruite et al. 2001 for a review) and several models have been proposed, e.g., Protection Motivation Theory (Rogers 1983) and the Health Belief Model (Janz and Becker 1984; Rosenstock 1974).

Prominent in this literature is the Parallel Processing Model proposed by Leventhal (1970) that was later modified and expanded by Witte (1992, 1994) into the Extended Parallel Process Model (EPPM). The EPPM provides a description of how persons deal with threatening information. In essence, the EPPM argues that threatening information can stimulate two cognitive appraisals: danger control or fear control. Danger control occurs when persons believe they are vulnerable to the health threat and believe they can effectively ameliorate the threat, i.e., they have perceived self and response efficacy. Ultimately, the danger control process elicits protection motivation that causes the person to confront the danger and engage in adaptive behavior. Fear control occurs when persons believe that they are vulnerable to a health threat but cannot effectively ameliorate the threat or that they are incapable of performing the behavior. When fear control dominates, it elicits defensive motivation that causes the person to focus on coping with fear not danger. Instead of changing behavior, persons defensively avoid the health issue or deny the risk associated with their current behavior.

The major implication of the EPPM is that messages designed to promote health behavior need to motivate persons to engage in danger control. That is, a health message should cause the person to perceive a serious threat, convince the person of the efficacy of the health behavior, and convince the person that he/she can perform the behavior. Many recent studies have demonstrated the effectiveness of this persuasion strategy with a variety of health topics, e.g., Gore and Campanella (2005) with messages promoting meningitis vaccinations; Murray-Johnson et al. (2004) with a procedure to promoting utilization of hearing protection, and McKay et al. (2004) with messages about cardiovascular risk. Earlier work examined in a meta-analysis by Witte and Allen (2000) provided support for the model.

Yet what about situations when fear control process are initiated? The EPPM’s suggests that fear control often leads to a rejection of the message or denial of the threat. Regrettably, in the real world many variables predispose persons to engage in defensive motivation. Often, perceived response efficacy is difficult to create because many health behaviors have only limited effectiveness, i.e., they only partially ameliorate the disorder. Further, other health behaviors are designed to detect disorders and have no immediate ability to reduce threat (see Millar 2004 for discussion of disease detection behaviors). Beyond the characteristics of the behavior, personality characteristics may predispose persons to engage in fear control. For example, Ruiter et al. (2004) found a relationship between fear control and a person’s inclination to engage in and enjoyment of cognitive activity (need for cognition). Persons who are low in the need for cognition are more likely to engage in fear control.

Hence, it seems probable that fear control process will operate or at least partially operate with many health behaviors. This raises an important question—is it possible to motivate adaptive behavior when fear control dominates? Conceivably a health message could be constructed that would cause defensive motivation to enhance acceptance of message. That is, we could employ a person’s desire to control fear to motivate him/her to accept the message and perform an adaptive response. We are proposing that if a person has the motive to control fear then a message that targets this motive can be successful in changing behavior. There is considerable evidence that targeting a persuasive messages at the functional bases of the attitude is an effective means of changing attitudes and behavior (e.g., Bailis et al. 2005; Clary et al. 1994; Petty and Wegener 1998). A straightforward method to do this might be to present the health behavior as both a means to control the health threat and a means to control fear. In a manner, we are broadening the concept of response efficacy to encompass both coping with threat and coping with fear.

Masculinity and Fear Control

Are there populations likely to engage in fear control? In this context relatively few individual differences variables have been examined with the notable exceptions of state anxiety and need for cognition (Ruiter et al. 2004). When it comes to propensity for danger control, an individual difference that might warrant attention is masculinity. Masculinity has been conceptualized as an instrumental orientation that is characterized by independence, assertiveness, and dominance behaviors (Bem 1981). This construct has been widely studied and related to many divergent behaviors. For example, masculinity has been implicated in criminal behavior (Walters 2001), eating disorders (Meyer et al. 2001), and loneliness (Cramer and Neyedley 1998). Although masculine traits are more associated with males this personality orientation is not necessarily related to gender. Indeed Twenge (1997) found that since Bem’s original work gender differences are disappearing. At this point in time, it may be more appropriate to label masculinity as an instrumental orientation recognizing that these traits fluctuate within a person regardless of gender.

