Customer satisfaction as suggested by [21] is a “slippery term”, due to its highly subjective nature and the inconsistency in definitions that exist [22] [23]. This makes the measurement, interpretation and comparison of customer satisfaction results challenging [24] [25]. Notwithstanding, tourist satisfaction is acknowledged as a principal indicator of a destinations’ performance and competitiveness [26] [27], and warrants measurement, interpretation and comparison, to inform policy makers, marketing decision makers, and tourism agencies.

Extant literature highlights that tourism satisfaction studies typically either focus on tangible, functional attributes that are readily measurable [28] [29], or on intangible, psychological aspects of satisfaction, which are more difficult to measure [30] [31]. However, functional attributes have gained more research attention compared to psychological attitudes [32]. In addition, wide ranging opinions exist as to the main attributes of a holiday that attract a tourist to a destination: scenery, hospitality, receptiveness, cost, climate, activities, entertainment and sports facilities [32] ; the physiography and climate, culture and history; range and mix of activities and entertainment available, events offered, superstructure, human relationships [33] ; and natural, social, historical, recreational and shopping facilities, food and shelter [34] [35].

Extant literature proposes various theories, approaches and instruments for measuring customer satisfaction [8], and more specifically for measuring the level of customer satisfaction of tourists. The main research instruments developed to analyse service quality and customer satisfaction are the Importance-Performance Analysis [36], SERVQUAL [37], SERPERF [38], and the HOLSAT holiday satisfaction [21]. The Importance-Performance Analysis uses the confirmation/disconfirmation concept [21] developed by [39] and [40], and has been criticised on the basis that empirical research suggests consumer satisfaction is a function of both expectations of performance and actual performance [36]. The SERVQUAL instrument measures the difference between customer’s expectations that an excellent (ideal) service would provide and customer’s perceptions of the actual service received [41], and has been widely criticised on the basis that an inbuilt bias towards dissatisfaction exists [21]. Some researchers [38] [40] suggest the removal of the expectations from the SERVQUAL and use only of the perceptions of actual performance to determine satisfaction of respondents. In line with this suggestion, the SERPERF developed by [38] measures only the service performance, and has also been criticised on the basis that “performance alone cannot give a full picture of satisfaction” [21].

Numerous researchers [42] [43] [44] [45], have used a range of instruments to measure the level of customer satisfaction (gap analysis based on expectations and perceptions of service), and applied this approach to the measurement of individual services of a destination, in the tourism sector [16]. Nevertheless, the approach adopted by these researchers focuses on the services delivered by a particular hotel, as opposed to assessing the overall holiday experience [16].

Considering the research instruments and approaches mentioned, and their obvious criticisms, the instrument adopted for use in this research is the HOLSAT model, because this model “overcomes some of the limitations of other models when satisfaction is measured for a destination rather than a specific service” [16]. The HOLSAT instrument addresses the multidimensional nature of tourist’s satisfaction with a specific destination through a comparison of a broad range of attributes (namely the physical resort and facilities; ambience; restaurants, bars, shops and light life; transfers; heritage and culture; and accommodation) of expectation against the holiday performance (“experience”) [21] [46]. However, the HOLSAT instrument does not utilise a fixed menu of attributes supposedly generic to all destinations; rather allows for customisation of attributes to best suit the particular destination under study. This flexibility in attributes addresses the concerns raised by [47] that fixed dimensions were not necessarily applicable in all destinations. Additionally, the use of both positive (favourable impressions of the destination) and negative (unfavourable impressions of the destination) attributes of a destination, is a peculiar characteristic of the HOLSAT instrument. This allows tourists to express both positive and negative aspects of a holiday [21] [46].

The HOLSAT instrument features 56 items (49 positive and 7 negative holiday attributes) representing the 6 categories of attributes, with tourists asked to rate their expectations before their holiday, during their holiday (performance measure 1), and after their holiday (performance measure 2), based on these items, on a 5-point Likert scale. “The scale was labelled and scored [strongly disagree (−4), disagree (−2), no opinion (0), agree (+2), strongly agree (+4), and an additional box was included labelled not applicable]” [21]. Mean scores for the expectations and performance are calculated, as is the mean sum of differences between the expectations and performance, for each respondent, and for each attribute item. In order to assist interpretation of the attribute item mean scores, the attribute item numbers are plotted in a matrix and numbered data points result. To further aid interpretation, a diagonal “draw” line is provided in the matrix to indicate where perception of performance (X-axis) equals expectations (Y-axis). For positive attribute items, numbered data points located above the “draw” line indicate “loss”, that is a loss of satisfaction, while numbered data points below the “draw” line indicate “win”, that is gain in satisfaction. For negative attribute items, numbered data points located above the “draw” line indicate “win”, that is gain in satisfaction, while numbered data points below the “draw” line indicate “loss”, that is loss of satisfaction. The further a numbered data point is away from the “draw” line, the greater is the loss or gain of satisfaction for that specific attribute [16]. Inferential analysis (paired samples t-test) is also used to determine the probability of difference between the tourist’s expectations and performance scores.

