Recreational travel behavior and COVID-19: Insights from expected utility and the theory of planned behavior

Abstract

Understanding what factors play a role in people’s decisions to travel during a pandemic is important to public health officials and to stakeholders in the travel and tourism industry in the United States (US) and worldwide. This study examines factors influencing people’s decisions to cancel/postpone recreational travel within the US amidst the COVID-19 pandemic. Our conceptual framework extends the Expected Utility model, commonly used in economics to model decisions under risk and uncertainty, to incorporate subjective norms and perceived behavioral control from the Theory of Planned Behavior. Our results suggest that risk perceptions, subjective norms, and concerns over transmitting COVID-19 to others play a significant role in the decision to cancel and postpone recreational travel. Results also suggest that perceived behavioral control may be less relevant to travel decisions when traveling involves elevated health risks.

Keywords

COVID-19 Pandemic expected utility theory theory of planned behavior risk perception subjective norms perceived behavioral control

Introduction

In addition to general risks a person traveling for leisure faces including being injured in an accident or developing stomach problems, traveling during a pandemic poses qualitatively different risks of contracting and spreading a potentially lethal disease. This paper examines factors that influence people’s decisions to cancel or postpone recreational travel within the United States (US) amidst the COVID-19 pandemic. Understanding what factors play a role in people’s decisions to travel during a pandemic is important because of the consequences on both personal and public health and on the travel and tourism industry. Travel is a strong force in the emergence and spread of disease. One of the first actions of public health officials worldwide at the outset of the COVID-19 pandemic was to issue stay-at-home orders restricting travel and gatherings to reduce activities associated with community spread of COVID-19, including population movement and close person-to-person contact outside the household (Center for Disease Control and Prevention (CDC), 2020). During the COVID-19 pandemic, travel restrictions and decision by millions of individuals to cancel or postpone travel imposed huge costs to stakeholders in the travel and tourism industry in the US and worldwide. According to the US Travel Association, since the beginning of March 2020 through the end of the year, the pandemic resulted in $492 billion in cumulative losses for the US travel economy (US Travel Association, 2021).

In this paper, we model traveler behavior using a conceptual framework based on Expected Utility Theory (EUT) where individuals weigh the utility derived from traveling with the disutility of being infected by COVID-19 (or infecting others), using subjective risk perceptions. The EUT model is expanded to incorporate subjective norms and perceived behavioral control from the theory of planned behavior (TPB) (Huang et al., 2020). Subjective norms are the perception of other people’s approval of an individual’s particular behavior (Ajzen, 1991), while perceived behavioral control refers to an individual’s perception of the capacity to perform the behavior (Bish et al., 2000). We hypothesize that, in addition to the perceived risk of contracting or transmitting COVID-19 while traveling, the decision to engage in recreational travel depends on subjective norms as well as perceived behavioral control, particularly given public appeals from government to engage in social-distancing and travel restrictions. Our analysis tests the relative strength of these predictors.

Many previous studies on tourism, leisure, and hospitality management rely on the TPB, particularly to analyze consumer behavior (e.g., Joo et al., 2020; Petrescu and Bran, 2020; Ulker-Demirel and Ciftci, 2020; Meng et al., 2020; Huang et al., 2020; Lee et al., 2019). However, within this literature, it has been noted that the TPB fails to fully reflect behavioral change (Matsumori et al., 2019), disregarding some underlying factors addressed through behavioral economic models, especially centered on uncertainty. The Expected Utility Theory (EUT) is one such model commonly used in economics for making decisions under uncertainty. Similarly, EUT does not capture the social psychological factors involved in decision-making, which are better accounted for within the TPB. Therefore, to more accurately explain behavioral change, the current work uses the EUT and the TPB in a complementary fashion. To our knowledge, only two works have combined the EUT and TPB frameworks to explain behavioral change. Matsumori et al. (2019) used an expected utility model to extend the theory of planned behavior so as to derive a decision-theoretic model of behavioral change. Similarly, Borges et al. (2015) used the TPB to extend EUT to explain farmers’ innovation adoption decisions. Not only were these works outside the realm of travel and tourism, they were theoretical in nature, stopping short of empirically evaluating their proposed models. Our study is the first to combine EUT with TPB in an effort to explain behavioral change concerning leisure travel from theoretical and empirical perspectives.

To estimate an empirical model of travel sentiment and behavior based on the conceptual framework, we used survey data collected in two waves, the first one in June–July 2020 and the second one in October 2020. The survey was administered through Qualtrics, Inc. who randomly selected a representative sample of 1454 US residents (541 for wave 1, and 913 for wave 2) from its proprietary panel covering 48 states in the US. The survey gathered information on travel behavioral change, subjective norms, perceived behavioral control, risk perception, health risks, financial risks, and a wide range of socio-demographic variables. One advantage of analyzing the same survey conducted at two different points in time is that it allows us to test if the strength of the predictors of travel decisions changes over time, perhaps as people were getting more informed, developing coping strategies, or experiencing COVID-19 fatigue.

Our results suggest that risk perceptions and subjective norms have played a significant role in the decision of US residents to cancel or postpone recreation travel due to the COVID-19 threat. The effect of risk perceptions on the decision to cancel recreational travel is statistically significant and has decreased over time (i.e., from wave 1 to wave 2). The decline was not found to be statistically significant; however, perhaps because of the short period between waves. Unsurprisingly, health risks play a greater role in the decision to cancel recreational trips than financial risks. Concern about infecting others matters greatly: all individuals who canceled leisure trips in our survey indicated that they were concerned about transmitting COVID-19 at their travel destination or upon their return. On the other hand, perceived behavioral control was not a significant factor for canceling/postponing recreational travel in our study.

Our contribution to the literature is threefold. First, we evaluate consumer behavior in an industry that is especially susceptible to risk, namely tourism, by combining two theoretical frameworks, EUT and TPB, commonly used (although separately) to explain decisions under risk. Previous papers have combined the health belief model (HBM) and TPB (Bae and Chang 2020; Huang et al., 2020).¹ Instead, our model extends the expected utility model, and is the first of its kind in explaining tourist behavior. We note that the proposed model is not limited to health risks and tourist behavior and that it could be applied to a wide range of risky situations that affect behavior. Second, we extend the concept of subjective norms in TPB to account not only for others’ approval of one’s decision to travel but also to account for the externality imposed on others by one’s behavior. That is, our model allows for individuals caring about the possibility of infecting others. Third, we account for a potential temporal change in leisure tourist behavior in the face of evolving uncertainty by using data collected at different time/waves.

The remainder of the paper is organized as follows: the Background section provides information and a brief literature review in regard to COVID-19 and its impact on tourism. The Conceptual Modeling section extends the expected utility model to account for subjective norms and perceived behavioral control. The Empirical Modeling section introduces the econometric model (binary logistic regression) used to examine and test travel behavior hypotheses based on the conceptual model. The Data section provides information about data collection, variable construction, and summary statistics. The Results section provides the main results. The Discussion and Conclusion section provides policy implications and our final thoughts.

