Acting on surprise: emotional response, multiple-channel information seeking and vaccination in the H1N1 flu epidemic

Abstract

A survey conducted among college students (N = 205), one of the groups most susceptible to the H1N1 influenza virus, examined how people responded to the unexpected public health crisis from their initial emotional reaction, to information seeking, to vaccination behavior. Results showed that surprise upon first learning of the H1N1 flu was positively related to multiple-channel information seeking. People who used multiple channels to get updated about the pandemic reported more favorable attitudes about getting vaccinated against the H1N1 flu and were more likely to regard getting vaccinated as the norm. These cognitive antecedents were positively related to their actual vaccination behavior.

Keywords::

• surprise
• multiple-channel information seeking
• TPB
• H1N1 flu vaccination

The unusual and virulent nature of the H1N1 flu epidemic surprised and frightened the populations worldwide. The disease was first found in Mexico in April 2009, and was quickly identified as a major global public health risk. The epidemic spread rapidly to many other parts of the world, due largely to international passenger travel. On 11 June 2009, the World Health Organization (WHO) signaled that a global pandemic of a novel influenza A (H1N1) was underway by raising the worldwide pandemic alert to its highest level—Phase 6 (WHO, 2009).

As a newly found virus, the epidemic evoked a wide range of emotions (Forbes, 2009). When the epidemic first broke out, there was little immunity against the virus in the general population, and no vaccine was available (Centers for Disease Control and Prevention [CDC], 2010). Some individuals infected early on even died from the flu (CDC, 2010). In addition, the new flu virus demonstrated many unique characteristics. Unlike the seasonal flu, young adults were far more susceptible to the H1N1 virus (CDC, 2009). The deadly nature of the virus and its unusual profile prompted a range of emotions including surprise, which may have triggered many more information seeking behaviors than those that are commonly associated with the seasonal flu. As previous studies suggested, unexpected pandemics often motivate more information seeking from a variety of sources (Holton, 2010; Morahan-Martin, 2004).

Past research in crisis communication has acknowledged the power of emotion in influencing information seeking behaviors (Boyle et al., 2004; Kubey & Peluso, 1990). However, most of these studies viewed emotional reactions as a single entity or divided them broadly into two categories, positive and negative emotions. Surprise, as a discrete emotion, has not been adequately studied for its relationship with information seeking. However, defined as an emotional response to a stimuli-schema discrepancy (Alden, Mukherjee, & Hoyer, 2000), surprise often leads to investigative and exploratory activities with the goal to understand the schema discrepant event (Horstmann, 2006; Meyer, Reisenzein, & Schützwohl, 1997). Therefore, surprise has a natural connection with information seeking, as the search for information could be the means to understanding the unexpected event.

The current study examines the effect of surprise, a less studied emotional response in the literature of information seeking and processing, on information seeking behaviors in the context of the H1N1 flu epidemic. Surprise is a natural emotional reaction to an unexpected public health crisis. Given that young adults were particularly susceptible to the H1N1 flu virus, this study explores how the surprise that college students felt after first learning of the H1N1 flu pandemic influenced them to seek information related to the epidemic with the guidance of the model of risk information seeking and processing (RISP). We also link information seeking with cognitive appraisals of the vaccination decision using the theory of planned behavior (TPB). According to TPB, cognitive antecedents including attitudes towards the behavior, subjective norms and perceived behavioral control all predict a behavioral decision (Ajzen, 1991, 2002). Therefore, in this study, we investigate how information seeking from multiple sources may influence attitudes, subjective norms, and perceived behavioral control. Given that information seeking is the basis for acquiring related beliefs and attitudes, we expect that such activities would be associated with these cognitive antecedents of the actual vaccination behavior.

Literature review

Surprise and multiple-channel information seeking

The RISP model was developed to investigate how information seeking can be activated. According to the model, emotional responses, which may change people's perceptions about the adequacy of the information needed about a social issue and thus change the gap between their current knowledge and the sufficiency threshold, have been identified as an important predictor of information seeking (Griffin, Dunwoody, & Neuwirth 1999).

