Tourism Economics, 2015, 21 (1), 205–221 doi: 10.5367/te.2014.0443

Why do tourists persist in visiting the same destination? ANTÓNIA CORREIA

CEFAGE, Faculty of Economics, University of Algarve, Campus de Gambelas 80005–328 Faro, Portugal. E-mail: [email protected]. (Corresponding author.) ANDREAS H. ZINS

MODUL University, Vienna, Austria. E-mail: [email protected]. FRANCISCO SILVA

University of Azores, Portugal. E-mail: [email protected]. Capturing and retaining tourists are the main driving forces behind tourism marketing research. Nevertheless, research on how to retain tourists and why they persist in repeating the same destination is not consensual. Following the early work of Ehrenberg and the recency– frequency–monetary value paradigm, this study applies a Poisson distribution model to estimate the past frequency of revisiting Portugal based on information collected from international repeat visitors surveyed at all airports of Portugal in 2012. Results from estimating the model show mixed effects for recency, country of origin and destination region. Recommendation behaviour could not be identified as an explanatory variable for past visitation frequency. Keywords: repeat tourists; loyalty; Poisson models; Portugal

Retaining tourists is one of the main challenges for marketing research given the benefits of repeat visitors to sustained tourism destination development (Oppermann 2000). Among the benefits most often cited are: (a) the consensus that retaining is more cost-effective than capturing new markets; (b) the tendency to increased profits due to the lower susceptibility of repeat tourists to price; and (c) repeat tourists are one of the best referrals mechanism through word of mouth (Reichheld and Sasser, 1990; Shoemaker and Lewis, 1999). The ” and marketing benefits of repeat tourists justify the great effort economic tourism destinations make to retain tourists (Gitelson and Crompton, 1984; Darnell and Johnson, 2001). This effort is also evident in the tourism literature; in fact research about repeat tourists abounds, even though some of these studies present drawbacks that limit the understanding of repeat behaviour patterns. Most of the extant literature assesses behavioural patterns based on intentions.

206

TOURISM ECONOMICS

However, there is disagreement in the literature about this: Van den Putte (1991), Eagly and Chaiken (1993) argued that intentions could be a good predictor of behavioural patterns, whereas Wind and Lerner’s (1979) work proves that intentions are weak if not spurious determinants of behavioural patterns. In fact the definition of behavioural intentions presents slight but critical differences from the definition of attitudes/behaviour. Behavioural intention is the plan an individual makes to perform or not perform a certain behaviour (Fishbein and Ajzen, 1975), whereas behaviour refers to the actions individuals adopt that change their relationship with their environment. If intentions may not contribute to the understanding of behavioural patterns, it is even less likely that intentions may be used as a proxy of loyalty, which is the one of the most common drawbacks of the literature. One of the fundamentals of loyalty programmes is the recency–frequency– monetary value (RFM) (Hughes, 1996). RFM proposes that the most likely loyal consumers are those most recent, frequent and who spend more (Hughes, 1996). The notion of recency crosses the borders of consumption life cycle theory (Deaton, 2005), in which a decreasing likelihood of repurchase is expected over a time period. Hence, the individuals most likely to repurchase are the most recent ones, the recency being the period since a customer’s last purchase. Frequency is the number of purchases made in a certain period; monetary value is the amount of money the customer spends during a certain period. These presuppositions have been highlighted in the tourism literature, even though most are not directly related to loyalty and repeat behaviour assessments under the RFM paradigm (Woodside and MacDonald, 1994). Although there is evidence that frequency of visit may be correlated with loyalty (Oppermann, 1999), both constructs do not mean exactly the same. Jacoby (1971) argues that repeat purchasing behaviour is a necessary but non-sufficient condition for brand loyalty; further, Jacoby and Kyner (1973) prove that repeat purchasing behaviour and brand loyalty are different functional concepts moderated by different dynamics. These, and other research findings not outlined here, follow the same presupposition: there is a behavioural pattern that should be emphasized when assessing loyalty (Dick and Basu, 1994). Under these contextual settings it is evident that the repeat frequency visit entails a new stream of research towards a better understanding of repeating tourists’ behavioural patterns that till now has been scarcely approached in tourism literature, grounded on the RFM paradigm and Ehrenberg (1955) theory. This study explores the effects of tourists’ recency, frequency and spending behaviour on tourists’ revisit behavioural patterns. By transferring scientific contributions from other disciplines, this research, even without any ambition to trace a theoretical framework, is a step towards bringing new approaches to the discussion which should be considered while assessing tourists’ repeat behavioural patterns.

From repeat behavioural patterns to loyalty Observed consumer purchase behaviour is the most important information sources for predicting consumer behaviour (Bellman et al, 2000). Further, consumer behaviour is assumed to be mainly driven by their preferences. Hence, the most common research relies on choice models based on stated preference

Why do tourists persist in repeat visits?