Background

As of 2017, the US travel and tourism sector provided 7.8 million jobs and $1.62 trillion in total travel and tourism output, or 2.8% of the country’s GDP (NTTO, 2018). Travel disruptions caused by natural disasters, terrorism, and health and safety crises profoundly impact tourism (Ghaderi and Henderson, 2013; Samitas et al., 2018). To curb the spread of COVID-19, lockdowns (in the form of stay-at-home orders) and travel restrictions were implemented worldwide and in about 42 states in the United States since mid-March 2020 (CDC, 2020). In most if not all US states and/or cities within states during 2020, restrictions to various degrees were placed on eating and drinking establishments, sport and entertainment venues, and other hospitality and tourism-based businesses. These measures combined with people’s reluctance to travel, stay in hotels, or go out to eat or drink had enormous negative economic impacts on the hospitality and tourism industry. The cumulative loss in the US travel economy² since the beginning of March 2020 through the end of the year amounted to $492 billion, resulting in about $64 billion loss in tax revenue at the federal, state, and local levels (US Travel Association, 2021).

The study of health risk perceptions and behavior of individuals regarding travel is important from an economic standpoint, and also for public health reasons as travel poses a threat for the spread of infectious disease (Jonas et al., 2011; Aro et al., 2009; Lopez-Velez and Bayas, 2007). Regarding COVID-19, Gössling et al. (2020) state that there is a need to “understand the behavioral demand responses of tourists in the short- and longer-term.” There is also a need to better understand how different types of travelers (e.g., business vs. leisure travelers) perceive and assume health risks (Aro et al., 2009). Several studies have assessed the impact of COVID-19 on the tourism industry (Kaushal and Srivastava, 2020; Gössling et al., 2020; Hoque et al., 2020; Higgins-Desbiolles, 2020; Lee and Chen 2020; Boto-García and Leoni, 2021; Watson and Deller, 2021). Kaushal and Srivastava (2020) examined the challenges facing tourism and hospitality amid the pandemic and the lessons the tourism industry can learn from COVID-19 conditions. Gössling et al. (2020) examined whether COVID-19 was an “unknowable risk” and assessed the reported impacts of COVID-19 on global tourism. Bae and Chang (2020) used the health belief model (HBM) and an extended theory of planned behavior (eTPB) to examine the impact of risk perception of COVID-19 on behavioral intention towards “untact” tourism.

Conceptual model

We extend the expected utility theory (EUT) model using the theory of planned behavior (TPB) to conceptualize the decision to engage in recreational travel during the COVID-19 pandemic. EUT is the standard “workhorse” model in economics to explain decision making under risk and uncertainty (Kahneman and Tversky, 1979; Schoemaker, 1982; Starmer, 2000; Harrison and Rutström, 2009). Expected utility is measured by weighting the utility of alternative actions with their probabilities. When the outcomes of individual decisions are uncertain, EUT says that individuals will choose the decision that maximizes their expected utility.

Our baseline model is exclusively based on the EUT. In our application, the decision of whether to travel for recreation is made in an uncertain context, where traveling provides utility but also increases the probability of getting sick with COVID-19. We define two states C = {C₁ = contract COVID-19, C₂ = do not contract COVID-19} and a set of two actions T = {t₁ = yes, travel, t₂ = no, do not travel/postpone}, where the subjective probability of a state is contingent on the decision to travel; $P (C_{j} | t_{k})$ . Ordinarily, an individual who stays at home has a lesser chance of contracting COVID-19. Therefore, we assume that the probability of not contracting COVID-19 conditional on not traveling, $P (C_{2} | t_{2})$ , is greater than the probability of not contracting COVID-19 given traveling, $P (C_{2} | t_{1})$ , which in turn is greater than the probability of contracting COVID-19 given traveling, $P (C_{1} | t_{1})$ , which is greater than the probability of contracting COVID-19 given not traveling, $P (C_{1} | t_{2})$ . That is, $P (C_{2} | t_{2}) > P (C_{2} | t_{1}) > P (C_{1} | t_{1}) > P (C_{1} | t_{2})$ and $0 < P (C_{j} | t_{k}) < 1$ . This assumption is based on the premise that COVID-19 spread through the air and individuals are prone to contracting it by being close to someone who has it, and traveling might increase the chances of being around someone with COVID-19. Therefore, an individual who chooses to stay-at-home (not traveling) is less likely to contract COVID-19 compared to an individual who chooses to travel. Still, we assume that the chances of contracting COVID-19 while traveling are low, although they are higher than the probability of contracting COVID-19 while not traveling.

Total utility is assumed to be additive and to depend on the utility of state j, $U (C_{j})$ , and the utility derived from travel, $U (t_{k})$ , so that $U (C_{j}, t_{k}) = U (C_{j}) + U (t_{k})$ . We assume that the utility derived by an individual in state C₁ “contracting COVID-19” is less than zero $(U (C_{1}) < 0)$ , because whether an individual choose to travel or not, there is disutility from being infected with COVID-19, which may lead to sickness and being hospitalized, with even a possibility of death. The utility that an individual gets from state “not contracting COVID-19” is assumed to be greater than zero , regardless of their action to travel or not. Likewise, we assume the utility an individual derives from traveling is greater than zero $(U (t_{1}) > 0)$ , indicating that travelers naturally enjoy recreation and leisure travel. Furthermore, we assume the utility of an individual not traveling can be either zero or lesser than zero $(U (t_{2}) \leq 0)$ given the individuals’ preference for traveling.

The expected utility can then be written as

E [U | t_{k}] = \sum_{j = 1}^{2} P (C_{j} | t_{k}) U (C_{j}, t_{k})

(1)

where action k = 1: travel, 2: do not travel/postpone, and state j = 1: contract COVID-19, 2: do not contract COVID-19. That is

E [U | t_{1}] = \sum_{j = 1}^{2} P (C_{j} | t_{1}) U (C_{j}, t_{1}) = P (C_{1} | t_{1}) U (C_{1}) + P (C_{2} | t_{1}) U (C_{2}) + U (t_{1})

(2)

and

E [U | t_{2}] = \sum_{j = 1}^{2} P (C_{j} | t_{2}) U (C_{j}, t_{2}) = P (C_{1} | t_{2}) U (C_{1}) + P (C_{2} | t_{2}) U (C_{2}) + U (t_{2})

(3)

When deciding whether to go ahead with travel or cancelling/postponing a trip, the individual will compare equations (2) and (3) and cancel/postpone travel when

E [U | t_{1}] < E [U | t_{2}]

. We note, however, that the signs and magnitudes of

E [U | t_{1}]

and

E [U | t_{2}]

in equations (2) and (3) are ambiguous, as they depend on the impacts that getting sick or not has on the general utility, on the subjective probability of contracting COVID-19 while traveling, and on the utility derived from travel.