The modified HOLSAT survey instrument utilised for this empirical study consists of two parts and comprises 41 attribute items. The first part of the questionnaire (Part A.) contains 7 demographic questions, to measure the demographic characteristics of respondents. The second part of the questionnaire comprises 31 positive and 10 negative attribute items derived from the HOLSAT survey developed by [21], and from a modified HOLSAT survey [16]. Attribute items were also modified to suit the context of the tourist destination being studied. While the HOLSAT survey groups the attribute items into the following categories: physical resort and facilities; ambience; restaurants, bars, shops and nightlife; transfers; heritage and culture; and accommodation; it is suggested that this approach tends to somewhat focus on the characteristics of the hotel [21]. Hence, in this research, the “Five A’s” categorisation of attributes, as suggested by [16], was adopted: “attractions”, “activities”, “Amenities”, “Accommodation”, and “Accessibility”. While this categorisation was similar to the HOLSAT developers [21] categorisation, it is suggested that a broader view of the holiday (beyond the hotel) was possible. The second part of the questionnaire requires respondents to rate each attribute item according to a five-point Likert scale. “The scale was labelled and scored [strongly disagree (−4), disagree (−2), no opinion (0), agree (+2), strongly agree (+4)]” [21]. Respondents were asked to submit a rating for both their expectations (Part B.1.) and the performance (“experience”) (Part B.2.) for each of the 41 holiday attributes items. Table 1 shows a sample sheet from the questionnaire, namely Part B.1. Expectations.

A pilot test of the modified HOLSAT survey instrument was conducted with 20 tourists in Nadi, Fiji, in October 2016, to check the usability of the instrument and to gain valuable feedback from respondents regarding the wording of the instrument. Minor wording changes were suggested and were incorporated into the questionnaire. The questionnaire was designed in English.

The target population for this study was international tourists (n = 200) that had arrived in Fiji, and were staying in hotels. The public-intercept, self-completion questionnaire data collection occurred during December 2016 (a peak period when tourists visit Fiji), in the western division of Fiji (Nadi and Lautoka) and in the eastern division (Suva). Table 2 shows 245 tourists were intercepted, with 35 tourists declining to participate in the survey (response rate of 86.0 percent), and 10 questionnaires found to be unusable (completion rate of 95.0 percent), resulting in 200 usable questionnaires.

The Cronbach’s alpha is one of the most commonly used measures of internal consistency (reliability) for research where we have used a Likert scale and wish to determine whether the scale is reliable. In order to study the internal consistency of the questionnaire and items, the Cronbach’s alpha value and Cronbach’s alpha coefficients if item deleted were calculated. The overall value of the Cronbach’s alpha for the modified HOLSAT survey instrument comprising 41 items was 0.945. Considering the alpha value was above 0.800, it is suggested that the questionnaire has a good degree of internal consistency [48]. The Cronbach’s alpha coefficients if item deleted range between 0.943 to 0.945, and confirm that if we removed any of the items from the questionnaire it would result in a lower Cronbach’s alpha. Therefore, it is suggested not to remove any items from the questionnaire. Overall, results suggest that the modified HOLSAT survey instrument comprising 41 items and used in this research, can be considered a reliable measure of tourist’s perception of satisfaction with Fiji as a holiday destination.

Demographic profiling of the 200 respondents (Table 3) shows that the majority were males (n = 134, 67.0 percent) followed by females (n = 66, 33.0 percent), and were predominantly in the age ranges of 26 - 35 years (n = 69, 34.5 percent), 18 - 25 years (n = 67, 33.5 percent), and 35 - 50 years (n = 48, 24.0 percent). The

education level attained by respondents was primarily undergraduate degree (n = 146, 73.0 percent). Most respondents were married (n =133, 66.5 percent), and the dominant country of origin was either Australia (n = 101, 50.5 percent) or New Zealand (n = 39, 19.5 percent). For most of the respondents this was not their first visit to Fiji (n = 124, 62.0 percent), and the primary purpose of their visit was for a holiday (n = 131, 65.5 percent) or business (n = 46, 23.0 percent).