The most relevant feature of masculinity for the present research is its impact on responses to emotional distress. Conway and his colleagues have found that high masculinity persons engage in more distraction when feeling sad and focus more on positive events when feeling distress than do persons low in masculinity (Conway and Dube 2002; Conway et al. 1990, 1991). Consistent with Conway’s findings, Oliver and Toner (1990) demonstrated that when high masculinity persons are depressed they attempt to focus on positive behavior as opposed to negative feelings. Furthermore, Jakupcak et al. (2005) findings suggest that high masculinity persons avoid emotions altogether. A type of emotional distress that high masculine persons particularly try to avoid is feeling vulnerable due to the experience of pain, grief, or fear (Jansz 2000). The thrust of this research indicates that persons high in masculinity are more motivated to control feelings of fear than persons low in masculinity.

Purpose of this Research

The purpose of this study is to explore whether persons can be motivated to perform adaptive health behavior when fear control processes are operating. It is our contention that when persons are disposed to control fear, communications that put forward the health behavior as a way to abate fears would lead to acceptance of the message. Further, based on a review of the evidence, we would suggest that persons high in masculinity are predisposed to avoid negative emotions such as fear. This leads us to the prediction that for persons high in masculinity, messages that address fear may be more influential than messages that exclusively focus on the efficacy of the behavior. To test this prediction, we presented fear control, danger control, or informational messages to persons who were high or low masculinity.

Methods

Participants

One hundred and fifteen female and 57 male participants were recruited from a large urban university in the southwest United States. Participants were recruited on a voluntary basis and offered class credit in exchange for participation. No monetary compensation was given for participation. The average age of the participants was 21 and the range of ages was 18–61 years of age. Fifty-two percent of the participants were of European descent, 12% were of Hispanic descent, 20% were of Asian descent, 6% were of African descent, and 10% were from other ethnic groups. On average participants had approximately one and half year of post-secondary education. Participants were randomly assigned to receive one of the six messages.

Materials

Six messages encouraging the performance of health behaviors were used in the study. Three messages encouraged detection skin cancer self-examinations (disease detection behavior), and three messages encouraged wearing sunscreen when exposed to bright sunlight. Skin cancer is presumed to be a relevant concern to our sample of young adults because of their frequent exposure to UV rays and the potentially high-risk social norms previously demonstrated in research utilizing this age group (e.g., maintaining a tanned complexion, changes in attitudes toward sun protection during adolescence) (see Saraiya et al. 2004 for a review).

In both message groups, there was an informational message that was constructed to provide basic information about the performance of the behavior. For example, the message about skin self-examination pointed out that the purpose of a regular skin exam is to become familiar with markings on your body and that it takes only about 10 min for perform. The informational message about wearing sunscreen pointed out that it is an effective way to reduce UV exposure and is relatively inexpensive to buy and only takes minutes to apply. The danger control messages emphasized the effectiveness of the behavior in reducing the health risk. For example, the danger control message about skin self-examination pointed out that self-examinations are often the first step in detecting skin cancer and that skin cancer it is almost always curable if it is caught in the early stages. The danger control messages about sunscreen pointed out that reducing UV exposure can significantly reduce the risk of developing skin cancer. The fear control messages emphasized the effectiveness of the behavior in reducing the negative emotions associated with skin cancer. For example, the fear control message indicated that regularly wearing sunscreen will allow you to avoid some of the worry and distress associated with skin cancer.

Procedure

Participants were informed that the purpose of the study was to investigate general and specific reactions to some current health issues. To study these reactions the participants would be required to indicate their general state of health and performance of health behaviors on a paper-based health behavior survey. After reassuring the participants their responses would be completely confidential, the experimenter seated the participants at desks and presented the study materials to the participants. At the beginning of the session all participants were required to complete the 60-item Bem Sex Role Inventory (BSRI) (Bem 1974). Then participants were asked to short demographic questionnaire and three items on their general state of health. Following this, participants were asked to complete a 15-item general health behavior survey, which assessed the frequency of their adherence to healthy behaviors on a 9-point scale with endpoints of 9 (always) and 1 (never). Imbedded with the other health behaviors there was an item on how often they wore sunscreen and an item on how often they performed skin self-examinations. After the initial questions, the participants were asked to read a message that encouraged the performance of the behavior.