207

theory. Stated preference theory assumes that individuals are able to order their preferences across different alternatives. Nevertheless in some circumstances individuals are not able to state clearly what their preferences will be. Moreover their attitude may differ from what they stated. Another alternative to understanding consumers’ choice is revealed preference. Revealed preference does not ask directly what they prefer; rather their choices are observed indirectly, which is the best way to assess choice determinants. Grounded on quantitatively observed behaviour, Ehrenberg (1955) concluded that patterns of consumer purchases follow a negative binomial distribution. Further developments were presented by Goodhardt et al (1984), who link the phenomenon of repetition to consumption loyalty, with the purchasing frequency following a binomial distribution explaining the repetition of the act. The so-called ‘Dirichlet Model’ of repeat-purchase behaviour is capable of providing estimates of repeat-purchase patterns for a stationary competitive repertoire market, which can do so from just a few inputs. By providing brand-by-brand estimates of expected performance if the market were stationary, it can be used to assess nonstationary brands that have initiated major marketing interventions. Thus, the ‘Dirichlet’ approach can be used to provide a natural benchmark against which the impact of a loyalty programme can be assessed. Ehrenberg’s repeat-buying theory is a descriptive theory based on consumer panel data receiving strong empirical support since the late 1950s. In a consumer panel history the sequence of his or her purchase history is known. Thus buyer behaviour can be predicted from the penetration and the average purchase frequency without taking into account non-buyers. Ehrenberg (1955) states that repeat-buying behaviour can be adequately described by formalizing the purchase incidence process for a single brand. Furthermore, this author proves that purchase incidence follows a Poisson distribution with a certain long-run average. In some circumstances, consumers purchase a brand repeatedly because they are truly fond of the brand. In other cases, consumers make repeat purchases of a brand simply because the brand provides an adequate solution to a problem. Thus, the reason for repeat purchasing is likely to be inertia or true loyalty (Ehrenberg, 1955). Dick and Basu (1994) suggested that inertia or habit may be defined as spurious loyalty. True loyalty refers to repeating purchase incidence when the consumer is fond of a brand (Cunningham, 1967) whereas spurious loyalty, in contrast, represents repeat purchasing without attachment to brand attributes (Day, 1969). However, there is a tendency to assess loyalty through indicators of repeat purchasing sequences or purchase market shares, which do not allow for a distinction between true and spurious loyalty. Although distinguishing typologies of loyalty is beyond the scope of this research it may be assumed that loyalty should be measured by the probability of buying the same brand based on recency, frequency, monetary value and satisfaction. The first three determinants are grounded on the RFM paradigm with the last representing a proxy of brand attachment.

Tourists repeating behavioural patterns (TRBP) Woodside and MacDonald (1994) argue that tourists may or may not return to a familiar destination. Those who return are looking for familiar destinations

208

TOURISM ECONOMICS

whereas the others who decide not to return are trying to avoid the familiarity effect. Tourist taxonomies based on revisit frequency appear in Oppermann (2000), among others. Gitelson and Crompton (1984) classified repeat visitors into three subgroups: infrequent, frequent and very frequent, although the extent of the frequency is not clearly specified. Later on and following the work of Gitelson and Crompton (1984), Oppermann (1999) correlated the frequency of the visit with loyalty, presenting the following typologies: somewhat loyal (infrequent), loyal (regular) or very loyal (annual and biannual). Jang and Feng (2007) were one of the first authors to defend a trichotomous TRBP tourist segmentation: continuous repeater (travellers with consistently high revisit intentions over time); deferred repeater (travellers with low revisit intentions in the short-term but high revisit intentions in the long-term); and continuous switcher (travellers with consistently low revisit intentions over time). Among the three segments, deferred repeaters tend to reinforce visit patterns. Thus, they are also potential switchers who tend to visit more than one destination, showing split loyalties and displaying an increasing tendency to revisit the destination after their initial visit and these intentions may be mixed with loyalty. The authors asserted that brand loyalty is rooted in high involvement with the product category and that inertia results from a lack of involvement. Referring to Ehrenberg theory, Sharp and Sharp (1997) used the Dirichlet model to examine a major loyalty programme in Australia. It is assumed that when consumption is rewarding, repetition is more likely to take place although this should not be assumed as a rule, since in some circumstances tourists may prefer to have other experiences than to repeat the same one, even when they are very pleased with the previous visit (Mazanec and Zins, 1996). Further Cohen and Houston (1972), argue that loyalty to a given tourist destination may not necessarily be explained by the repetition of the experience itself, but instead, by the extent to which a particular set of needs and aims are more or less satisfied. Certain kinds of tourists, with a more allocentric nature, may not repeat a certain holiday, even if they are satisfied with it, Cohen (1972). Conversely, more psychocentric and conservative tourists, who are keen on local routines, to which they become attached, tend to more easily repeat their holiday destination (Cohen, 1972). It is under this controversial and challenging framework that our research arose, aiming to explain why and how tourists persist in visiting the same destination by the means of a Poisson distribution model. The Poisson distribution is a better representation of hazard events than using a negative binomial distribution (Cameron and Trivedi, 2013). The fundamentals of the research design are presented in the following section.

Research design and hypotheses To our knowledge, this approach has not yet been followed by any author in tourism research. To test the model, a database of 5,422 observations, obtained from a survey among international tourists undertaken throughout 2012 in all Portuguese airports, was employed. With the aim of also making some comparisons between nationalities and regions of the country, dummy variables were

Why do tourists persist in repeat visits?

209

created and the combined effect of nationals, regions, satisfaction and number of years visiting the country or the region were introduced in the Poisson distribution model. In the Poisson regression model the number of visits (y) has a Poisson distribution with a conditional mean that depends on tourists’ characteristics according to the structural model: µi = E(yi|Xi) = exp(Xiβ),

(1)

Taking the exponential of Xβ forces the expected count µ to be positive; this is required for the Poisson distribution. The likelihood function for the Poisson regression model is: N N exp(–µ )µ yi i i L(β|y, X) = Π Pr(yi| µi) = Π ———— yi ! , i=1 i=1

(2)

where µ = exp(Xβ). After taking the log, numerical maximization can be used. Since the likelihood function is globally concave, if a maximum is found it will be unique. For the Poisson regression model, the expected value of y for a vector of variables X is: µ = E(y|X) = exp(Xβ).

(3)

The partial derivative of E(y|X) with respect to xk (marginal effect) can be computed using the chain rule: ∂E(y|X) ∂exp(Xβ) ∂Xβ ——— = ———– . —— = exp(Xβ)βk . ∂x k ∂Xβ ∂x k

(4)

The value of the marginal effect depends on the levels of all explanatory variables. The factor or percent variation change in the expected count can be computed from the parameters of the model. Equation (1) can be rewritten as: E(yi|X, xk) = exp(β0).exp(β1x1) … exp(βkxk) … exp(βKxK).