Extended expected utility model

The expected utility model in equations (1–3) is a stylized model that highlights how rational individuals would choose between alternative options in a risky situation. In practice EUT can make faulty predictions when, for example, decisions are influenced by the opinions or the welfare of relevant other people in one’s life such as family and friends. We hypothesize that, in addition to the risk perception of contracting COVID-19 while traveling, the decision to engage in recreational travel depends on subjective norms as well as perceived behavioral control. TPB incorporates attitudes, subjective norms, and perceived behavioral control to evaluate intentions to perform a behavior (Ajzen, 1991; Bish et al., 2000). Thus, we extend the EUT model by adding two additional variables: subjective norms $(U_{o t h e r s} (t_{k}))$ , and perceived behavioral control $(P B C)$ from TPB, where $U (t_{1})$ is a function of PBC; that is, the utility an individual derives from travel is also dependent on their PBC, $U (t_{1} | P B C)$ . With this extension, equations (2) and (3) become

E [U | t_{1}] = \sum_{j = 1}^{2} P (C_{j} | t_{1}) U (C_{j}) + U (t_{1} | P B C) + U_{o t h e r s} (t_{1})

(4)

E [U | t_{2}] = \sum_{j = 1}^{2} P (C_{j} | t_{2}) U (C_{j}) + U (t_{2}) + U_{o t h e r s} (t_{2})

(5)

In equations (4) and (5), the subjective norms term can be positive or negative depending on whether the approval/disapproval of others reinforces or contradicts the individual’s decision to travel or not.

Moreover, the effect that the “utility” of others has on an individual’s decision to travel, $U_{o t h e r s} (t_{k})$ , could be further decomposed into two-parts. The first component is the approval/disapproval of friends and family of the travel behavior itself $U_{f} (t_{k})$ —this is the part accounted for by TPB. The second component (and a contribution of this paper) reflects the negative externality that traveling might have on others in terms of infecting them and increasing community transmission $U_{i n f e c t o t h e r s} (t_{k})$

E [U | t_{1}] = \sum_{j = 1}^{2} P (C_{j} | t_{1}) U (C_{j}) + U (t_{1} | P B C) + U_{f} (t_{1}) + U_{i n f e c t o t h e r s} (t_{1})

(6)

E [U | t_{2}] = \sum_{j = 1}^{2} P (C_{j} | t_{2}) U (C_{j}) + U (t_{2}) + U_{f} (t_{2}) + U_{i n f e c t o t h e r s} (t_{2})

(7)

Empirical model

For the empirical application, we assume, as it is common in the literature, that a binary decision (to travel in our case), can be modeled by a sigmoid function, for example, a logistic function (Harrison and Rutström, 2009; Chakravarty and Roy, 2009). Thus, we estimated a binary logistic regression model

T_{i r d} = \frac{e x p (Z_{i r d})}{1 + e x p (Z_{i r d})} + ε_{i r d}

(8)

where

T_{i r d}

is the probability of canceling or postponing travel for individual i in region r on survey date d. 1 is assigned to individuals who canceled or postponed their trip and 0 otherwise.

Z_{i r d}

is a linear combination of potential determinants of travel.

We estimate two different models. In the baseline model, $Z_{i r d} = α + γ R P_{i r d} + θ X_{i r d} + η_{r} + η_{d y}$ , where $R P_{i r d}$ measures individual risk perception; $X_{i r d}$ is a vector of socio-economic variables (age, gender, education, income, race, marital status, having children, and employment status); $η_{r}$ denotes region fixed effects, $η_{d y}$ are day-of-week dummies, and $ε_{i r d}$ is the error term, distributed by the standard logistic distribution. In the more comprehensive model, $Z_{i r d} = α + γ R P_{i r d} + δ S N_{i r d} + υ P B C_{i r d} + θ X_{i r d} + η_{r} + η_{d y}$ , where, in addition, $S N_{i r d}$ is the subjective norm (i.e., the utility of relevant others) and $P B C_{i r d}$ denotes perceived behavioral control. The relationship between corresponding variables from the conceptual and empirical models is explained within Appendix Table A1.

Data

We collected primary data through an online survey administered through Qualtrics, Inc. which was designed to gather information on the impact of COVID-19 on travel sentiment and behavior. The data collection was drawn randomly by Qualtrics, Inc. for a representative sample of the US population covering 48 states. The final sample covers all US states except North Dakota and New Hampshire. Data collection was in two waves: wave 1, from June 23 to 1 July 2020 (about 3 months after the first state in the US declared a mandatory stay-at-home order), with 541 respondents; and wave 2, from October 1 to 15 October 2020, with 913 respondents. Total respondents are thus 1454 for both waves. Data collection at different time periods during the pandemic allows us to examine travelers' behavioral changes as the severity of the pandemic measured by the total numbers of confirmed cases varies over space and time and as people developed coping strategies over time. It took respondents an average time of approximately 16 minutes to complete the survey.

The description of the variables and the summary statistics are displayed in Table 1. To measure behavioral change, one question on the survey asked: “For 2020, did you cancel or postpone any recreation or leisure overnight trips throughout the US after learning about the COVID-19 threat?” with answers “Yes-cancel,” “Yes-Postpone,” “No.” 66.44% of the respondents indicated that no, they had not canceled/postponed recreation trips, while 33.56% indicated they had. Figure 1 shows the histogram of the number of canceled trips, indicating that the number of intended recreation trips canceled or postponed in 2020 with highest frequency is 1 trip with 43.54%, while the average number of canceled trips is 2.75.

Table 1.

Definition of variables and descriptive statistics.

