Introduction

In today's highly competitive business environment, airlines, like all other companies, need to create and follow a good competitive strategy in order to gain an advantage over their competitors. Reports show that only a few airlines are profitable (Sampere 2016) and the credit quality of the industry is at risk these days (Piad 2020). At this point, a fitted competitive strategy can help them to cope with the observed problems in the industry (Acquaah and Yasai-Ardekani 2008; Heizer and Render 2014). For instance, an airline’s decisions such as flight route, airport selection, or in-cabin service offerings are formed by its competitive strategy. There are three main competitive strategies known as cost leadership, differentiation and focus (Tanwar 2013). In this context, cost leadership calls for delivering the cheapest travel services, while differentiation requires unique and full service at a premium price. Focus strategy, on the other hand, involves paying particular attention to a niche market by using either differentiation or low-cost strategies (Allen and Helms 2006).

It has been discovered that the service quality expectations are higher in airline industry compared to other industries (Hussain et al. 2015). This situation reveals service quality as a key performance indicator for airlines (Heisig et al. 2016). For this reason, service quality measurements are performed on a regular basis, and necessary actions are planned accordingly. In addition, official institutions like the US Department of Transportation publish statistics about quality rankings of airlines along with online comparisons (Shine and Joseph 2018). These reports show that airline service quality is worsening (Lebeau 2015), illustrating the challenges of achieving high quality of service in the airline industry today. Moreover, it is also not easy to measure an airline's service quality (Young et al. 1994); it is intangible and measured by the subjective judgments of customers (Parasuraman et al. 1985). At this point, SERVQUAL is the most common method recommended for this purpose (Celik et al. 2014; Pakdil and Aydın 2007). It has five dimensions as tangibles, assurance, responsiveness, empathy, and reliability.

An airline's competitive strategy can affect all five dimensions of service quality, but it is not easy to say exactly in which direction this effect is. For instance, full-service airlines can offer much more services along with additional space and comfort. Thus, we can expect them to provide a better quality of service. Nevertheless, the situation may also be the other way around, as their customers have much higher expectations than the customers of low-cost airlines (Choi et al. 2015; Dhar 2015).

Similarly, the intention to recommend is considered a good predictor of performance (Finn et al. 2009). It refers to the situation in which an airline's customers want to recommend the airline to others. In other words, it represents word-of-mouth communication (Kumar et al. 2007). The relevant literature shows some focus on the relationship between service quality and intention to recommend in the context of competitive strategies (Arasli et al. 2005; Hellier et al. 2003; Kim et al. 2007; Ladhari et al. 2011; Park et al. 2004; Teeroovengadum et al. 2016; Yuen and Thai 2015). Nevertheless, the results are unclear and indeed reveal some mixed evidence (Baker 2013; Kos Koklic et al. 2017). Namely, it is not possible to say that full-service airlines offer better service quality, and as expected, their intention to recommend is higher. Thus, this study aims to answer the following two questions to fill the gaps observed in the relevant literature.

  1. 1.

    Is there a difference in service quality and intention to recommend between low-cost and full-service airlines?

  2. 2.

    Which factors are significant for the intention to recommend in the airline industry?

In fact, many studies have been conducted on the airline service quality in Turkey and other countries (Deveci et al. 2018; Gupta 2018; Perçin 2018). This study differs from them in a number of ways. First, to the best of our knowledge, none of the previous studies clearly answers the above-mentioned two questions. Second, this study proposes a new hybrid methodology in this context. To do this, a survey instrument is developed under fuzzy environment and then it is validated by using exploratory factor analysis (EFA). In previous studies, on the other hand, either crisp survey instruments are used with EFA or fuzzy survey instruments are proposed without any form of statistical validation (Gupta 2018; Perçin 2018).

The remainder of this study is organized as follows. Section 2 reviews the previous literature related to service quality of airlines with emphasis on its relationship with competitive strategies and the intention to recommend. The research design is clarified in Sect. 3, and then the analysis results and findings are provided in Sect. 4. Finally, conclusions are presented in Sect. 5.

Literature review

We can divide the relevant literature into three main categories as shown in Table 1. In the first category, the studies that focus on service quality measurements in the airline industry are considered. The second category of studies deals with the service quality and its relationship with customer satisfaction and loyalty in different settings. Finally, the third category includes studies focusing on the relationship between the service quality, competitive strategies, and intention to recommend, which are closer to our study.