(5)

If xk changes by ∂: E(yi|X, xk + ∂) = exp(β0).exp(β1x1) … exp(βkxk)exp(βk∂) … exp(βKxK),

(6)

the factor change in the expected count for a change of ∂l n xk equals: exp(β0).exp(β1x1) … exp(βkxk)exp(βk∂) … exp(βKxK) E(yi|X, xk + ∂) —————–– = ———————————————————— exp(β0).exp(β1x1) … exp(βkxk) … exp(βKxK) E(yi|X, xk) = exp(βk∂) .

(7)

For a change of ∂ ln xk the expected count increases by a factor of exp(βk∂) holding all other variables constant. For a ∂ unit change in the expected count changes by a factor of exp(βk), holding all other variables constant. Alternatively, the percent variation change in the expected count for a ∂ unit change in xk, holding all other variables constant, can be computed as:

210

TOURISM ECONOMICS

E(yi|X, xk + ∂) – E(yi|X, xk) ——————————— × 100 = [exp(βk × δ) – 1] × 100. E(yi|X, xk)

(8)

The effect of a variable can also be assessed by computing the discrete change in the expected value of y for a change in xk starting in xS and ending at xE: ΔE(y|X)

——— = E(y|X, xk = xE) – E(y|X, xk = xS) . Δxk

(9)

The effect of a binary variable is obtained by letting xk change from 0 (xS) to 1 (xE), that is: ΔE(y|X) = E(y|X, xk = 1) – E(y|X, xk = 0) .

(10)

The magnitude of the discrete change depends on the levels of all explanatory variables in the model. The other explanatory variables are evaluated by their mean values. Variables in the model were selected based on the RFM paradigm and Ehrenberg theory. Thus, the Poisson model explains tourists’ revisiting patterns based on socio-demographic variables, duration of the travel relationship with Portugal, average spending, tripographic variables, expectations, satisfaction with the attributes of the destination and behavioural intentions, in the case of Portugal.

Hypotheses Tourists who visited Portugal in 2012 were asked how many times they had been to Portugal before. This past trip frequency follows a Poisson distribution and is based on the utility that the tourist receives on previous visits. Thus, the utility underlying the destination choice is partially covered by the following hypotheses. Hypothesis 1 (socio-demographic characteristics): the frequency of revisiting Portugal is explained by tourists’ socio-demographic characteristics such as age, income, employment status and nationality. This hypothesis follows the proposal of Niininen and Riley (1998) who demonstrate that demographic variables influence repetitive buying behaviour. In particular these authors found that a more conservative lifestyle and ageing tourists positively influence persistence in having holidays at the same destination. This is also a very traditional hypothesis in demand models that at least helps to validate the socioeconomic characterization of the questionnaire respondents. Hypothesis 2 (tripographic characteristics): the frequency of revisiting Portugal is explained by tourists’ tripographic characteristics of previous visit, such as: length of stay; accommodation type and size of travelling party. Length of stay has been shown to play a role in the way tourists perceive and assess the destination visited (Gokovali et al, 2007; Barros et al, 2010). Thus,

Why do tourists persist in repeat visits?

211

it is likely that repeat visits to the same destination are influenced by the length of stay of previous visits. The type of accommodation, travel companion and average spending on previous visits also affects the likelihood to revisit the destination. Valle et al (2008) show that the intention to return is influenced by these variables. Furthermore, it has been demonstrated that tourists travelling with family are more likely to choose familiar destinations, these being the tourists who might also be willing to pay a higher price (Moutinho, 2000). Hypothesis 3 (expectations, satisfaction and behavioural intentions): the frequency of revisiting Portugal is explained by tourists’ satisfaction, expectations and behavioural intentions. Even though satisfaction may be termed a necessary condition favouring holiday repetition it may not be a sufficient condition itself. Gitelson and Crompton (1984) argue that even if satisfied with their visit, tourists may not, under certain circumstances, be willing to return, even if they are able to recommend the place and its attractions (see also Mazanec and Zins, 1996). On the one hand, less satisfied tourists may repeat their holiday in the same place, due to an inertia factor, motivated by the avoidance of taking on a new decision process, even when facing the risk of disappointment with the result of a new choice being made (Oppermann, 1999, 2000). Further, it has been argued that satisfaction and previous expectations influence future visits to the same destination (Tian-Cole and Crompton, 2003). On the other hand, if the tourist does not reveal a high involvement or high satisfaction with the destination, his or her loyalty is spurious (Dick and Basu, 1994; Zins, 2001) and therefore difficult to be retained for a long period. Furthermore, even if behavioural intentions may not result in effective revisit, extant research shows that tourists with positive intentions are more likely to be loyal to the destination (Baloglu and Erickson, 1998). Hypothesis 4 (recency): the frequency of revisiting Portugal is explained by the recency of their relation with the destination. This hypothesis entails the presuppositions of customer life cycle theory, which states that customers start a relation with a brand that emerges, grows, matures, declines and ends. Yet it is likely that the more recent the relation within the destination and tourists, the higher the revisit frequency (Oppermann, 2000). Yet it could be expected that time affects tourist retention and loyalty (Oppermann, 1999). It was hypothesized that the duration of the relationship with the destination may determine different revisiting patterns, as such five dummies were created to account for different elapse time levels with the first representing tourists who started their relationship with Portugal in the 1950s or 1960s, and the last dummy the most recent ones who started to visit in 2000 or later. Furthermore, the cross-effects of this elapse time by nationalities and by regions were also assessed by multiplying the five dummies with the seven regions and with the three nationals. Hypothesis 5 (monetary value): The frequency of revisiting Portugal differs by average daily spending.