Variable	Type	Description	Frequency (%)
			Combined Waves (N=1454)	Wave 1 (N=514)	Wave 2 (N=913)
Travel decision	Categorical	For 2020, did you cancel or postpone any recreation or leisure overnight trips throughout the U.S. after learning about COVID-19?
		Yes-cancel/Yes-postpone	488 (33.56)	194 (35.86)	294 (32.20)
		No	966 (66.44)	347 (64.14)	619 (67.80)
			Mean (std. dev)
Number of canceled trips	Continuous	Number of intended recreation or leisure trips canceled or postponed in 2020	2.75 (2.99)	2.64 (2.83)	2.82 (3.09)
Risk perception (rp)	Categorical	What is the probability that traveling within the U.S. in the next 6 months will lead you to?1=not probable, 2=somewhat probable, 3=neutral, 4=somewhat probable, and 5=probable
		rp1 - Be around others with COVID-19	3.056 (1.43)	3.19 (1.42)	2.979 (1.430)
		rp2 - contract COVID-19	2.979 (1.38)	3.05 (1.38)	2.94 (1.38)
		rp3 - Be hospitalized due to COVID-19	2.72 (1.37)	2.76 (1.35)	2.697 (1.389)
Subjective norms (sn)	Categorical	Please continue to indicate your level of agreement with the following statements.1=strongly disagree, 2=Disagree, 3=neutral, 4=Agree, and 5=Strongly agree			3.02 (1.25)
		sn1 - most people who are important to me think I should travel within the U.S. in the near future	2.95 (1.28)	2.82 (1.34)
Variable	Type	Description	Mean (std. dev)
			Combined Waves (N=1454)	Wave 1 (N=514)	Wave 2 (N=913)
		sn2 - the people in my life whose opinions I value would approve of me traveling within the U.S. in the near future	3.09 (1.25)	2.989 (1.28)	3.15 (1.23)
		sn3 - most people who are important to me would travel within the U.S. in the near future	3.11 (1.27)	3.03 (1.28)	3.16 (1.27)
Perceived behavioral control (pbc)	Categorical	Please indicate your level of agreement with the following statements.1=strongly disagree, 2=Disagree, 3=neutral, 4=Agree, and 5=Strongly agree
		pbc1 - it is easy for me to travel within the U.S. in the near future	3.25 (1.24)	3.16 (1.28)	3.31 (1.21)
		pbc2 - whether or not I travel within the U.S. in the near future is completely up to me	3.82 (1.09)	3.85 (1.09)	3.80 (1.09)
		pbc3 - if I wanted to, I could travel throughout the U.S. in the near future	3.68 (1.15)	3.72 (1.16)	3.66 (1.15)
		pbc4 - I have complete control over traveling throughout the U.S. in the near future	3.61 (1.17)	3.61 (1.18)	3.62 (1.17)
		pbc5 - it is possible for me to travel throughout the U.S. in the near future	3.62 (1.16)	3.64 (1.17)	3.61 (1.16)
Transmit to others	Binary	1=if concern of inadvertently transmitting COVID-19 to people at the destination or to relatives and friends upon return	0.165 (0.37)	0.181 (0.385)	0.16 (0.363)
Variable	Type	Description	Mean (std. dev)
			Combined Waves (N=1454)	Wave 1 (N=514)	Wave 2 (N=913)
Health risk (hr)	Categorical	In your opinion, how serious do you think the health risks of COVID-19 are to you.0=Not at all serious/Slightly serious, 1=Moderately serious/very serious/extremely serious	0.72 (0.45)	0.723 (0.448)	0.72 (0.45)
Financial risk (fr)	Categorical	In your opinion, how serious do you think the financial risks of COVID-19 are to you.0=Not at all serious/Slightly serious, 1=Moderately serious/very serious/extremely serious	0.69 (0.46)	0.684 (0.465)	0.7 (0.46)
Variable	Type	Description	Frequency (%)
Age	Categorical	18–24	260 (17.88)	93 (17.19)	167 (18.29)
		25–29	202 (13.89)	75 (13.86)	127 (13.91)
		30–34	167 (11.49)	46 (8.50)	121 (13.25)
		35–39	165 (11.35)	64 (11.83)	101 (11.06)
		40–44	139 (9.56)	56 (10.35)	83 (9.09)
		45–49	108 (7.43)	37 (6.84)	71 (7.78)
		50–54	98 (6.74)	36 (6.65)	62 (6.79)
		55–59	75 (5.16)	32 (5.91)	43 (4.71)
		60–64	73 (5.02)	31 (5.73)	42 (4.60)
		65–69	89 (6.12)	39 (7.21)	50 (5.48)
		70 and above	78 (5.36)	32 (5.91)	46 (5.04)
Income	Categorical	Under $24,999	361 (24.83)	140 (25.88)	221 (24.21)
		$25,000-$34,999	246 (16.92)	72 (13.31)	174 (19.06)
		$35,000-$49,999	165 (11.35)	71 (13.12)	94 (10.30)
		50,000–74,999	280 (19.26)	91 (16.82)	189 (20.70)
		$75,000-$99,999	173 (11.90)	67 (12.38)	106 (11.61)
		$100,000-$149,999	113 (7.77)	44 (8.13)	69 (7.56)
		150,000 and above	116 (7.98)	56 (10.35)	60 (6.57)
Education	Categorical	Grade school	29 (1.99)	9 (1.66)	20 (2.19)
Variable	Type	Description	Frequency (%)
			Combined Waves (N=1454)	Wave 1 (N=514)	Wave 2 (N=913)
		High school	447 (30.74)	164 (30.31)	283 (31.00)
		Some degree	368 (25.31)	127 (23.48)	241 (26.40)
		Associated degree (2years)	219 (15.06)	92 (17.01)	127 (13.91)
		Bachelor’s degree (4years)	210 (14.44)	74 (13.68)	136 (14.90)
		Graduate degree (post degree/Masters)	181 (12.45)	75 (13.86)	106 (11.61)
Marital status	Categorical	Married	543 (37.35)	222 (41.04)	321 (35.16)
		Single	542 (37.28)	186 (34.38)	356 (38.99)
		Divorced	139 (9.56)	57 (10.54)	82 (8.98)
		Living as married or partnered	151 (10.39)	53 (9.80)	98 (10.73)
		Widowed	56 (3.85)	16 (2.96)	40 (3.38)
		Separated	23 (1.58)	7 (1.29)	16 (1.75)
Gender	Categorical	Female	781 (53.71)	293 (54.16)	488 (53.45)
		Male	643 (44.22)	233 (43.07)	410 (44.91)
		Others	30 (2.06)	15 (2.77)	15 (1.64)
Race	Categorical	White	946 (65.06)	397 (73.38)	549 (60.13)
		Black	292 (20.08)	69 (12.75)	223 (24.42)
		Hispanic	72 (4.95)	22 (4.07)	50 (5.48)
		Two or more race	54 (3.71)	21 (3.88)	33 (3.61)
		Asian	48 (3.30)	15 (2.77)	33 (3.61)
		Native Hawaiian	9 (0.62)	3 (0.55)	6 (0.66)
		Others	33 (2.27)	14 (2.59)	19 (2.08)
Employment status	Categorical	Employed full-time	614 (42.23)	222 (41.04)	392 (42.94)
		Not employed	298 (20.50)	117 (21.63)	181 (19.82)
		Retired	189 (13.00)	83 (15.34)	106 (70.21)
		Employed part-time	216 (14.89)	73 (13.49)	143 (15.66)
		Disable	116 (7.98)	36 (6.65)	80 (8.76)
		Furlong	21 (1.44)	10 (1.85)	11 (1.20)
With children	Categorical	Do you have children under 18 living at home?
		No	892 (61.35)	345 (63.77)	547 (59.91)
		Yes	562 (38.65)	196 (36.23)	366 (40.09)

Figure 1.

Histogram of the number of intended recreation or leisure trips cancelled or postponed in 2020.