Table 1 The relevant literature
Full size table

The use of SERVQUAL and Multi-Criteria Decision Making (MCDM) methods to measure service quality is quite popular in the airline industry. Chou et al. (2011) suggest using fuzzy set theory to express the weights of evaluation criteria in a Taiwanese airline case study. Similarly, Tsaur et al. (2002) come up with a SERVQUAL-based model combined with fuzzy AHP and TOPSIS. Basfirinci and Mitra (2015) on the other hand utilize SERVQUAL with the Kano model to obtain quality scores of airlines from the US and Turkey. More recently, Atalay et al. (2019) put forward a hybrid method based on fuzzy importance, performance, and impact analysis (FIPIA) with information entropy for the service quality of airlines in Turkey. There are also studies using different models than SERVQUAL (Chen 2016; Gupta 2018; Liou et al. 2011; Perçin 2018). To carry off this, Liou et al. (2011) propose 28 survey items under eight dimensions in Taiwan by using a modified VIKOR. Similarly, Perçin (2018) employs a combined fuzzy model of DEMATEL, ANP, and VIKOR by considering 16 quality criteria under five dimensions. Chen (2016) utilizes DEMATEL and ANP by using 12 criteria under four dimensions. In a more recent study, Gupta (2018) presents a model of best worst and VIKOR by examining 29 criteria under seven dimensions.

Besides measuring the service quality, its relationship with customer satisfaction and loyalty is also investigated. Namukasa (2013) views service quality in three dimensions as pre-flight, in-flight, and post-flight services in Uganda. His/her results point out that all three dimensions have a notable impact on customer satisfaction; and customer satisfaction has a significant impact on loyalty. Similarly, Saha and Theingi (2009) consider the relationship between service quality, satisfaction, and behavioural intensions (word-of-mount or repurchase intensions) in Thailand for three low-cost airlines. They indicate that service quality is a noteworthy determinant of satisfaction; and quality and satisfaction have a remarkable impact on behavioural intensions. From a wider perspective, Hussain et al. (2015) study the relationship between corporate image, customer expectations, service quality, perceived value, customer satisfaction, and brand loyalty in a Dubai-based airline. They disclose that service quality, perceived value, and brand image have a positive impact on customer satisfaction; and customer satisfaction may lead to brand loyalty. Moreover, Tsafarakis et al. (2018) come up with a MCDM-based ordinal regression for quantifying passenger satisfaction, and outline the critical service dimensions that need to be improved for Greek flag carrier Aegean airlines.

Finally, some studies focus on full-service and low-cost airlines for service quality and intention to recommend. They reveal the relationship with a variety of evidence. For example, Koklic et al. (2017) study the link between airline tangibles, quality of personnel, customer satisfaction, intention to purchase and intention to recommend. They find that customers regard low-cost airlines as providing inferior quality. However, they have also found out that customers are still more satisfied with the low-cost airlines. Similarly, Baker (2013) finds that low-cost airlines outperform full-service airlines in terms of flight on time, passengers denied boarding, passengers complaints, and mishandled baggage. Additionally, Bubalo and Gaggero (2015) find out that low-cost airlines contribute to reducing delays at European airports and improve the service quality in terms of on-time performance of all flights landing. On the other hand, O’Connell and Williams (2005) find out no difference between full-service and low-cost airlines in the attitudes and perceptions of customers from European and Asia. From a different perspective, Azadian and Vasigh (2019) explore the possibility of convergence of the low-cost and full-service airlines in the USA and find some level of convergence between full-service airlines and Southwest Airlines, but not for other low-cost airlines.

Research methodology

This study uses a hybrid method, combining fuzzy rating scales and statistical models to take advantage of both approaches. Compared to standard surveys with the Likert scale, fuzzy rating scales provide a better tool for considering respondents’ ambiguous opinions, especially in the evaluation of service quality (Jónás et al. 2018; Lizarelli et al. 2021). That is, survey items in the service quality scale are mostly vague and difficult to put into precise numerical values. Therefore, verbal expressions, which can be expressed with triangular fuzzy numbers, are chosen in the evaluations (Liu et al. 2015). On the other hand, statistical models make it possible to check the validity and reliability of the survey data and allow for easier data analysis. Thus, to make it possible, we defuzzify or clarify the survey data. Indeed, this type of clarification is a preferred method by many researchers (see e.g. Leon and Martin 2020).

The hybrid approach preferred in this study is presented in Fig. 1.

Fig. 1
figure 1

Research model

Full size image

In the first phase of Fig. 1, a questionnaire is developed for the service quality, and then a survey study is conducted with airline customers in Turkey. The evaluations are obtained via linguistics terms, which are subsequently converted to triangular fuzzy numbers. Later, Explanatory Factor Analysis (EFA) is performed on defuzzified data for validity and reliability check (Dobni 2008). Accordingly, the questionnaire is finalized. In the second phase, service quality scores and the intention to recommend in full-service and low-cost airlines are compared. Meanwhile, a binary logistic regression model is constructed to extract the significant factors for intention to recommend.