Using the RISP model, scholars have tested the relationship between multiple emotional responses and information seeking intentions with a valence model in which emotional reactions are categorized as either positive or negative (Kahlor, 2010; Yang & Kahlor, 2013; Yang et al., 2011). The typical negative emotions that have been examined include worry, fear, and anxiety. These studies have concluded that negative emotions can increase information seeking intentions in general across communication contexts (Kahlor, 2007, 2010; Yang & Kahlor, 2013; Yang et al., 2011). Investigations into the effects of positive emotions on information seeking have been inconclusive. Whereas optimism motivated the seeking of information about clinical trials (Yang et al., 2010, 2011), other positive emotions including excitement, happiness and hope showed no significant relationship with intentions to seek information about global warming (Yang & Kahlor, 2013). Surprise, as a discrete emotion, has been under-investigated in terms of its relationship with information seeking in RISP studies. However, the connection between surprise and information seeking is potentially strong, making the understanding of the connection clearly of interest.

Surprise is defined as “a primary individual-level response to stimulus-schema incongruity” (Alden et al., 2000). In other words, surprise can be induced by stimuli that are incongruent with familiar schemas (Alden et al., 2000) or unexpected events (Stiensmeier-Pelster, Martini, & Reisenzein, 1995). Understanding the stimuli or unexpected events then leads to a series of appraisal and reappraisal processes. For instance, people tend to pause in their ongoing activities and reallocate their cognitive attention. They analyze and evaluate the unexpected event in terms of its cause, its significance for their well-being and its relevance to their ongoing activities (Meyer et al., 1997).

These cognitive processes demonstrate a link between surprise and information seeking. As previous research has suggested, when people feel surprised, they engage in investigative and exploratory activities to understand the unexpected event (Meyer et al., 1997). These activities may include seeking information to become more informed about the unexpected event. Tiedens and Linton (2001) may be the scholars who are more supportive of this view even though their study focused on information processing rather than information seeking. According to them, since surprise is an uncertainty-associated emotion, it often drives people to engage in efforts that can restore their confidence and certainty. In the context of information processing, when such an emotion is elicited, individuals are more likely to be involved in the systematic information-processing mode (Tiedens & Linton, 2001). Therefore, it is also reasonable to propose that surprise activates the systematic and effortful search for information that must necessarily precede the processing of this information.

Multiple-channel information seeking is operationalized as a type of information seeking associated with a great deal of effort. Information seekers go beyond their habitual use of information sources and purposely enhance their efforts by acquiring information from multiple channels (Griffin et al., 1999). This type of information seeking, prompted by a perceived insufficiency of information, is a typical example of non-routine information seeking, because it is effortful rather than convenient. Griffin et al. (1999) conceptualized multiple-channel information seeking when developing the RISP model. However, the concept has rarely been measured in the published studies of RISP. This type of information seeking is compatible with our view of systematic and effortful information seeking. Therefore, we propose a positive relationship between surprise upon first hearing about the H1N1 flu epidemic and multi-channel information seeking.

H1

=

Surprise upon first hearing of the H1N1 pandemic is positively related to multi-channel information seeking.

Multiple-channel information seeking and preventive behavior

RISP has been considered an antecedent of preventive behaviors because such activities can alter the cognitive structure of a behavioral decision (Griffin et al., 1999). Griffin et al. (1999) argue that the way of seeking and processing risk information will result in different cognitive structures of a behavioral decision. According to them, people who systematically and effortfully seek information develop a more stable cognitive structure by bringing more behavioral beliefs to make the behavioral decisions. In contrast, people who seek information effortlessly generate a much simpler cognitive structure to determine whether the behavior should be performed. The stability between these cognitive appraisals and the behavioral decision will also be less stable (Griffin et al., 1999).

The TPB specifies the cognitive structure of a behavioral decision by arguing that the conscious decisions people make to engage in health behaviors are determined by their cognitive appraisals including their attitudes towards the behavior, subjective norms and perceived behavioral control (Ajzen, 1991, 2002). The link between information seeking and these cognitive appraisals is not new. In the context of the H1N1 flu epidemic, previous studies found that seeking information from media and governmental sources helped create knowledge about the flu and increased the perceived effectiveness of hygienic behaviors to prevent flu infection (Liao, Cowling, Lam, & Fielding, 2011), a behavioral belief that promotes the performance of such behaviors.