In order to capture risk perceptions, the survey asked: “What is the probability that traveling within the U.S. in the next 6 months will lead you to: 1. Be around others with COVID-19, 2. Contract COVID-19, 3. Be hospitalized with COVID-19?”, with responses on the scale of not probable, somewhat improbable, neutral, somewhat probable, and very probable). In addition, the survey had two questions that allow us to evaluate the relative importance of health and financial risks in the decision of whether to cancel or not travel: Health Risk (“in your opinion, how serious do you think the health risks of COVID-19 are to you?”), Financial Risk (“in your opinion, how serious do you think the financial risks of COVID-19 are to you?”), with responses on a 5-point Likert scale: “not at all serious, slightly serious, moderately serious, very serious, extremely serious” that were recoded into two categories of “not serious” and “serious.”

Subjective norms were assessed based on the level of agreement in a 5-point Likert scale with the following statements: 1. “Most people who are important to me think I should travel within the U.S. in the near future,” 2. “The people in my life whose opinions I value would approve of me traveling within the US in the near future.” 3. “Most people who are important to me would travel within the US in the near future.” The concern of disease transmission, Transmitting to others, was measured with the yes/no answer to the following question: “In your decision to cancel or postpone your US overnight recreation or leisure trips, which factors played a role?” with “concern of inadvertently transmitting COVID-19 to people at the destination or to relatives and friends upon my return” being an option.

Perceived behavioral control was assessed based on the level of agreement in a 5-point Likert scale with the following statements: 1. “It is easy for me to travel within the US in the near future.” 2. “Whether or not I travel within the US in the near future is completely up to me.” 3. “If I wanted to, I could travel throughout the US in the near future.” 4. “I have complete control over traveling throughout the US in the near future.” 5. “It is possible for me to travel throughout the US in the near future.”

As it is apparent from the description of the key variables above, for each of the constructs of interest, there are multiple questions that we could use. The answers to those questions are correlated with each other, and Table 2 displays the correlation matrix. One way to tackle the correlation while retaining as much information as possible is by converting the data into an index. The most common statistical method to do the decorrelation and reduce the dimensionality of the data is principal component analysis (PCA) (Jolliffe and Cadima, 2016). Therefore, we used PCA for the construct of risk perception, subjective norms, and perceived behavioral control. Values of the Kaiser-Meyer-Olkin (KMO) test (which measures how well suited the data is for factor analysis) provides values of sampling adequacy of 0.67, 0.75, and 0.85, for risk perceptions, subjective norms, and perceived behavioral control, respectively. As a rule of thumb, values above 0.6 indicate the sampling is adequate for PCA (Dodge, 2008).

Table 2.

Pairwise correlations of the key variables of interest.

Variables	rp1	rp2	rp3	sn1	sn2	sn3	pbc1	pbc2	pbc3	pbc4	pbc5	hr	fr
rp1	1.00
rp2	0.61	1.00
rp3	0.51	0.77	1.00
sn1	−0.05	−0.10	−0.06	1.00
sn2	−0.09	−0.15	−0.10	0.78	1.00
sn3	−0.06	−0.14	−0.12	0.74	0.78	1.00
pbc1	−0.05	−0.14	−0.11	0.55	0.52	0.50	1.00
pbc2	−0.01	−0.07	−0.09	0.13	0.15	0.15	0.35	1.00
pbc3	0.01	−0.06	−0.07	0.28	0.29	0.30	0.49	0.54	1.00
pbc4	−0.03	−0.11	−0.09	0.25	0.27	0.29	0.43	0.56	0.65	1.00
pbc5	0.02	−0.07	−0.10	0.35	0.36	0.38	0.54	0.47	0.66	0.59	1.00
Hr	0.18	0.30	0.33	−0.18	−0.21	−0.19	−0.22	−0.01	−0.08	−0.02	−0.10	1.00
Fr	0.14	0.21	0.21	−0.06	−0.09	−0.07	−0.10	−0.02	−0.04	−0.04	−0.05	0.54	1.00

This table display the correlation matrix among the key variables of interest risk perception (rp1 – rp3), subjective norms (sn1 – sn3), perceived behavioral control (pbc1 – pbc5), health risk (hr), and financial risk (fr). This shows that rp1 to rp3 are highly correlated, sn1 to sn3 are highly correlated, and pbc1 to pbc5 are highly correlated. For definition of the variables, please see Table 1.

Other questions on the survey gathered socio-economic information (such as age, marital status, or employment). Our sample is slightly younger than the US population (25.38% of respondents are between 25 and 34 years of age, compared to a corresponding 13.9% from the US Census, 2019). 65.06% are white; 57.12% are either partially or fully employed, and 53.17% are female. This compares to corresponding US Census figures of 72%, 60.2% and 49.2%, respectively (United States Census, 2019).

Results

Table 3 presents the results from the logistic regression displayed for the EUT baseline model (equations (2) and (3)) in column 1, and the extended model (equations (4) and (5)) in column 2. The regression coefficients in Table 3 represent the marginal effect of the variables of interest.

Table 3.

Logistic regression result of the decision to cancel or postpone recreation overnight travels within the U.S.

Dependent variable: Travel decision	Baseline model	Risk types baseline model	Extended model	Risk types extended model
Variables	(1)	(2)	(3)	(4)
Risk perception	0.0567***		0.0598***
	(0.0171)		(0.0175)
Wave dummy	−0.0660*	−0.0731**	−0.0672**	−0.0743**
	(0.0337)	(0.0343)	(0.0342)	(0.0341)
Risk perception x wave dummy	−0.0321		−0.0325
	(0.0231)		(0.0232)
Health risk		0.0772**		0.0834**
		(0.0347)		(0.0351)
Financial risk		−0.0159		−0.0138
		(0.0224)		(0.0223)
Subjective norms			0.0140***	0.0110**
			(0.00441)	(0.00537)
Perceived behavioral control			0.00140	0.000490
			(0.00733)	(0.00736)
Demographic variables	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes
Day-of-week	Yes	Yes	Yes	Yes
Observations	1454	1454	1454	1454
Log likelihood	−815.51116	−822.64279	−814.19051	−821.8631
Pseudo-R-Squared	0.1210	0.1133	0.1224	0.1142
AIC	1637.022	1651.286	1634.381	1649.726
BIC	1652.869	1667.132	1650.227	1665.572

*** p<0.01, ** p<0.05, * p<0.1, Standard errors in parentheses.

The logistic regression outputs in the table are the marginal effect values.

Standard errors are clustered at the region level. All the models control for covariates such as demographic variables (age, income, education, marital status, gender, employment status, children, and race). Also, dummies for region and day-of-week are included, which controls for the unobserved differences across regions and day of the week. Risk perception, subjective norms, and perceived behavioral control variables used in the analysis are the single indices calculated using principal component analysis of the group of questions that provide each variable: rp1-rp3 for risk perceptions, sn1-sn3 for subjective norms, and pbc1-pbc5 for perceived behavioral control. For definition of the variables, please see Table 1.