Assuring about content validity is very important in designing of a questionnaire. It requires that the survey items in a questionnaire cover the major content of a construct. This is usually achieved through an extensive review of literature (Reyes et al. 2016). To this end, a careful review of literature on service quality and airline industry is performed. It shows that SERVQUAL is widely adopted in measuring service quality of an airline. At this point, we extract the survey items from the related literature. Subsequently, the extracted survey items are discussed one by one with five practitioners and five academicians at different times. Consequently, 23 items are obtained for measuring the service quality of an airline. They are presented along with their source of inspiration in Table 2.

Table 2 Survey items for measuring service quality of airlines
Full size table

The verbal assessments of respondents are converted to triangular fuzzy numbers according to the seven-grade linguistics scale presented in Table 3. A triangular fuzzy number is represented with three parameters of \(l\), \(m\), and \(u\). Here, \(m\) refers to the most probable value, and \(l\) and \(u\) refer to the minimum and maximum values. For example, “Poor” in Table 3 is described by the triangular fuzzy number of (0, 0.1, 0.3). It means that the most probable value for the quality score symbolized by the variable “Poor” is 0.1, and the lower and upper bounds are 0 and 0.3, respectively.

Table 3 Fuzzy linguistics scale (Wang et al. 2016)
Full size table

Analysis of results and findings

The questionnaire shown in Table 2 is used to collect the data. It is a self-administrated questionnaire and performed at the beginning of 2017. The data is collected from Istanbul Ataturk International Airport and Istanbul Sabiha Gokcen International Airport at different times and dates. The survey is requested from the respondents with an air travel experience over the last year as we focus on perceived service quality (Jain and Gupta 2004). Accordingly, 204 responses are obtained, but a data screening process indicates that 173 responses are usable for data analysis. At this point, Hair et al. (1998) advice a sample size of 100 or larger with a minimum sample size of five-to-one ratio for observations and survey items. Similarly, Tabachnick and Fidell (2013, p.617) encourage to have a sample size in the range of 100–200 for most cases. Therefore, it shows that the sample size of this study, which includes 173 responses and twenty-three questionnaire items, is at an acceptable level. Descriptive information of the respondents is presented in Table 4.

Table 4 Descriptive information of the respondents
Full size table

Table 4 presents that the survey study is performed with the customers of five airlines in Turkey. Those are Turkish Airlines, Pegasus, Atlas Global, Onur Air, and Borajet. Here, while Turkish Airlines is full-service airline, the others follow a low-cost strategy (Acar and Karabulak 2015). Hence, Table 4 shows that 100 respondents are from a full-service airline, while the remaining 73 respondents are customers of low-cost airlines.

In data analysis, firstly, the best non-fuzzy performance (\(BNP\)) value of each survey item is calculated for EFA. It is done by using Eq. 1 (Çebi and Otay 2016).

$$BNP_{{\tilde{q}_{i} }} = \frac{{\left[ {\left( {u_{i} - l_{i} } \right) + \left( {m_{i} - l_{i} } \right)} \right]}}{3} + l_{i}$$
(1)

In making EFA, IBM SPSS version 25 is used in accordance with the guidelines by Reyes et al. (2016), Johnson and Wichern (2007, p.519), and Hair et al. (1998, p.90). Then, the survey data is found appropriate for EFA with a Kaiser–Meyer–Olkin Measure of Sampling Adequacy of 0.922 and a significant Bartlett's Test of Sphericity value at five percent. Meanwhile, EFA reveals five factors for airline service quality according to a priori criterion. At this point, generalized least squares are used for factor extraction along with Equimax rotation. The results show that 75.22 percent of the total variance is explained by those five factors. Nevertheless, three items (1, 7, and 8) have been discovered to have significant loadings on more than one factor; thus, they have been removed from the questionnaire. After recalculation of the loadings, the factors are labeled as reliability, responsiveness, empathy, tangibles, and assurance.

Subsequently, the reliability of each factor is checked with Cronbach’s alpha. Hair et al. (1998, p.118) reports that lower limit for Cronbach’s alpha is 0.70 for reliability of a factor. Our analysis reveals that items in each factor is consistent with a Cronbach’s alpha value of at least 0.864. In addition, total scale reliability is found acceptable with a very high Cronbach alpha value of 0.959. These results are presented in Table 5.

Table 5 EFA analysis results
Full size table

After EFA, the factor score of each quality dimension is calculated by the SPSS software. Factor scores with a range of −3 and 3 are chosen here over summated scales as they are used for subsequent statistical analysis (Tabachnick and Fidell 2013). Later, an ANOVA test is performed to compare the quality performances of full-service and low-cost airlines. The results are shown in Table 6.