212

TOURISM ECONOMICS

It is noteworthy that loyalty reduces consumer sensitivity to price variations (Krishnamurthi and Papatla, 2003) and thus more frequent tourists are likely to spend more on the same destination than first-time visitors. This hypothesis is the corollary of RFM. Hypothesis 6 (heterogeneity): the frequency of revisiting Portugal differs by destination region and by nationalities. Barros et al (2008) have outlined that tourism choice always entails heterogeneity which should be considered when deciding what statistical models to apply (see also Mazanec, 2000). Later, Correia and Kozak (2012) demonstrated that domestic tourists have different perceptions of the regions in which they stay while in the Algarve, in the south of Portugal. In order to account for the so-called heterogeneity, a set of dummies was created, one for each nationality, divided into three groups: (a) Germany and the UK, the most traditional markets for Portugal; (b) Central Europe; (c) Northern Europe. Since tourists coming from these different countries stay in different regions, dummies for the most typical touristic regions of Portugal were created: Algarve, Lisbon, Azores and Madeira. Moreover, to account for the cross-effect of satisfaction by regions, the overall satisfaction was multiplied by each of the four dummy regions. Furthermore, the nationalities were analysed in each region in which they stayed. This resulted in 3*7 regions, giving 21 dummy variables. On the basis of this set of hypotheses, the expected value of y, past frequency of revizitation, is explained by a set of variables generated by a questionnaire collected at Portuguese airports at the time of tourists’ departure. The variables are illustrated in Table 1. To estimate the Poisson distribution model we use the software Stata 7.

Survey methods The empirical study was carried out by means of a questionnaire applied throughout 2012 in all Portuguese airports. A stratified random sample of passengers travelling on 37 different routes was selected. Overall we collected 30,000 observations from which 20,077 were considered valid. The number of valid cases represents 67% of the whole sample, which according to Dillman (1978) is a very acceptable level. The central aim of this research was to determine passenger flows as well as habits, motivations, satisfaction and economic impacts of tourists travelling to Portugal via these routes. Foreign tourists in Portugal totalled 1.4 million, from which 70% travel to Portugal on these routes. The sample comprises 20,077 observations, which corresponds to a sampling error of 0.6% with a confidence interval of 95%, although as the aim of this research relies on repeat tourists for estimating the model only 5,422 cases were considered. The remaining questionnaires not considered for the present research were discarded because of uncompleted fields, incorrect completion or simply because the tourists are not repeat visitors. Pre-tests and a number of procedures were adopted to ensure the generalizability of the data, meaning that the findings are applicable to a more general population. The general characteristics of this population are the following: international

Nominal Quantitative

Region in Portugal

Average spending

Source: Adapted from Initiative Monitoring survey, UALG, Ana Airports and Portuguese Tourism Board.

Where did you usually spend your holidays in Portugal? How much did you usually spend in Portugal?

Where is your permanent residence?

Ordinal

Satisfaction

Nominal

Ordinal Ordinal

Return Recommendation

Quantitative

Ordinal

Expectations

Number of years visiting Portugal Country of residence

Nominal

Type of accommodation

Ordinal

Nominal Nominal Quantitative

Family average income Employment situation Average stay

Overall satisfaction

How many times have you visited Portugal? How old are you?

Quantitative Nominal

Y Age Family average monthly income Employment situation How many nights do you usually spend in Portugal? What type of accommodation did you usually choose? Before travelling, what was your expectation about Portugal? Do you plan to visit Portugal again? Would you recommend Portugal to friends and relatives? How satisfied are you with the following attributes? Shopping, information available, transportation facilities, closeness to home, gastronomy, climate, relaxing environment, sightseeing and excursions. Overall how satisfied are you with this destination? When was your first trip to Portugal? (year)

Question, description

Table 1. Characterization of the variables. Variable Type of variable

1 – very dissatisfied; 2 –dissatisfied; 3 – satisfied; 4 –very satisfied; 5 –extremely satisfied 1 – since 1970s; 2 – since 1980s; 3 – since 1990s; 4 – since 2000s 1 – Germany and UK; 2 –Central Europe; 3 – Nordic countries 1 – Lisbon; 2 – Algarve; 3 – North; 4 – Azores; 5 – Madeira Numeric

1 – very dissatisfied; 2 –dissatisfied; 3 – satisfied; 4 – very satisfied; 5 – extremely satisfied

1 – hotel; 2 – aparthotel; 3 – rented house; 4 – family/ friends’ house; 5 – own house 1 – very low; 2 – low; 3 – no expectations; 4 – moderate; 5 – high 1 – no; 2 – probably; 3 – certainly 1 – no; 2 – probably; 3 –certainly

Numeric 1 – less than 30 years, young; 2 – 30–64, adults; 3 – more than 64, seniors. 1 – Less than €5,000/month; 2 – €5,000 or more 1 – active; 2 – unemployed; 3 – student; 4 – retired Numeric

Scale

Why do tourists persist in repeat visits? 213

214

TOURISM ECONOMICS

tourists are adults (41.2%), young (30.8%) or seniors (27.6%), with an upper level of education (75.2% hold an university degree or higher), with high purchasing power (74.9% earn more than 3500 euros per month, on average), mostly employed, they came from Germany or the UK (43.5%), from North Europe (29.5%) or from Central Europe (27.5%). These tourists have visited Portugal more than four times on average, numbers ranging from 2 to 255 visits. The most frequent tourists have started their relation with Portugal in 1950s. On average the relation of international tourists with Portugal started in 2006. These repeat tourists spend €108, on average, nevertheless the average spending range was €80–1,300, daily. The scaled variables were developed based on a five-point Likert-type scale at a low level, totally unsatisfied to extremely satisfied. The set of explanatory variables considered in this study attempts to capture the key determinants of revisit frequency, based on the theoretical framework and the literature review, as illustrated in Table 1.