Behavioral Intentions and Attitude

After the participants had finished reading the message, the participants’ intentions to perform the behavior were measured on three scales. The scales had with endpoints of 1 (unlikely) and 9 (likely), 1 (improbable) and 9 (probable), and 1 (impossible) and 9 (possible). Then the participants’ attitudes about the behavior were assessed with four scales. Participants gave their reaction to the behavior on three scales with endpoints of 1 (dislikable) and 9 (likable), 1 (unpleasant) and 9 (pleasant), and 1 (positive) and 9 (negative). In addition, participants were as to indicate how much they agreed with the message on a scale with endpoints of 1 (disagree) and 9 (agree).

Perception of Behaviors and Messages

When the attitude scales were completed, the participants were asked indicate how effective “sunscreen” or “self-examination” is in reducing skin cancer on three scales. These scales had endpoints of 1 (not effective) and 9 (very effective), 1 (not helpful) and 9 (very helpful), and 1 (not useful) and 9 (very useful). Then participants were asked to indicate the amount of distress associated with the behaviors on three nine-point scales with endpoints of 1 (less distress) and 9 (more distress), 1 (less worry) and 9 (more worry), and 1 (less anxiety) and 9 (more anxiety). Participants were also asked to evaluate the message on three nine-point scales. These scales had end points of 1 (weak) and 9 (strong), 1 (low quality) and 9 (high quality), and 1 (not persuasive) and 9 (persuasive).

Behavioral Indicator

As the participant was preparing to leave the study, the experimenter told the participant he/she could take a pamphlet that contained more information about the health behavior. Then the experimenter pointed to a table located by the exit that had a number of pamphlets on it. The pamphlet in sunscreen conditions was labeled facts about sunscreen and contained information about how to select a good sunscreen. The pamphlet in the self-examination conditions was an instruction manual on how to perform a self-examination. After the participant had taken or not taken a pamphlet, he/she was questioned by the experimenter about his/her understanding of the experimental hypotheses, then fully briefed, and released from the experiment.

Results

Behavioral Intentions

To examine the main hypothesis participants were classified into a high masculinity group or low masculinity group on the bases of a median split (median = 96). See Bem (1978) for methods of scoring the BSRI. The scores from the three behavior intention scales were averaged to form a single index (Chronbach alpha = .91) and examined a 2 (Behavior Type: sunscreen versus skin examination) × 2 (Masculinity) × 3 (Argument Type) analysis of covariance (ANCOVA). The covariate was constructed from the initial questions about how often the participants performed different health behaviors. In the sunscreen conditions the participants’ responses to how often they wore sunscreen was used as a covariate and in the skin-examination conditions the participants’ responses to how often they had performed a skin cancer self examination was used as a covariate. The covariate was significantly related to the behavioral intention measure, r(169) = .38, p < .01. As expected, a significant interaction between Argument Type and Masculinity was obtained, F(2, 156) = 4.93, p = .008, ηp2 = 10). The fear-reducing message created greater intention to perform the behavior for persons high in masculinity than for persons low in masculinity, F(1, 156) = 4.09, p < .05, η2 = .06.1 The danger control and informational messages produced no significant differences between the high and low masculinity groups, (see Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs10865-007-9113-8/MediaObjects/10865_2007_9113_Fig1_HTML.gif
Fig. 1

Mean behavioral intention (+SE) as a function of masculinity, argument type, and behavior type

In addition, to the predicted interaction, two other significant effects emerged. First, a main effect of Behavior Type was found, F(1, 156) = 5.56, p < .05, ηp2 = .04. Participants indicated more intention to perform self-examinations (M = 6.97) than wear sunscreen (M = 6.21). Second, a significant interaction between Masculinity and Behavior Type was found, F(1, 156) = 4.47, p < .05, ηp2 = .03. With the detection behavior (skin cancer self-examination) persons high in masculinity had less intention to perform the behavior (M = 6.59) than persons low in masculinity (M = 7.34), F(1, 156) = 5.19, p < .05, ηp2 = .04. With the promotion behavior (wearing sunscreen) there was no significant difference between high and low masculine persons’ intentions to perform the behavior, (p > .05).