Table 4 shows the positive attribute item mean score (M) and standard deviation (SD) for respondents’ perceptions of performance (“experience”) and expectations, the mean difference between perceptions of performance and expectations, and the paired-samples t-test results. Mean results show that perceptions of performance range between item 38 “I would be able to visit a Fijian village” (M = 2.30) and item 4 “The resort would be safe and secure” (M = 3.66); and, the expectations range between item 17 “Transfer vehicle would be on-time and air conditioned” (M = 2.18) and item 1 “The beaches and ocean would be clean” (M = 3.54). Standard deviation results for perceptions of performance range between item 11 “Shops would be cheap”, item 22 “I would be able to sample local food and drink”, and item 29 “Hotel staff would be friendly” (SD = 0.81), and item 38 “I would be able to visit a Fijian village) (SD = 2.26); and, the expectations range between item 1 “The beaches and ocean would be clean” and item 29 “Hotel staff would be friendly” (SD = 0.87) and item 9 “Restaurants would be cheap” (SD = 2.21). These results suggest wide ranging variation in opinions of respondents in terms of perceptions of performance and expectations of the positive attributes of their holiday.

Paired-samples t-tests were conducted to compare perceptions of performance and expectations relating to the positive holiday attributes. Table 4 also shows that for 20 of the 31 positive holiday attributes (items 2, 4, 9 - 13, 16 - 17, 22 - 27, 29, 31, 38 - 40) there was a statistically significant difference in the scores for perceptions of performance and expectations (p < 0.05).

Of the 20 positive holiday attributes suggesting significant differences, 13 items showed statistically significant positive differences between perceptions of performance and expectations suggesting that respondents’ expectations had been exceeded and they were delighted with the performance experienced: item 4 “The resort would be safe and secure: t = 2.76, p < 0.05; item 9 “Restaurants would be cheap” t = 3.05, p < 0.05; item 10 “Bars would be cheap” t = 2.56, p < 0.05; item 11 “Shops would be cheap” t = 2.81, p < 0.05; item 12 “Nightlife would be cheap” t = 2.29, p < 0.05; item 13 “ Hotel meals would be of high quality” t = 2.16, p < 0.05; item 16 “The arrival airport would be modern and efficient” t = 2.08, p < 0.05; item 17 “Transfer vehicle would be on-time and air conditioned” t = 2.03, p < 0.05; item 22 “I would be able to sample local food and drink” t = 3.20, p < 0.05; item 25 “I would be able to shop in local markets” t = 2.31, p < 0.05; item 26 “I would be able to buy cheap Fijian artefacts” t = 2.27, p < 0.05;

*Bolded numbers show statistically significant t-test results.

item 27 “I would be able to experience an island cruise” t = 2.09, p < 0.05; item 29 “Hotel staff would be friendly” t = 2.28, p < 0.05.

Of the 20 positive holiday attributes suggesting significant differences, 7 items showed statistically significant negative differences between perceptions of performance and expectations suggesting that respondents’ expectations had not been met and they were dissatisfied with the performance experienced: item 2 “Check in and check out times would be flexible” t = −2.44, p < 0.05; item 23 “Water sports would be available” t = −4.16, p < 0.05; item 24 “I would be able to experience traditional Fijian singing and dancing” t = −1.99, p < 0.05; item 31 “The room would have a good view” t = −5.15, p < 0.05; item 38 “I would be able to visit a Fijian village” t = −6.07, p < 0.05; item 39 “I would be able to visit historical sites” t = −4.25, p < 0.05; and item 40 “I would be able to visit a rainforest” t = −4.89, p < 0.05.

For the remaining 11 positive holiday attributes there was no significant difference in the respondents’ scores for perceptions of performance and expectations (p > 0.05), suggesting that respondents’ expectations had been sufficiently met/close to being met that they were satisfied: items 1 “The beaches and ocean would be clean” t = 0.43, p > 0.05; item 3 “The resort would be clean” t = 0.44, p > 0.05; item 5 “The resort would have a variety of restaurants” t = 0.95, p > 0.05; item 6 “The resort would have a variety of bars” t = 0.62, p > 0.05; item 7 “The resort would have a variety of shops” t = 0.83, p > 0.05; item 8 “The resort would have a variety of nightlife” t = −0.65, p > 0.05; item 15 “In-flight service would be high quality” t = 0.31, p > 0.05; item 30 “The room would be quiet” t = 0.12, p > 0.05; item 32 “The hotel and my room would have good internet connectivity” t = 0.05, p > 0.05; item 33 “Facilities in the room would function properly” t = −0.31, p > 0.05; and item 34 “The room would have a comfortable bed and clean bedding” t = 0.37, p > 0.05.