Results A Poisson regression model was estimated with the number of visits to the final destination (‘How many times have you been to your final destination?’ being set as the dependent variable). Table 2 shows the results of this as well as the marginal effects for categorical variables. We confirmed that the dependent variable follows a Poisson distribution using a Kolmogorov–Smirnov test with Lilliefors correction (81.642, p = 0.000). The Poisson regression model was considered the appropriate model to fit the data given that we found no evidence of over-dispersion (variance greater than the mean) in the dependent variable. In this situation the Poisson model is more efficient than the Negative Binomial Regression Model (see, as an example, Wooldridge, 2002). The loglikelihood value of the estimated Poisson model is –21,309.096. The overall fit of the model is reasonably good, with a Chi-square statistic value of 26,156.06 for 34 degrees of freedom and a level of significance of 0.000 due the sample size. The model also presents a pseudo-R2 = 0.3803. The goodness of fit of the Poisson model estimated allow us to compute the marginal effects. They represent the change in the probability of an observation of being classified in each specific category of the dependent variable, according to the values of the predictors. The list of the independent variables, the respective coefficient, parameter significance and marginal effects computed by Equation (10) are also available in Table 2. There it can be find explanatory variables able to test the six hypotheses outlined. Table 3 shows the factor change, computed using Equation (7) and the percentage change, computed using Equation (8), for the non-categorical independent variables. Given the model specification, positive values for the parameters imply that the length of stay increases with increasing values in the respective variable. A negative value for the parameters implies a negative relationship. With only one exception it was possible to find a significant relationship with the number of visits at 5% or even 1% significance level: the intention to recommend Portugal for sure does not show a significant effect on the number of visits.

Why do tourists persist in repeat visits?

215

Table 2. Estimation results for the Poisson regression model and marginal effects for categorical variables. Variable H1 – Socio-demographic Adults Seniors Family average monthly income More than €5,000 Employment situation Active H2 – Tripographic Average stay (number of nights) Type of accommodation chosen Aparthotel Rented house H3 – Tourists’ behaviour Moderate expectations Intention to return to Portugal Probably Certainly Intention to recommend Portugal Probably Certainly Satisfaction Shopping Information available Transportation facilities Closeness to home Gastronomy Climate Relaxing environment Sightseeing and excursions H4 – Recency Number of years visiting Portugal German and UK tourists visiting Portugal since 1970s Central Europe tourists since 1970s Algarve tourists since 1970s Algarve tourists since 2000s Lisbon Tourists since 2000s Azores Tourists since 2000s H5 – Monetary value Average spending (euros) by tourists on a daily basis H6 – Heterogeneity German or UK Nordic tourists’ satisfaction in Algarve Nordic tourists’ satisfaction in Madeira Central European satisfied tourists Tourists’ satisfaction in the Algarve Tourists’ satisfaction in the Azores Constant

SD

0.0995 0.1689

0.0140 0.0147

7.1200 11.5000

0.0000 0.0000

0.7040 1.2380

0.0272

0.0115

2.3700

0.0180

0.2036

0.1068

0.0122

8.7600

0.0000

0.8134

0.0017

0.0008

2.1300

0.0330

0.0161

0.4810 0.0840

0.0117 0.0297

41.1500 2.8200

0.0000 0.0050

3.6410 0.5165

–0.1036

0.0129

–8.0100

0.0000 –0.7952

0.1581 0.2977

0.0246 0.0242

6.4300 12.2900

0.0000 0.0000

–0.0818 –0.0038

0.0322 0.0305

–2.5400 –0.1300

0.0110 –0.5932 0.9010 –0.0288

0.0600 –0.0667 –0.0652 0.0698 0.1140 –0.1943 0.0893 –0.1335

0.0125 4.7900 0.0177 –3.7600 0.0118 –5.5000 0.0172 4.0500 0.0138 8.2800 0.0155 –12.5400 0.0168 5.3100 0.0171 –7.8200

0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.0592 –0.5485

0.0004 135.0500 0.0413 –13.2800

0.0000 0.6083 0.0000 –4.0793

–0.2180 0.2537 –0.0826 –0.1577 0.2944

0.0395 0.0402 0.0154 0.0338 0.0594

–5.5200 6.3100 –5.3600 –4.6700 4.9600

0.0000 –1.6210 0.0000 1.8867 0.0000 –0.6144 0.0000 –1.1726 0.0000 2.1892

–0.0004

0.0001

–5.4300

0.0000 –0.0392

–0.0632 –0.1784 –0.0635 –0.0788 0.0674 –0.1881 0.8814

0.0253 –2.5000 0.0361 –4.9400 0.0308 –2.0600 0.0095 –8.3400 0.0050 13.4000 0.0145 –12.9900 0.0468 18.8200

Source: Authors’ analysis of UALG survey data.

z

P>|z|

Δ E(y| X)

Coefficient

0.0130 0.0000 0.0390 0.0000 0.0000 0.0000 0.0000

1.0060 2.0367

0.4508 –0.5083 –0.4825 0.5062 0.8257 –1.5337 0.6464 –1.0454

–0.4697 –1.3267 –0.4723 –0.5864 0.5016 –1.3988

216

TOURISM ECONOMICS

Table 3. Marginal effects: factor change and percentage change for the Poisson regression model. Variable: Average stay (number of nights) Number of years visiting Portugal Average spending in euros by tourist on a daily basis

Factor change 1.0017 1.0610 0.9996

Percentage change 0.1706 6.0973 –0.0402

Source: Authors’ analysis of UALG survey data.