Behavioral Indicator

A similar pattern of results was obtained with the behavioral indicator (taking a pamphlet about either sunscreen or skin self-examination). A logistic regression analysis found a significant interaction between Masculinity and Argument Type, b = 1.21, t = 2.11, p < .05 (Odds Ratio = 3.34). The fear-reducing message created greater intention to perform the behavior with person high in masculinity than with persons low in masculinity, b = 1.24, t = 2.22, p < .05 (Odds Ratio = 3.45). With danger control there was a non-significant tendency for low masculinity persons to have a greater intention to perform the behavior than high masculinity persons, b = .51, t = 1.78, p = .07 (Odds Ratio = 1.66). Informational messages produced no significant differences between the high and low masculinity persons, (see Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs10865-007-9113-8/MediaObjects/10865_2007_9113_Fig2_HTML.gif
Fig. 2

Mean behavior indicator (+SE) function of masculinity, argument type, and behavior type

Attitudes

The scores from the four attitude scales were averaged to form a single attitude index (Chronbach alpha = .78) that was examined in the same three factor ANCOVA used with behavior intentions. A significant interaction between Masculinity and Argument Type was found, F(2, 159) = 3.18, p < .05, ηp2 = .04. The fear-reducing message created more positive attitudes about the behavior for persons high in masculinity (M = 6.36) than for persons low in masculinity (M = 5.48), F(1, 159) = 5.44, p < .05, η2 = .10. With the danger control message there was not a significant difference between high (M = 5.64) and low masculinity (M = 5.93) participants, F < 1. Further, informational messages produced no significant differences between persons high and low masculinity, F < 1.

Perception of Messages and Behaviors

The three scores from the questions about the quality of the messages were averaged to form a single attitude index (Chronbach alpha = .86) that was examined in the standard three factor ANCOVA. A main effect for Masculinity was the only significant effect found, F(1, 159) = 5.72, p < .05, ηp2 = .04. Persons high in masculinity believed the messages were higher in quality (M = 7.31) than persons low in masculinity (M = 6.87).

To explore the how participants perceived the behaviors two indexes were created. First, an index of the amount of distress associated with the behaviors was created by averaging the participants’ responses to the three questions on distress (Chronbach alpha = .94). Second, an index of how effective the behavior was in combating the health problem was created by averaging responses to the three questions on effectiveness (Chronbach alpha = .96). Both of these indexes were examined separately in the standard three factor ANCOVA. When distress was examined a significant main effect for the type of behavior was obtained, F(1, 159) = 9.10, p = .003, ηp2 = .05. The detection behavior (skin cancer self-examination) was associated with more distress (M = 3.19) than wearing sunscreen (M = 2.38). In addition, a significant main effect for masculinity was found, F(1, 159) = 4.21, p < .05, ηp2 = .03, with persons high in masculinity reporting less distress (M = 2.51) than person low in masculinity (M = 3.06). When effectiveness was examined only a main effect for the type of behavior was found, F(1, 158) = 6.27, p = .01, ηp2 = .04.2 Wearing sunscreen was perceived as more effective (M = 8.11) than skin self-examination (M = 7.94).

Other Analyses

The participants’ responses to the three items on how they perceive their general state of health were averaged to create a single indexes (Chronbach alpha = .78) and analyzed in a 2 (Masculinity) × 3 (Argument Type) analysis of variance (ANOVA). A main effect for masculinity was the only significant effect obtained, F(1, 159) = 10.57, p < .01, η2 = .06. Persons high in masculinity reported being more healthy (M = 7.10) than person low in masculinity (M = 6.48). In addition, the participants’ responses to each of the items in the 15-item general health behavior survey were analyzed in separate one factor (Masculinity) ANOVAs. A Bonferroni correction was applied to these analyses because of their post hoc nature. With the correction, no significant differences were found.

Discussion

The findings provided support for our predictions. With both the behavioral intention measure and the behavior indicator measure, high masculinity participants were more influenced by the fear control messages than low masculinity participants. With informational messages, there was a non-significant tendency for this effect to reverse. An analogous but weaker pattern of results was obtained with the attitude measure. These finding suggest that a person’s desire to control fear could be employed to motivate acceptance of the message. Additionally, the findings point to the possibility that individual differences may influence the propensity for persons to engage in danger control when confronted with a health threat.