Given that multiple-channel information seeking is a typical systematic and effortful approach to acquiring information, we hypothesize that it will have a significant positive relationship with the cognitive antecedents of vaccination behavior, including attitudes, subjective norms and perceived behavioral control. As the TPB suggests, these cognitive antecedents positively predict the vaccination decision. Therefore, we posit the following hypotheses:

H2	=	Multiple-channel information seeking is positively related to the TPB antecedents of vaccination behavior, including (a) attitudes towards the vaccination behavior, (b) subjective norms, and (c) perceived behavioral control.
H3	=	TPB antecedents, including (a) attitudes towards the vaccination behavior, (b) subjective norms, and (c) perceived behavioral control, positively predict vaccination behavior.

Method

Data collection and sample

An online survey was conducted immediately following the 2009–2010 flu season in Spring 2010. As young people were especially susceptible to the H1N1 flu virus (American College Health Association, 2010), the study was conducted among college students. In total, 205 college students who were enrolled in an introductory communication class at the Pennsylvania State University were recruited to participate in the study. The sample consisted of more female (77.1%) than male students (22.9%). The majority of the respondents were Caucasian American (84.9%), and the average age was 20.2 (SD = 1.19).

Instrumentation

Surprise

Using scales developed by Shen and Dillard (2007), surprise was measured by asking to what degree the respondents felt astonished and surprised when they first learned about the disease. Seven-point scales were used ranging from 0 = none of this feeling to 6 = a great deal of this feeling. The scale was reliable (r = .77, p < .01). An overall mean was computed.

Multiple-channel information seeking

Respondents were asked how frequently they used different communication sources to get updated about the H1N1 flu during the epidemic until they decided whether they would or would not be vaccinated against the flu. Seven-point scales were used ranging from 0 = used very little to 6 = used very much. The communication sources included (1) traditional mass media, (2) family/friends through online social media, (3) family/friends through one-to-one communication, (4) doctors/healthcare professionals, and (5) school administration/announcements.

We measured the number of channels, in which the respondents used to seek information about the H1N1 flu, by recoding the five information seeking frequency variables above for each respondent. When the rating for each source was higher than the mid-point of the scale (3), it was recoded as high frequency = 1, while when the rating was lower than the mid-point of the scale, it was recoded as low frequency = 0. The number of information sources that the respondents used with high frequency was calculated by summing the five newly recoded variables, which could possibly range from 0 to 5. The new variable reflected the efforts respondents made to intentionally expand their number of information sources. It also excluded accidental exposure as a possible confounding variable in understanding this phenomenon.

Attitudes

TPB variables were measured by asking people's thoughts at the moment when they decided whether or not to get the H1N1 flu vaccine. Using 7-point scales, attitudes toward vaccination were measured by asking respondents whether they thought getting the H1N1 vaccine would be harmful/beneficial, unpleasant/pleasant, bad/good, and worthless/valuable when they made the decision (α = .83). An overall mean was computed for attitudes.

Subjective norms

Subjective norms were measured using five items focusing on the respondents' perceptions of the approval from important others to become vaccinated at that moment. Sample items included whether most people who were important to the respondents thought they should/should not get the H1N1 flu or approved/disapproved of the vaccination decision. Respondents were also asked to evaluate on a scale ranging from extremely unlikely to extremely likely whether people who were important to them had received the H1N1 flu shot. The responses on this scale ranged from 0, indicating the lowest approval, to 6, indicating the highest approval (α = .88). An overall mean was computed for subjective norms.

Perceived behavioral control

Two items were used to measure the respondents' perceived ability to get the H1N1 flu shot. They were asked about the degree to which they agreed with statements that if they wanted to (1) they were able to or (2) it was easy for them to get the vaccine (0 = strongly disagree, 6 = strongly agree). As the two-item scale was reliable (r = .78, p < .001), an overall mean for perceived behavioral control was computed.

Vaccination

At the end of the questionnaire, the respondents were asked whether they had been vaccinated against the H1N1 flu. Twenty-two percent of the participants reported that they received the H1N1 flu shot during the epidemic.