As hypothesized, socio-demographic variables have a positive effect on expected revisit pattern, which means that tourists with relatively high budgets tend to exhibit a higher repeat visit pattern, these being the tourists in active employment situation. That is, for tourists with an average monthly income of more than €5,000 the expected number of visits increases by 0.2036. Accordingly, when tourists’ employment situation is active the number of expected visits increases by 0.8134. Further, adults and seniors present a positive and significant coefficient, although the seniors’ coefficients are higher. This result is reinforced by marginal effects: an increase of 0.704 for the adults and of 1.238 for the seniors, suggesting that senior tourists are more likely to become true loyal tourists in Portugal, with a high repeat visit pattern. This is in accordance with the findings of Ryan (1995), although it seems quite surprising since senior tourists are mostly recognized as novelty-seekers (Hsu et al, 2007). Overall, H1 is accepted: socio-demographic tourist characteristics affect the past repeat visit patterns. Concerning H2, it is the type of lodging of previous visits that influences the past frequency of visit most, confirming that repeat visitors are looking for familiarity in the type of lodging used – apart hotels and rented houses. That is, when tourists choose to stay in an aparthotel the number of past visits to the final destination increases by 3.641, or by 0.5165 if the tourists decide to stay in a rented house, compared to other types of accommodation. This is in accordance with Dias et al (2013) who suggest that the more informal the type of lodging the more comfortable the tourists may feel (away from home but feeling at home). Hence H2 is accepted, tripographic characteristics partially influence the past repeat visit pattern. H3 reflects the notion that the past frequency of revisiting Portugal is explained by tourists’ expectations, satisfaction and behavioural intentions. The results show mixed effects, echoing the disagreement already outlined in the literature review. Negative effects are related to expectations and behavioural intentions to return. When tourists have moderate expectations about Portugal and when their intentions to return are not tacitly assumed, the past frequency of revisit is likely to decrease. Through marginal effects it is possible to conclude that having moderate expectations towards the final destination decreases the number of visits by 0.7952. This is an intuitive result that groups the most likely spurious loyal tourists (Dick and Basu, 1994). Tourists planning to return to Portugal have a higher number of expected visits (1.0060 visits if they answer ‘probably’ and 2.0367 visits if they are ‘sure’) whereas tourists planning to recommend Portugal have a smaller number

Why do tourists persist in repeat visits?

217

(minus 0.5932 if they answer ‘probably’ and minus 0.0288 if they are ‘sure’ to recommend). It has been argued that satisfaction is a condition for repeat destinations (Kozak, 2003), this satisfaction being a proxy of the attachment the tourists have to the destination. This condition is necessary but not sufficient due to the novelty effect (Cohen, 1972). Concerning the sample of repeater tourists in Portugal, the results are mixed: satisfaction with some attributes is significant and positively affects the expected frequency of revisit, whereas others negatively affect the frequency of revisiting. A decreasing effect is expected due to the influence of other attributes. More specifically, tourists considering commerce as a satisfactory or very good aspect when deciding to repeat a visit to Portugal have a higher number of past visits (the number increases by 0.450). It is also shown that tourists who are satisfied with the closeness to home (with a marginal effect of 0.5062), gastronomy (with a marginal effect of 0.8257) and the relaxing environment (with a marginal effect of 0.6464) show a higher number of past visits to the final destination. These represent the competitive factors in Portugal with the potential to retain the most frequent tourists. Further, the expected number of visits turns out to be smaller for tourists satisfied or very satisfied with the information available (minus 0.5083). It is surprising that the weather has a negative effect on frequency of visits, as climate is the most consensual competitive advantage in Portugal, this may suggest that the promotion around the climate is inflated. Overall H3 is partially accepted, expectations and satisfaction with Portuguese tourism destination attributes influences the estimated number of past revisits even if in opposite axes. Intentions to return partially affect the repetition phenomena. Following the presuppositions of the RFM paradigm, H4 and H5 were proposed with past visits happening more recently, as more frequent tourists likely to spend more on the same destination than the first-time visitors. The effect is more evident in recency than in monetary value. Generally the number of years visiting Portugal increases the expected number of visits (each additional year after the first trip to the final destination increases the estimated number of past visits by 6.1%). However, German and UK tourists visiting Portugal since the 1970s, Central Europe tourists visiting Portugal since the 1970s and Azores tourists since the 1960s have a negative impact in the expected number of visits. This may suggest that the relationship with the destination is ceasing, which is not surprising considering the duration of the relationship these tourists maintained with Portugal; more than 30 years of repeat visits contradicts all the patterns of destination life cycles (Butler, 1980). At the regional level tourists visiting Algarve or Lisbon since 2000, presenting a life cycle of 12 years or less, were about to cease this relationship. Monetary value moderates negatively the frequency of repeat visits: for each additional euro spent daily, the number of expected visits decreases by 0.0402%. H4 and H5 are supported, providing empirical evidence in the RFM paradigm, since coefficients are statistically significant. Last but not least, heterogeneity was assessed by nationality and by region visited and, further, the cross effect of nationals visiting a specific region of Portugal and the satisfaction provided by each region were assessed. The results show that heterogeneity exists and the effects are mixed. A thorough analysis

218

TOURISM ECONOMICS

of the marginal effects obtained outlined important tendencies. For example, the most traditional markets in Portugal – Germany and the UK, as well as Central Europe – are likely to decrease their frequency of revisiting the country, illustrating that these markets are about to reach the decline stage, which in the near future should mean a ceased relationship. Concerning heterogeneity, it is suggested that tourists from Germany or the UK are less likely to persist in repeating the visit, at least on a regular basis, with the expected number of visits reduced by 0.4697. The overall satisfaction with the Algarve may lead to an increase in the expected number of visits. In the Azores the expected number of visits decreases with tourists’ satisfaction. Also, for Nordic tourists visiting Algarve or Madeira, the expected number of visits decreases with tourists’ satisfaction. As the coefficients are all significant, it may be assumed that H6 is supported. These results bring Portuguese tourism policy into question.

Conclusion and strategic implications The estimation results suggest a number of findings based on visits to several Portuguese touristic destination regions. First, the results emphasize the erosion of Portugal as a tourism destination for traditional travellers who began their trips in the 1970s. Overall, time contributes positively to a higher number of repeat visits. This is almost an intuitive result. However, this dynamic is moderated by differences among certain nationalities and Portuguese regions. Tourists to the Algarve, having started to travel there in the 1970s, show a very positive reinforcing effect when it comes to repeat visits. However, this does not apply uniformly to all tourist segments. Overnight guests from Central Europe counterbalance this positive effect. In contrast, more recent tourists (who first travelled to Portugal since 2000) show evidence for increased repeat visit behaviour the shorter the time the relationship has existed. This effect holds true for tourists in the Algarve and Lisbon but not for those in the Azores. Second, concerning changing patterns in tourism demand, the decreasing effect of monetary value on average spending in Portugal is also noteworthy, suggesting that the longer people stay the lower their daily expenditure is. This may prioritize capturing tourists with a higher purchasing power as well as retaining repeat tourists who are less sensitive to price. Moreover, the preference for a self-catering/aparthotel or rented house accommodation, as opposed to a traditional hotel stay, leads to a depletion of traditional lodgings. These results, which are cause and consequence of a new paradigm of travelling, also call for a reassessment of traditional lodging, to take into account an increasing tendency in the phenomenon of tourism repetition. Third, the average length of stay, which in familiar destinations may lead to a wish to stay longer and longer, suggests an increasing marginal utility or the non-satiation principle of revealed preference. An increasing marginal utility is expected in destinations where innovations and emerging activities feed the optimal level of the novelty tourists seek, even in familiar destinations (Correia et al, 2009). Fourth, the probability of return for a holiday in Portugal is evident for adults (increases the expected number of visits by 0.704), even greater for

Why do tourists persist in repeat visits?