Beyond the predicted results the type of behavior presented (sunscreen versus self-examination) produced a two unpredicted effects. First, participants indicated that thinking about skin self-examinations caused more distress than thinking about wearing sunscreen. This finding is consistent with past research indicating that behaviors designed to detect illness cause more anxiety than other types of health behaviors (e.g., Erblich et al. 2000; Lerman et al. 1991; Millar 2005; Millar and Millar 1996; Murray and McMillan 1993). Second, wearing sunscreen was perceived as more effective than skin self-examination. This is consistent with the notion that detection behaviors (e.g., self-examinations) do not provide the immediate ability to reduce threat.

The masculinity variable also produced a two unexpected effects. High masculinity persons reported less distress associated with the health behaviors than low masculinity persons. This may seem at odds with our contention that high masculinity persons focus on fear control, i.e., why focus on fear control if you are not distressed? However, there is reason to suspect self-reports of distress made by high masculinity persons. These reports may simply reflect a stereotypic reluctance by high masculinity persons to report or admit weakness (Kaplan and Marks 1995; Pollack 1999; Radley et al. 2000). In addition, high masculinity persons reported being more healthy that low masculinity persons. It is difficult to interpret this difference because it could reflect an actual difference in health or it could simply reflect the tendency for high masculinity persons to focus on more positive events (Conway et al. 1991).

Lastly, an unexpected interaction between masculinity and type of behavior was found. High masculinity persons indicated less intention to perform the skin cancer self-examination (detection behavior) than persons low in masculinity. An examination of our finding provides one possible explanation. We found that skin self-examinations were viewed as more distressing and less effective than wearing sunscreen. If as evidence suggests, persons high in masculinity have a lower tolerance for distress than persons high in masculinity (e.g., Conway and Dube 2002; Conway et al. 1990; Conway et al. 1991) then the highly masculine participants engaged more in danger control and rejected our health messages about self-examinations.

As frequently happens with research, the present findings both have limitations and generate questions for future research. One limitation of the current is study it that fear was not directly measured. Fear was not measured prior to the presentation of the messages because it was possible that the measure would have influenced responses to the messages, e.g., the measurement of fear might prime memories of fear causing participants to be more receptive to the fear-reduction messages.

One question that needs answering, also suggested by Ruiter et al. (2004), is whether there are other individual difference variables beyond masculinity and the need for cognition (see Cacioppo and Petty 1982) that influence the propensity to engage in fear control. For instance, how does a person’s locus of control (see Rotter 1954; see also Murray and McMillan 1993) relate to danger control? Plausibly an external locus of control (i.e., the belief that events are controlled by external forces) reduces feelings of self-efficacy thus increasing the likelihood of fear control. Conversely, one with an internal locus of control would arguably engage in danger control, which leads to the more adaptive behaviors. Williams-Piehota et al. (2004) have already demonstrated the effectiveness of matching messages to locus of control beliefs. Further, emotional stability, a component of many personality factor models (Costa and McCrae 1985), may influence the tendency to engage in danger control, i.e., a person with high emotional stability may be more liable to use danger control than a person lower in emotional stability. Distress tolerance is another in individual difference that might influence the tendency to engage is danger control. Distress tolerance is a person’s ability to experience and cope with negative emotion (Simons and Gaher 2005). Persons low in distress tolerance who attempt to avoid negative emotions might be more likely to engage in fear control. An additional question generated by the present findings is whether characteristics of the health behavior can influence the likelihood of danger control. The current findings already hint that the detection behaviors might be more likely to stimulate danger control.

Finally, what is the relationship between fear control and danger control processes? The EPPM focuses on situations were either the fear or the danger control process dominates. For example, EPPM proposes that fear control process begins to dominate at a critical point when the individual realizes that the behavior is ineffective or that he/she is incapable of performing the behavior. However, the present findings raise the possibility that in some situations or with some persons that both fear and danger control processes operate simultaneously. Even when the fear control process is not dominant, it might still be beneficial to add a fear control component to traditional health promotion messages.

Footnotes
1

Three participants failed to complete one of the behavioral intention scales and were not included in this analysis.

 
2

One participant failed to complete one of the effectiveness scales and was not included in this analysis.

 

Copyright information

© Springer Science+Business Media, LLC 2007