Control variables

Demographic variables including gender and age were measured at the beginning of the questionnaire. Previous studies suggest that gender and age often influence health information seeking behaviors and vaccination intentions and behavior. For instance, health information seeking behaviors are often more frequent among females than males, and increase with age (National Cancer Institute, 2005). Indeed, research has demonstrated that age is a predictor of H1N1 flu vaccination intentions (Myers & Goodwin, 2011), and both age and gender matter in vaccination decisions in regular flu seasons (CDC, 2011). Therefore, in this study, we controlled for gender and age in our data analysis.

Data analysis

We examined three sub-processes in response to the H1N1 flu epidemic including (1) how surprise influenced the multiple-channel information seeking (H1), (2) how multiple-channel information seeking influenced the cognitive antecedents of vaccination behavior (H2), and then (3) how the cognitive antecedents influenced the vaccination behavior (H3). Given that the dependent variables for H1 and H2 were continuous variables, we conducted linear regression analyses to test H1 and H2. Given that the dependent variable for H3 was a nominal variable, we used a logistic regression analysis to test H3. Gender and age were included as the covariates in these analyses. To prevent false positives, each regression analysis was performed with and without covariates. All variables measured in the current study are reported in the instrumentation section and the results are reported below.

Additional structural equation modeling analysis was performed using Mplus 7.1. The model used surprise as the independent variable, multiple-channel information seeking as the mediating variable, and all three cognitive antecedents of the vaccination behavior as the dependent variables. The categorical outcome—vaccination—was not included in the model. Table 1 presents the descriptive statistics for the variables included in the model.

Table 1 Correlation matrix and descriptive statistics.

Key variables	1	2	3	4	5
1. Surprise	1
2. Multiple-channel information seeking	.44^*^*^*	1
3. Attitudes	.14	.24^*^*	1
4. Subjective norms	.06	.12	.51^*^*^*	1
5. Perceived behavioral control	− .02	.12	.08	.20^*^*	1
M	2.62	2.63	3.49	3.29	4.68
SD	1.51	1.67	1.23	1.43	1.30

Note: ^**p < .01, ^***p < .001.

Results

H1 proposed that surprise was positively related to multiple-channel information seeking. As Table 2 shows, surprise was associated with multiple-channel information seeking (β = .44, p < .001) and its effect remained significant after controlling for gender and age (β = .41, p < .001). In other words, people who reported a higher level of surprise upon first hearing of the H1N1 flu epidemic used more channels to seek information to get updated about the epidemic. Therefore, H1 was supported.

Table 2 Relationship between surprise and multiple-channel information seeking.

	Model 1 (without covariates)		Model 2 (with covariates)
	β	p	β	p
Gender			.16	^*
Age			− .04
Surprise	.44	^*^*^*	.41	^*^*^*

Note: ^*p < .05, ^***p < .001.

H2 proposed that multiple-channel information seeking behavior had a positive relationship with TPB variables. As Table 3 shows, multiple-channel information seeking was positively related to attitudes (β = .24, p < .01). Furthermore, after controlling for gender and age, multiple-channel information was positively related to attitudes (β = .27, p < .001) and subjective norms (β = .14, p < .05). In other words, individuals who used more channels to get updated about the H1N1 flu epidemic reported more favorable attitudes, and higher subjective norms of receiving the H1N1 flu vaccination. However, multiple-channel information seeking was not related to perceived behavioral control. Therefore, as the results suggest, H2(a) was supported; H2(b) was partially supported; and H2(c) was rejected (Table 3).

Table 3 Relationships between multiple-channel information seeking and cognitive antecedents of the vaccination behavior.

	DV1: Attitudes		DV2: Subjective norms		DV3: Perceived behavioral control
	β	p	β	p	β	p
Model 1 (without covariates)
Multiple-channel information seeking	.24	^*^*	.12		.12
Model 2 (with covariates)
Gender	− .12		− .11		.07
Age	− .09		− .11		− .09
Multiple-channel information seeking	.27	^*^*^*	.14	^*	.09

Note: ^*p < .05, ^**p < .01, ^***p < .001.