Equations (6) and (7) were not included in the regression because the variable “Transmitting to others” is perfectly correlated with the outcome variable (travel decision). Table 4 showing the cross-tabulation between these two variables indicates that everyone who canceled or postponed their travel was concerned about transmitting COVID-19.

Table 4.

Cross-tabulation of transmits concern and travel decision.

concern of inadvertently transmitting COVID-19 to people	Indicate whether the respondent cancel or postpone any recreation or leisure trip
concern of inadvertently transmitting COVID-19 to people	0=No	1=Yes	Total
No	966	248	1214
Yes	0	240	240
Total	966	488	1454

For the baseline model (column 1 Table 3), the estimate for risk perception is statistically significant at a 1% level. This indicates that a 1-point increase in respondents' risk perception index leads to a 5.67% increase in the probability to cancel/postpone overnight trips within the US. The wave dummy exhibits a negative sign and statistically significant at 10% suggesting that US residents may have experienced some fatigue from COVID-19, and they are less likely to cancel or postpone leisure travel when asked later in the year (October vs. July 2020) despite the country still being hit by the pandemic. Also, the interaction term of risk perception and wave shows a negative coefficient that is not statistically significant, perhaps because the two survey waves are close together.

In column 2, we differentiate between different risk types, namely health and financial risks. Health risk is statistically significant at a 5% level with a coefficient of 0.0772. This is an intuitive result indicating that the probability of canceling or postponing leisure travel is larger for individuals who take the health risk of COVID-19 seriously compared to those who do not. In contrast, the financial risk variable is not statistically significant, suggesting that health risks are a more important factor in people’s decision to cancel/postpone recreation trips amidst COVID-19. As in column 1, the wave dummy variable is statistically significant at 5%. The interaction variable between the risk perception and wave dummy is not reported here for simplicity as it was not statistically significant (see Appendix Table A2, column 1 for full set of results).

Column 3 of Table 3 shows the estimates of the extended EUT model. In the extended model, we further decomposed subjective norms into two variables; the relevant others' perception and the negative externality on others of traveling (based on equations (6) and (7)). As indicated above, however, all those who cancel or postpone recreational travel are concerned about transmitting COVID-19 (Table 4). In column 3, risk perception is statistically significant at a 1% level, and of comparable magnitude to the estimate in column 1. The coefficient estimates on the subjective norms variable is statistically significant at 1%. It indicates that as the level of agreement with important others approving of them to recreationally travel, their probability of canceling/postponing recreation trips increases by 1.40%. Perceived behavioral control (PBC) is not statistically significant suggesting that PBC is not an important reason for canceling or postponing trips in our sample. The Akaike information criterion (AIC) and Bayesian information criterion (BIC) model selection criteria show that the extended model is preferred (Table 3). The interactions between the wave dummy and the TPB variables in the extended model (subjective norms, and perceived behavioral control), reported in Appendix A Table A2 column 2, are not statistically significant.

Finally, column 4 displays the coefficients for risk type. As in column 2, while health risk is found to be statistically significant, financial risk is not.

Discussion and conclusions

This paper shows that risk perceptions and subjective norms are significant factors influencing traveler’s decisions to cancel or postpone recreation trips in the US in both June–July 2020 and October 2020. Further investigation into the nature of the travel risks suggests that it is health risks rather than financial risks perception that changed travelers’ decisions. This is consistent with Jonas et al. (2011) who found that health risk perception ranks high among other types of risk perception (although their study focuses on tourists traveling to low-income countries). Our finding reflecting the importance of COVID-19 health risk perceptions is consistent with more recent studies on travel intention and vacation travel behavior change during the pandemic (Perić et al., 2021; Bratić et al., 2021; Neuburger and Egger 2020; Falahuddin et al., 2021).

In the way of theoretical implications, our results also indicate that an EUT model extended to incorporate TPB constructs is preferred to the baseline EUT model to explain travelers' behavior during the COVID-19 pandemic. EUT is concerned with individual utilities and subjective probabilities while TPB measures behavioral intention using attitudes, subjective norms, and perceived behavioral control. Our results indicate that both subjective probabilities (those reflecting health risks) and subjective norms affect individuals’ decision to travel, while perceived behavioral control does not affect that decision. This last result is consistent with Arikkatt and Mohana (2021) and suggests that perceived behavioral control (PBC) may be irrelevant when travel decisions involve elevated health risk, such as with the COVID-19 pandemic. PBC found to be insignificant may be as a result of the health risks. This might be greater than an individual perception of easiness or difficulty to travel for leisure during COVID-19 pandemic. Despite this potential explanation, Aschwanden et al. (2021) recently found that perceived behavioral control was a significant predictor of individuals’ preventative behaviors during the pandemic.

One of the TPB constructs included in the extended EUT model is subjective norms. A large literature in behavioral studies has studied the role of social norms in decision making (Baron, 1992; Bicchieri, 2005; Bicchieri and Xiao, 2009; Tesar, 2020). Bicchieri (2005), categorized social norms into two groups: empirical expectation (what an individual expects others to do) and norm expectation (what we think other expect us to do). The subjective norms considered in our paper are related to norm expectation. In addition, our study considers another element reflecting concern for others’ welfare that may be considered more altruistic and people willingness to contributes to a public good (external benefit or positive externality): the concern of transmitting COVID-19 to others in the destination or upon returning from travel. This element proves to be extremely important. Further, all surveyed individuals who canceled or postponed recreational travel were concerned about transmitting COVID-19, which is what Li et al. (2020) proposed would occur in their recent work concerning tourists’ behavioral changes in response to risk perceptions of transmitting COVID-19.

COVID-19 has tremendously affected the hospitality and tourism industry and has underscored that understanding of factors that lead to massive cancelation or postponement of recreational travel is important not only from a public health perspective but also from an economic perspective. Our findings suggest that strategies to influence travelers’ subjective norms and health risk perceptions, particularly the concern of infecting others, may be helpful as the recovery from the pandemic is ongoing. One possible strategy would be to enhance destination trust regarding health risks. One example of such actions can be promoting safety measures and precautions of COVID-19 at hotels, restaurants, and tourist attractions (Assaf and Scuderi, 2020). These efforts could take the form of visible postings of standards for social distance, vaccination, and mask requirements on-site at establishments but also on lodging, eateries and bars, and attraction webpages and social media platforms. Such efforts may allay some potential behavioral change and provide visitors with greater peace of mind in deciding to visit.

Application of the proposed model in this study is not limited to the COVID-19 pandemic. Our complementary model uniting EUT and TPB can be tested and applied in additional contexts where tourists are faced with risks and uncertain outcomes in deciding to travel during, for example, other viral disease outbreaks, natural disasters (e.g., floods, hurricanes, wildfires, etc.), times of war and armed conflicts, and terrorism. In each of these contexts, travelers’ potential for behavioral change can be examined, especially as we see each of these contexts currently occurring throughout the globe. Furthermore, subsequent research can examine how potential factors affecting U.S. residents’ travel decisions evolve over time as we recover from the pandemic, and as vaccines becomes widely available across the country. In so doing, additional model predictors may include potential travelers’ vaccine status, perceptions of vaccination, age, response efforts of transportation corporations, availability of health care, established restrictions, number of those infected with COVID-19 within the destination, etc. Others have offered these constructs as viable predictors within their works (see Chua et al., 2020; Parady et al., 2020; Shamshiripour et al., 2020; Zhang and Lu, 2021).