219

seniors (increases the expected number of visits by 1.238), clearly denoting an avoidance of the risk implied when searching for new holiday options. Here, a certain ‘conservative’ attitude seems to be associated with more advanced age groups. Individuals with higher incomes (over €5,000 monthly) exhibit a greater propensity to repeat visits to Portugal; this finding may be due to a lack of interest in finding cheaper alternatives, or to other factors, such as scarcity of time to plan holidays elsewhere, with some trust being conferred on an option already well known. Further, ‘commerce availability’ determines an increase of 0.4508 in effective holiday repetition. The availability for purchase of holiday ‘self-gifts’, at a very attractive price, at the holiday destination is an important factor encouraging repetition. If there is a decent amount of information available about the holiday destination, the ‘novelty’ effect seems to be completely offset, this acts in a counter-effective way in terms of holiday repetition (decreases the number of expected visits by 0.5083). The ‘closeness’ of the holiday destination to home seems to be of major importance (0.5062). Such an attribute forms part of a low ‘generalized cost’, a favourable combination of fare and travel time, such as that enabled by the recent emergence of cheap and frequent ‘low-cost’ airline services. Local gastronomy and the relaxing nature of the environment appear to have a similarly important effect to destination proximity on the phenomenon of holiday repetition (0.8257 and 0.6464). Nonetheless, it is noteworthy that satisfaction with local transport facilities, sightseeing and excursions, and particularly weather, denote a smaller propensity to repeat visits to Portugal, suggesting that tourists assume a sedentary situation. In contrast, the negative effect of the attribute ‘weather’ (–1.5337) appears to be rather confusing and somewhat illogical. Finally, we should comment on the influence of ‘nationality’ and ‘satisfaction’. Results for the nationality-set ‘dummy variables’ in our model pinpoint the decrease in importance of the British and German markets. As such, the model seems to suggest a complex transition period experienced by the Portuguese tourism sector, in which a traditional and formerly ‘secure’ and ‘faithful’ clientele is fading away, while a new generation of other nationalities is appearing rapidly. The challenge here is to secure the new tourists while trying to recapture at least some of the traditional customers. With regard to study limitations and future research, it is suggested to extend the research to other destinations, to develop a cross-country analysis, and to extend the conceptual framework to include loyalty and to examine why frequent tourists are more loyal. Furthermore, this research assumes that repeat visits are related to some sort of risk aversion; nevertheless, future research might assess the effects of perceived risk on the decision to return. References Baloglu, S., and Erickson, R.E. (1998), ‘Destination loyalty and switching behavior of travelers: a Markov analysis’, Tourism Analysis, Vol 2, pp 119–127. Barros, C., Butler, R., and Correia, A. (2008), ‘Heterogeneity in destination choice’, Journal of Travel Research, Vol 47, No 2, pp 235–246. Barros, C., Butler, R., and Correia, A. (2010), ‘The length of stay of golf tourism: a survival analysis’, Tourism Management, Vol 31, No 1, pp 13–21.