Figure 1 is a variant of the HOLSAT explanatory matrix for positive holiday attributes, with perception of performance (X-axis) plotted against expectations (Y-axis). In order to assist interpretation of the attribute item numbers detailed in Table 4, the attribute item numbers are plotted in the matrix and numbered data points result [16]. Figure 1 shows that the positive holiday attribute data points were predominantly closely clustered around the “draw” line indicating

tourist satisfaction, and tending below the “draw” line indicating a gain in satisfaction (tourist delight). Some data points were located above the “draw” line indicating a loss of satisfaction (tourist dissatisfaction).

[21] suggest that when interpreting the negative attributes of the HOLSAT model, where the difference between perceived performance and expectations is negative, it is an indication of tourist satisfaction. Conversely, positive differences are an indication of tourist dissatisfaction. Table 5 shows the negative attribute item mean score (M) and standard deviation (SD) for respondents’ perceptions of performance (“experience”) and expectations, the mean difference between perceptions of performance and expectations, and the paired-samples t-test results. Mean results show that perceptions of performance range between item 18 “I would have difficulty obtaining cash from my credit card or ATM” and item 37 “There might be power cuts” (M = 1.13), and item 19 “I would have to be careful what I eat and drink” (M = 3.75); and, the expectations range between item 18 “I would have difficulty obtaining cash from my credit card or ATM” (M = 2.47) and item 19 “I would have to be careful what I eat and drink” (M = 3.51). Standard deviation results for perceptions of performance range between item 14 “There would be a lack of public toilet facilities” (SD = 1.14), and item 18 “I would have difficulty obtaining cash from my credit card or ATM” (SD = 2.36); and, the expectations range between item 21 “Immigration and customs clearance would be slow and inefficient” and item 41 “There would be pollution in the towns and cities” (SD = 1.03), and item 18 “I would have difficulty obtaining cash from my credit card or ATM” (SD = 1.69). While these results

*Bolded numbers show statistically significant t-test results.

also suggest some variation in opinions of respondents in terms of perceptions of performance and expectations of negative holiday attributes, the variation in opinions is less than for the perceptions of performance and expectations of positive holiday attributes.

Paired-samples t-tests were conducted to compare perceptions of performance and expectations relating to the negative holiday attributes. Table 5 also shows that for all the negative holiday attributes (items 14, 18 - 21, 28, 35 - 37, and 41) there was a statistically significant difference in the scores for perceptions of performance and expectations (p < 0.05).

Of the 10 negative holiday attributes suggesting statistically significant differences, 2 items showed statistically significant positive differences between perceptions of performance and expectations suggesting that respondents’ negative expectations had been exceeded: item 14 “There would be a lack of public toilet facilities” t = 3.43, p < 0.05; and item 19 “I would have to be careful what I eat and drink” t = 2.54, p < 0.05..

Of the 10 negative holiday attributes suggesting significant differences, 8 items showed statistically significant negative differences between perceptions of performance and expectations suggesting that respondents’ negative expectations had not been realised: item 18 “I would have difficulty obtaining cash from my credit card or ATM” t = −10.13, p < 0.05; item 20 “Changing money would be difficult” t = −2.92, p < 0.05; item 21 “Immigration and customs clearance would be slow and inefficient” t = −10.99, p < 0.05; item 28 “There would be many beggars and street vendors” t = −8.96, p < 0.05; item 35 “There might be shortages of some food and drink” t = −5.58, p < 0.05; item 36 “There might be shortages of fresh water” t = −6.61, p < 0.05, item 37 “There might be power cuts” t = −11.17, p > 0.05; and item 41 “There would be pollution in the towns and cities” t = −8.81, p < 0.05.

Figure 2 is a variant of the HOLSAT explanatory matrix for negative attributes, with perception of performance (X-axis) plotted against expectations (Y-axis). In order to assist interpretation of the attribute item numbers detailed in Table 5, the attribute item numbers are plotted in the matrix and numbered data points result [16].

Figure 2 shows that 8 of 10 negative holiday attribute data points were dispersed above the “draw” line indicating a gain in satisfaction (tourist delight), with only 2 data points below the “draw” line indicating a loss of satisfaction (tourist dissatisfaction).