Table 6 Comparison of quality scores for full-service and low-cost airlines
Full size table

Table 6 exhibits that the factor scores are significantly greater for the full-service airline in tangibles, responsiveness, and empathy dimensions of service quality at a significance level of five percent. However, in reliability and assurance dimensions, no significant difference is observed. This means that the respondents in our study do not feel any performance gap between low-cost and full-service airlines with regard to reliability and assurance. Similarly, ANOVA results for age, gender, travel type, travel frequency, travel purpose, airline company, income level, and education level are all found to be insignificant at significance level of 5 percent in service quality.

Furthermore, when the other performance criterion, intention to recommend, is considered, we discover that positive recommendation is significantly greater for full-service airlines. The results are presented in Table 7. Table 7 indicates that while the positive recommendation rate is only 64.3 percent for low-cost airlines, it is 92 percent for full-services airlines.

Table 7 Positive customer recommendation
Full size table

A binary logistic regression model (BLRM) is constructed to understand the reason behind this very large amount of difference in the positive recommendation of customers presented in Table 7. By means of a BLRM, it is possible to reveal the significant factors in predicting intention to recommend along with their relative importance (Tabachnick and Fidell 2013). To this end, the guidelines by Hair et al. (1998, p. 314) are employed with IBM SPSS version 25. In the BLRM, the intention to recommend (0 = not to recommend, 1 = positive recommendation) is selected as the dependent variable, while the five service quality dimensions are determined as independent variables. In addition, gender, age, travel frequency, travel purpose, travel type, income level, airline company, and education level of participants are considered as control variables. The Hosmer and Lemeshow test shows that the model has a good fit with a significance value of 0.464. Subsequently, Cox and Snell R-squared value of 0.520 and Nagelkerke R-squared value of 0.828 confirm that the BLRM is suitable to use.

The results of the BLRM are presented in Table 8. It shows that not all the variables in the model are noteworthy in the prediction. That is, responsiveness, empathy, and tangibles are significant at a five percent significance level. This signifies that those three factors can be used to predict the positive recommendation of a customer. Nevertheless, assurance and reliability dimensions along with all control variables have been found insignificant at a five percent significance level.

Table 8 Binary logistic regression results
Full size table

The odds ratio (Exp(B)) presented in Table 8 is used to understand the direction of variables in predicting the positive recommendation. Namely, as long as its value is greater (lower) than one, any increase (decrease) in that variable will create an increase (decrease) in the probability of a positive recommendation (Szumilas 2010). For example, an odd ratio of 177.8 for empathy indicates that the probability of a positive recommendation increases about 177.8 times greater for one-unit increase in the factor score of empathy. Similarly, it can be stated that it is 66.2 times greater to have a positive recommendation for one-unit increase in the factor score of responsiveness. Thus, as all the quality variables are on the same scale (i.e. between −3 and 3), we can rank their importance level with respect to their odd ratios. Accordingly, they can be ranked, in a descending order of importance, as follows: empathy, tangibles, and responsiveness. Figure 2 shows their relative contributions in predicting a positive recommendation.

Fig. 2
figure 2

Importance level of quality dimensions for positive recommendation

Full size image

Conclusions and implications

This study focuses on exploring the relationship between service quality and intention to recommend in low-cost and full-service airlines by introducing a new hybrid methodology. To achieve this purpose, firstly, a fuzzy questionnaire has been developed for the service quality. EFA shows that the service quality of airlines in Turkey can be explained by five dimensions labeled as tangibles, empathy, responsiveness, reliability, and assurance. In the comparison, ANOVA results indicate that the full-service airlines have a higher service quality and higher intention to recommend than low-cost airlines. Later, a BLRM is also modeled to discover the significant factors for intention to recommend. The model points out empathy, tangibles, and responsiveness as the key factors. However, reliability and assurance dimensions have been found insignificant for intention to recommend, as well as the control variables of age, gender, travel type, travel frequency, travel purpose, airline company, income level, and education level.

The findings of this study have important implications for both academics and airline managers. In particular, the newly introduced hybrid methodology allows academics to analyze the relationship between service quality and intention to recommend in low-cost and full-service airlines. From a managerial perspective, on the other hand, the study stresses the importance of service quality for airlines. Namely, it reveals that the service quality is the only factor that matters to intention to recommend among all other variables considered in this study. This means whether it is a low-cost or a full-service airline or not, it needs to focus on improving the service quality to have a better rate of a positive recommendation.

Finally, this study has also certain limitations that need to be mentioned here. That is, this study has been performed with a limited data set in Turkey. In addition, the results may be specific to Turkey, and the airlines mentioned here as having a service quality is highly sensitive to regional factors. Therefore, in a future study, it is recommended to conduct a similar study in different countries to compare the results.