Our study does indicate that risk perceptions are an important factor explaining the decision to cancel travel, but that other predictors (such as concern of transmitting COVID-19) cannot be ignored. By extending the expected utility theory with the theory of planned behavior to explain decisions that involve risk (the decision to travel during a pandemic), our study contributes to the to the growing literature on COVID-19, and more broadly to the behavioral and decision-theoretic literature.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Data availability statement

The data that support the findings of this study are available upon request from the correspondence author.

ORCID iD

Kehinde E Ojo

Notes

Appendix

Appendix A

Table A2.

Logistic regression result of the decision to cancel or postpone recreation overnight travels within the U.S. (Estimate results with the full interaction variables).

Dependent variable: Travel decision	Risk type baseline model	Extended model	Risk type extended model
Variables	(1)	(2)	(3)
Risk perception		0.0584***
		(0.0173)
Subjective norms		0.000205	−0.00466
		(0.0202)	(0.0192)
Perceived behavioral control		0.00683	0.00367
		(0.0132)	(0.0136)
Wave dummy	−0.0692	−0.0662**	−0.0792
	(0.0574)	(0.0331)	(0.0498)
Risk perception x wave dummy		−0.0301
		(0.0227)
Subjective norms x wave dummy		0.0249	0.0279
		(0.0324)	(0.0339)
Perceived behavioral control x wave dummy		−0.0106	−0.00654
		(0.0127)	(0.0126)
Health risk	0.0620**		0.0582***
	(0.0268)		(0.0208)
Financial risk	0.00320		0.00899
	(0.0417)		(0.0453)
Health risk x wave dummy	0.0238		0.0387
	(0.0309)		(0.0469)
Financial risk x wave dummy	−0.0304		−0.0321
	(0.0757)		(0.0791)
Demographic variables	Yes	Yes	Yes
Region	Yes	Yes	Yes
Day-of-week	Yes	Yes	Yes
Observations	1454	1454	1454
Log likelihood	−822.51888	−813.29219	−820.60713
Pseudo-R-Squared	0.1135	0.1234	0.1155
AIC	1651.038	1632.584	1647.214
BIC	1649.383	1648.431	1663.06

*** p<0.01, ** p<0.05, * p<0.1, Standard errors in parentheses.

The logistic regression outputs in the table are the marginal effect values.

Authors biographies

Kehinde Elizabeth Ojo is a doctoral student in the Agricultural and Applied Economics Department at the University of Georgia. Her research interests are in applied economics, environmental economics, and development economics. She currently studies behaviors and attitudes towards travel and recreation, health, and well-being in times of COVID-19.

Susana Ferreira is a Professor in the Agricultural and Applied Economics Department at the University of Georgia. An environmental economist by training, she is interested in how environmental risk perceptions shape individual behavior.

John Salazar is an Associate Professor of Hospitality and Food Industry Management in the Agricultural and Applied Economics Department at University of Georgia. He is also the Coordinator of the Hospitality and Food Industry Management Program. His research focuses on tourism marketing, destination development, and hospitality management.

John Bergstrom is a Professor in the Agricultural and Applied Economics Department at the University of Georgia. His research and teaching programs focus on natural resource economics and policy with applications to private and public lands management, valuation of environmental goods and services, and energy production and consumption.

Kyle Maurice Woosnam is an Associate Professor in the Parks, Recreation and Tourism Management Program within the Warnell School of Forestry and Natural Resources at the University of Georgia. His research centers on sustainable tourism planning and management.

References

Ajzen

(1991) The theory of planned behavior. Organizational Behavior and Human Decision Processes 50(2): 179–211.

Arikkatt

Mohanan

(2021) Social distancing intention during the COVID-19 pandemic: the theory of planned behavior in Thai cultural context. Indian Journal of Health and Wellbeing 12(1): 93–99.

Aro

Vartti

Schreck

, et al. (2009) Willingness to take travel-related health risks—A study among Finnish tourists in Asia during the avian influenza outbreak. International Journal of Behavioral Medicine 16(1): 68.

Aschwanden

Strickhouser

Sesker

, et al. (2021) Preventive behaviors during the COVID-19 pandemic: associations with perceived behavioral control, attitudes, and subjective norm. Frontiers in Public Health 9: 662835.

Assaf

Scuderi

(2020) COVID-19 and the recovery of the tourism industry. Tourism Economics 26(5): 731–733.

Bae

Chang

(2020) The effect of coronavirus disease-19 (COVID-19) risk perception on behavioural intention towards ‘untact’ tourism in South Korea during the first wave of the pandemic (March 2020). Current Issues in Tourism 24(7): 1017–1035.

Baron

(1992) The effect of normative beliefs on anticipated emotions. Journal of Personality and Social Psychology 63(2): 320.

Bish

Sutton

Golombok

(2000) Predicting uptake of a routine cervical smear test: a comparison of the health belief model and the theory of planned behavior. Psychology and Health 15(1): 35–50.

Bicchieri

(2005) The Grammar of Society: The Nature and Dynamics of Social Norms. Cambridge, UK: Cambridge University Press.

10.

Bicchieri

Xiao

(2009) Do the right thing: but only if others do so. Journal of Behavioral Decision Making 22(2): 191–208.

11.

Borges

JAR

Foletto

Xavier

(2015) An interdisciplinary framework to study farmers’ decisions on adoption of innovation: insights from expected utility theory and theory of planned behavior. African Journal of Agricultural Research 10(29): 2814–2825.

12.

Boto-García

Leoni

(2021) Exposure to COVID-19 and travel intentions: evidence from Spain. Tourism Economics. DOI: 10.1177/1354816621996554.

13.

Bratić

Radivojević

Stojiljković

, et al. (2021) Should I stay or should I go? Tourists’ COVID-19 risk perception and vacation behavior shift. Sustainability 13(6): 3573.

14.

Center for Disease Control and Prevention (CDC) (2020) Timing of state and territorial COVID-19 stay-at-home orders and changes in population movement – United States, March 1- May 31, 2020. https://www.cdc.gov/mmwr/volumes/69/wr/mm6935a2.htm (accessed 25 January 2020).

15.

Chakravarty

Roy

(2009) Recursive expected utility and the separation of attitudes towards risk and ambiguity: an experimental study. Theory and Decision 66(3): 199.

16.

Champion

Skinner

(2008) The health belief model. Health Behavior and Health Education: Theory, Research, and Practice 4: 45–65.

17.