H3 proposed that the TPB cognitive antecedents would all positively predict the vaccination behavior. The results of the logistic regression analysis revealed that all of the TPB variables, including attitudes (Exp(B) = 1.86, p < .01), subjective norms (Exp(B) = 3.00, p < .001) and perceived behavioral control (Exp(B) = 1.61, p < .05), were positively related to the actual vaccination behavior. After controlling for gender and age, the effects of these variables remained significant: attitudes (Exp(B) = 1.84 p < .01), subjective norms (Exp(B) = 3.52, p < .001) and perceived behavioral control (Exp(B) = 1.88, p < .01). In other words, people who had more favorable attitudes about getting vaccinated, who agreed more that vaccination was the norm, and who perceived that they themselves were more capable of getting vaccinated were more likely to do so during the epidemic. Therefore, H3 was supported (Table 4).

Table 4 Relationships between TPB cognitive antecedents and the actual vaccination behavior.

	Model 1 (without covariates)		Model 2 (with covariates)
	Exp(B)	p	Exp(B)	P
Gender			.39
Age			1.51
Attitudes	1.86	^*^*	1.84	^*^*
Subjective norms	3.00	^*^*^*	3.52	^*^*^*
Perceived behavioral control	1.61	^*	1.88	^*^*

Note: ^*p < .05, ^**p < .01, ^***p < .001.

Figure 1 illustrates the path coefficients from the additional SEM analysis along with the significance level. A good model fit was evidenced, χ² (71) = 109.85, p < .01, CFI = .97, TPI = .97, RMSEA = .05, SRMR = .05. The path coefficients in Figure 1 indicate that, consistent with the regression analyses, surprise positively predicted multiple-channel information seeking (β = .49, p < .001), which in turn positively predicted attitudes (β = .25, p < .01), and subjective norms (β = .16, p < .05), but not perceived behavioral control (β = .10, p = .13).

Figure 1 SEM results for the model examining the relationships between surprise, multiple-channel information seeking, and cognitive antecedents of the vaccination behavior. Note: ^*p < .05, ^**p < .01, ^***p < .001.

Discussion

This study examined how the group most susceptible to the H1N1 flu virus responded to the pandemic through their initial surprise to their purposeful multiple-channel information seeking, and how multiple-channel information seeking helped this population made a conscious decision about whether or not to become vaccinated. The study added to the current research in this field by assessing the role that surprise, a less studied emotional reaction, played in individuals' responses to a real public health crisis. Moreover, this study is one of the few attempts to measure multiple-channel information seeking, a systematic fashion of information seeking, based on actual information seeking behaviors instead of information seeking intent. We demonstrated that this particular approach to information seeking contributed to a variety of cognitive appraisals that helped people decide to become vaccinated against the H1N1 flu. Given the role that surprise played in successfully activating the desirable multiple-channel information seeking, we regard the leveraging of this emotion as a unique opportunity in health prevention and promotion, as well as in information subsidy particularly during times of crisis.

The present study found that surprise was positively related to multiple-channel information seeking. Since surprise has been conceptualized to be an emotional response to unexpected events and schema-stimuli incongruity, it is reasonable that surprise led to behaviors that reduce incongruity, update knowledge schemas and restore confidence (Meyer et al., 1997; Tiedens & Linton, 2001). Multiple-channel information seeking is defined as a non-routine information seeking strategy through which people purposely go beyond their habitual use of information sources by enhancing the number of sources they use to obtain information (Griffin et al., 1999). Seeking information from multiple sources can help on checking the accuracy and credibility of the information. Furthermore, the positive relationship we established between multiple-channel information seeking and attitudes and stronger subjective norms demonstrated its positive effect on schema updating and knowledge generation.