220

TOURISM ECONOMICS

Bellman, S., Johnson, E., and Lohse, G. (2000), ‘Consumer buying behavior on the internet: findings from panel data’, Journal of Interactive Marketing, Vol 14, No 1, pp 15–29. Butler, R.W. (1980), ‘The concept of the tourist area life-cycle of evolution: implications for management of resources’, Canadian Geographer, Vol 24, No 1, pp 5–12. Cameron, A.C., and Trivedi, P.K. (2013), Regression Analysis of Count Data, 2nd edn, Econometric Society Monograph, No 53, Cambridge University Press, Cambridge. Cohen, E. (1972), ‘Toward a sociology of international tourism’, Social Research, Vol 39, pp 164– 182. Cohen, J.B., and Houston, M. (1972), ‘Cognitive consequences of brand loyalty’, Journal of Marketing Research, Vol 2, pp 97–99. Correia, A., and Kozak, M. (2012), ‘Exploring prestige and status on domestic destinations: the case of Algarve’, Annals of Tourism Research, Vol 39, No 4, pp 1951–1967. Correia, A., Pimpão, A., and Crouch, G. (2009), ‘Perceived risk and novelty-seeking behaviour: the case of tourists on low-cost travel in Algarve (Portugal)’, in Woodside, A., Harrill, R., and Crotts, J., eds, Advances in Culture, Tourism and Hospitality Research Vol 2, Emerald, Bingley, pp 1–26. Cunningham, S.M. (1967), ‘The major determinants of perceived risk’, in Cox, D.F., ed, Risk Taking and Information Handling in Consumer Behavior, Harvard University Press, Boston, MA, pp 82– 108. Day, G.S. (1969), ‘A two-dimensional concept of brand loyalty’, Journal of Advertising Research, Vol 9, pp 29–35. Darnell, A.C., and Johnson, P.S. (2001), ‘Repeat visits to attractions: a preliminary economic analysis’, Tourism Management, Vol 22, No 2, pp 119–126. Deaton, A. (2005), Franco Modigliani and the Life Cycle Theory of Consumption, Research Program in Development Studies and Center for Health and Wellbeing – Princeton University, Princeton, NJ. Dias, J., Correia, A., and Martinez, F. (2013), ‘From information sharing to vacation rental choices: the case of Albufeira – Portugal’, 8th CPTHL Symposium Consumer Psychology of Tourism Conference, Hospitality and Leisure Research, Istanbul, 1–5 June. Dick, A., and Basu, K. (1994), ‘Customer loyalty: towards an integrated framework’, Journal of the Academy of Marketing Science, Vol 22, No 2, pp 99–113. Dillman, D. (1978), Mail and Telephone Surveys: The Total Design Method, John Wiley & Sons, Hoboken, NJ. Eagly, A.H., and Chaiken, S. (1993), The Psychology of Attitudes, Harcourt Brace Jovanovich, Fort Worth, TX. Ehrenberg, A. (1955), ‘Measurement and mathematics in psychology’, British Journal of Psychology, Vol 46, pp 20–29. Jang, S., and Feng, R. (2007), ‘Temporal destination revisit intention: the effects of novelty seeking and satisfaction’, Tourism Management, Vol 28, pp 580–590. Fishbein, M., and Ajzen, I. (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Addison-Wesley, Reading, MA. Gitelson, R.J., and Crompton, J.L. (1984), ‘Insights into the repeat vacation phenomenon’, Annals of Tourism Research, Vol 11, No 2, pp 199–217. Gokovali, U., Bahar, O., and Kozak, M. (2007), ‘Determinants of length of stay: a practical use of survival analysis’, Tourism Management, Vol 28, No 3, pp 736–746. Goodhardt, G., Ehrenberg, A., and Chatfield, C. (1984), ‘The Dirichlet: a comprehensive model of buying behaviour’, Journal of the Royal Statistical Society, Vol 147, No 5, pp 621–655. Hughes, A.M. (1996), ‘Boosting response with RFM’, Mark Tools, Vol 5, pp 4–10. Hsu, C.H.C., Cai, L.A., and Wonga, K.K.F. (2007), ‘A model of senior tourism motivations – anecdotes from Beijing and Shanghai’, Tourism Management, Vol 28, No 5, pp 1262–1273. Jacoby, J. (1971), ‘Brand loyalty a conceptual definition’, in Proceedings of the American Psychological Association Vol 6, American Psychological Association, Washington, DC (1971), pp 655–656. Jacoby, J., and Kyner, D.B. (1973), ‘Brand loyalty versus repeat purchasing’, Journal of Marketing Research, Vol 10, pp 1–9. Kozak, M. (2003), ‘Measuring comparative destination performance: a study in Spain and Turkey’, Journal of Travel and Tourism Marketing, Vol 13, No 3, pp 83–110. Krishnamurthi, L., and Papatla, P. (2003), ‘Accounting for heterogeneity and dynamics in the loyalty–price sensitivity relationship’, Journal of Retailing, Vol 79, No 2, pp 121–135. Mazanec, J. (2000), ‘Mastering unobserved heterogeneity in tourist behavior research’, Tourism Analysis, Vol 5, No 2/4, pp 171–176.

Why do tourists persist in repeat visits?

221

Mazanec, J., and Zins, A. (1996), ‘Alpine winter tourists and guest satisfaction: an exploration of the typological relationships’, in Weiermair, K., ed, Alpine Tourism-Sustainability: Reconsidered and Redesigned, Institut für Tourismus und Dienstleistungswirtschft, Innsbruck. pp 186–213 Moutinho, L. (2000), Strategic Management in Tourism, CABI, Walllingford. Niininen, O., and Riley, M. (1998), ‘Repetitive buying behaviour in tourism, The Academy of Marketing Annual Conference, Cheffield. Oppermann, M. (1999), ‘Predicting destinations choice – a discussion of destination loyalty’, Journal of Vacation Marketing, Vol 5, No 1, pp 51–65. Oppermann, M. (2000), ‘Tourism destinations loyalty’, Journal of Travel Research, Vol 39, No 11, pp 78–84. Ryan, C. (1995),’Learning about tourists from conversations: the over-55s in Majorca’, Tourism Management, Vol 16, No 3, pp 207–215. Reichheld, F., and Sasser, W. (1990), ‘Zero defects: quality comes to services’, Harvard Business Review, Sept–Oct, pp 105–111. Sharp, B., and Sharp, A. (1997), ‘Loyalty programs and their impact on repeat-purchase loyalty patterns’, International Journal of Research in Marketing, Vol 14, No 5, pp 473–487. Shoemaker, S., and Lewis, R.C. (1999), ‘Customer loyalty: the future of hospitality marketing’, Hospitality Management, Vol 18, pp 345–370. Tian-Cole, S., Crompton, J.L. (2003), ‘A conceptualization of the relationships between service quality and visitor satisfaction, and their links to destination selection’, Leisure Studies, Vol 22, pp 65–80. Valle, P., Correia, A., and Rebelo, E. (2008), ‘Determinants of tourism return behaviour’, Tourism and Hospitality Research, Surrey Quarterly Review, Vol 8, No 3, pp 205–219. Van den Putte, B. (1991), ‘20 years of the theory of reasoned action of Fishbein and Ajzen: a metaanalysis’, unpublished manuscript, University of Amsterdam, Amsterdam. Wind, Y., and Lerner, D. (1979), ‘On the measurement of purchase data: surveys versus purchase diaries’, Journal of Marketing Research, Vol 16, pp 39–47. Wooldridge, J. (2002), Econometric Analysis Of Cross Section And Panel Data, The MIT Press, Cambridge, MA. Woodside, A.G., and MacDonald, R. (1994), ‘General system framework of customer choice processes of tourism services’, in Gasser, R.V., and Weiermair, K., eds, Spoilt For Choice. Decision Making Processes and Preference Change of Tourists. Intertemporal and Intercountry Perspectives, Kulturverlag, Thaur, pp 30–59. Zins, A.H. (2001), ‘Relative attitudes in customer loyalty models: some experiences in the commercial airline industry’, International Journal of Service Industry Management, Vol 12, No 3, pp 269–294.