Chua

Al-Ansi

Lee

, et al. (2020) Tourists’ outbound travel behavior in the aftermath of the COVID-19: role of corporate social responsibility, response effort, and health prevention. Journal of Sustainable Tourism 29(6): 879–906.

18.

Dodge

(2008) The Concise Encyclopedia of Statistics. Berlin, Germany: Springer Science & Business Media.

19.

Falahuddin

Tergu

Brollo

, et al. (2021) Post COVID-19 pandemic international travel: does risk perception and stress-level affect future travel intention? Jurnal Ilmu Sosial dan Ilmu Politik 24(1): 1–14.

20.

Ghaderi

Henderson

(2013) Japanese tsunami debris and the threat to sustainable tourism in the Hawaiian Islands. Tourism Management Perspectives 8: 98–105.

21.

Gössling

Scott

Hall

(2020) Pandemics, tourism and global change: a rapid assessment of COVID-19. Journal of Sustainable Tourism 29(1): 1–20.

22.

Harrison

Rutström

(2009) Expected utility theory and prospect theory: one wedding and a decent funeral. Experimental Economics 12(2): 133.

23.

Higgins-Desbiolles

(2020) Socialising tourism for social and ecological justice after COVID- 19. Tourism Geographies 22(3): 610–623.

24.

Hoque

Shikha

Hasanat

, et al. (2020) The effect of Coronavirus (COVID-19) in the tourism industry in China. Asian Journal of Multidisciplinary Studies 3(1): 52–58.

25.

Huang

Dai

(2020) Predicting tourists' health risk preventative behaviour and travelling satisfaction in Tibet: combining the theory of planned behaviour and health belief model. Tourism Management Perspectives 33: 100589.

26.

Jolliffe

Cadima

(2016) Principal component analysis: a review and recent developments. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 374: 2065.

27.

Jonas

Mansfeld

Paz

, et al. (2011) Determinants of health risk perception among low-risk-taking tourists traveling to developing countries. Journal of Travel Research 50(1): 87–99.

28.

Jones

Jensen

Scherr

, et al. (2015) The health belief model as an explanatory framework in communication research: exploring parallel, serial, and moderated mediation. Journal of Health Communication 30(6): 566–576.

29.

Joo

Seok

Nam

(2020) The moderating effect of social media use on sustainable rural tourism: a theory of planned behavior model. Sustainability 12(10): 4095.

30.

Kahneman

Tversky

(1979) Prospect theory: an analysis of decision under risk. Econometrica 47(2): 263–292.

31.

Kaushal

Srivastava

(2020) Hospitality and tourism industry amid COVID-19 pandemic: perspectives on challenges and learnings from India. International Journal of Hospitality Management 92: 102707.

32.

Lee

Chen

(2020) The impact of COVID-19 on the travel and leisure industry returns: some international evidence. Tourism Economics. DOI: 10.1177/1354816620971981.

33.

Lee

Gunarathne

(2019) Do green awards and certifications matter? Consumers’ perceptions, green behavioral intentions, and economic implications for the hotel industry: a Sri Lankan perspective. Tourism Economics 25(4): 593–612.

34.

Zhang

Liu

, et al. (2020) Seeing the invisible hand: underlying effects of COVID-19 on tourists’ behavioral patterns. Journal of Destination Marketing & Management 18: 100502.

35.

Lopez‐Velez

Bayas

(2007) Spanish travelers to high‐risk areas in the tropics: airport survey of travel health knowledge, attitudes, and practices in vaccination and malaria prevention. Journal of Travel Medicine 14(5): 297–305.

36.

Matsumori

Iijima

Koike

, et al. (2019) A decision-theoretic model of behavior change. Frontiers in Psychology 10: 1042.

37.

Meng

Chua

Ryu

, et al. (2020) Volunteer tourism (V.T.) traveler behavior: merging norm activation model and theory of planned behavior. Journal of Sustainable Tourism 28(12): 1947–1969.

38.

National Travel and Tourism Office (NTTO) (2018) United States Travel and Tourism Industry. Retrieved: https://travel.trade.gov/outreachpages/download_data_table/Fast_Facts_2017.pdf (accessed 8 January 2020).

39.

Neuburger

Egger

(2020) Travel risk perception and travel behaviour during the COVID-19 pandemic 2020: a case study of the DACH region. Current Issues in Tourism 24(7): 1003–1016.

40.

Parady

Taniguchi

Takami

(2020) Travel behavior changes during the COVID-19 pandemic in Japan: analyzing the effects of risk perception and social influence on going-out self-restriction. Transportation Research Interdisciplinary Perspectives 7: 100181.

41.

Perić

Dramićanin

Conić

(2021) The impact of Serbian tourists’ risk perception on their travel intentions during the COVID-19 pandemic. European Journal of Tourism Research 27: 2705.

42.

Petrescu

Bran

(2020) The use of smartphone for the search of touristic information. an application of the theory of planned behavior. Economic Computation & Economic Cybernetics Studies & Research 54(1): 125–140.

43.

Samitas

Asteriou

Polyzos

, et al. (2018) Terrorist incidents and tourism demand: evidence from Greece. Tourism Management Perspectives 25: 23–28.

44.

Schoemaker

(1982) The expected utility model: its variants, purposes, evidence and limitations. Journal of Economic Literature 20(2): 529–563.

45.

Shamshiripour

Rahimi

Shabanpour

, et al. (2020) How is COVID-19 reshaping activity-travel behavior? Evidence from a comprehensive survey in Chicago. Transportation Research Interdisciplinary Perspectives 7: 100216.

46.

Shim

You

(2015) Cognitive and affective risk perceptions toward food safety outbreaks: mediating the relation between news use and food consumption intention. Asian Journal of Communication 25(1): 48–64.

47.

Starmer

(2000) Developments in non-expected utility theory: the hunt for a descriptive theory of choice under risk. Journal of Economic Literature 38(2): 332–382.

48.

Tesar

(2020) How do social norms and expectations about others influence individual behavior? Foundations of Science 25(1): 135–150.

49.

Ulker-Demirel

Ciftci

(2020) A systematic literature review of the theory of planned behavior in tourism, leisure and hospitality management research. Journal of Hospitality and Tourism Management 43: 209–219.

50.

United States Travel Association (2021, January). Weekly travel; data report. https://www.ustravel.org/toolkit/covid-19-travel-industry-research (accessed 18 June 2021).

51.

United States Census (2019) Comparative demographics estimates. https://data.census.gov/cedsci/table?q=ACSCP1Y2019.CP05&tid=ACSCP1Y2019.CP05&hidePreview=true (accessed 22 June 2021).

52.

Watson

Deller

(2021) Tourism and economic resilience. Tourism Economics. DOI: 10.1177/1354816621990943.

53.

Zhang

(2021) Forecasting hotel room demand amid COVID-19. Tourism Economics. DOI: 10.1177/13548166211035569.