While we found surprise to be a potentially important emotional response that can be leveraged in health prevention and promotion, several issues must be discussed. Since we did not examine the relationship between other frequently used negative emotions and multiple-channel information seeking, it is difficult to determine whether the role surprise plays in directing multiple-channel information seeking is unique compared with other emotions. Valentino, Hutchings, Banks, and Davis (2008) argued that of a variety of emotions, fear and worry are the emotional responses that direct people's attention to new information rather than relying on existing information and predispositions (Valentino et al., 2008). However, the research into the relationship between fear and information seeking has been inconclusive. On the one hand, studies have found that fear positively predicts a preference for protective information (Nabi, 2003) and overall information seeking intentions (Kahlor, 2010; Yang et al., 2011). On the other hand, there are also studies suggesting no relationship between fear and information seeking (Hoffner, Fujioka, Ye, & Ibrahim, 2009), and even a positive relationship between fear and information avoidance (Miles, Voorwinden, Chapman, & Wardle, 2008). Research about worry in different contexts generally suggests that this emotional response increases information seeking intentions (Kahlor, 2007; Yang & Kahlor, 2013) as well as the breadth of actual information seeking behaviors, operationalized as the number of web pages visited in the process of information seeking (Valentino et al., 2008). Therefore, we might conclude that there is a similarity between the effect of surprise and worry on systematic information seeking. However, scholars have argued that a high level of worry is distracting, reducing people's ability to filter out irrelevant information and creating obstacles to encoding new information and updating schemas to generate knowledge (Valentino et al., 2008). This by-product effect of worry on information seeking and processing may result from the negative valence of the emotional response. We maintain that surprise will not lead to these outcomes because this stimuli-schema incongruity induced emotion leads to information seeking behaviors that simply seek to update schemas and generate new knowledge. Therefore, surprise may function more like the psychological construct of curiosity and not necessarily be associated with a negative appraisal of the unexpected event (Loewenstein, 1994).

Given that several months passed between the time people first learned about the H1N1 flu epidemic and the time the vaccine was available to the public (Forbes, 2009; WHO, 2009), it was interesting to find that the initial surprise individuals reported predicted information seeking behaviors throughout the epidemic until they decided whether to become vaccinated. The initial surprise reported may be subject to recall bias due to the cross-sectional design of the current study. However, the results suggest that at least the memory of the initial surprise was influential in prompting subsequent information seeking behaviors over a relatively long period of time. The findings indicate that surprise may not be an emotion that disappears easily. In this case, it seems to have reduced the people's confidence level about their own knowledge and motivated purposeful information seeking. Future studies should use a longitudinal design to confirm whether cognitive appraisals such as uncertainty, perceived information insufficiency and risk perceptions mediate the relationship between initial surprise and information seeking behaviors.

Nevertheless, important implications can be drawn from our findings. Recent analyses of health campaigns using threat appeals have suggested that such campaigns have proved less effective in promoting behavioral compliance (Prevention First, 2008). To overcome this challenge in health promotion and prevention, health care professionals could create knowledge-based campaigns with stimuli that correct well-known misconceptions or present new information incongruent with conventional views on various issues. These types of health messages may induce surprise and then facilitate people to engage in systematic information seeking and processing without scaring the audiences away or eliciting the perception of being manipulated. While such messages may not influence health persuasion outcomes directly, they may increase the desirable information seeking behaviors that lead to learning and cognitive changes.

This study is not free of limitations. Even though we specified the temporal reference point in all questions, such as “when you first learned about the H1N1 flu” when measuring surprise, and “when you decided to get or not get vaccinated” when measuring the TPB variables, and collected the data right after the flu season ended in Spring 2010, the cross-sectional design of the study may result in some ambiguity, making it difficult to claim causal relationships. The responses collected may be also subject to recall bias. As we proposed earlier, future studies should use a longitudinal design and then examine the subsequent effects of surprise on information seeking, cognitive attributions of the unexpected event, and related health perceptions and behaviors. Secondly, the study used a sample of college students. Even though this group was selected because they were identified as the high-risk group due to the unusual profile of the H1N1 flu virus, there could be selection bias associated with this sample. College students have a higher level of educational attainment than about two thirds of the American population (U.S. Census Bureau, 2012). They are often confident about their health status due to their young age and may have more resources than the general public from which they can seek information (Metzger, Flanagin, & Zwarun, 2003). Future studies should examine whether the findings of the current study hold true for non-college student samples as well.

Despite these limitations, this study extended the conceptual work on surprise and empirically tested its relationship with information seeking behaviors in an epidemic. This study found that surprise induced in an unexpected pandemic such as the H1N1 flu epidemic could activate information seeking from multiple channels. This type of information seeking behaviors was found to be positively related to cognitive appraisals that helped people make a decision to get vaccinated against the H1N1 flu.

Disclosure statement

No potential conflict of interest was reported by the authors.