How does ICT influence residents' attitudes towards tourism as a driver of development? A generalised ordered logistic regression analysis ICT for tourism as driver of development in lagging behind regions

This study investigates the residents' attitudes to tourism as a driver of regional development and the impact of digital technologies on these perceptions. This work considers a ‘ lagging behind ’ region of the EU and employs a generalised ordered logit regression. The main findings suggest that the positive attitude is affected by the digital technology use, education, non-job variables and touristic products such as landscape, food and wine and events. As a result, the digital technology and the enhancement of touristic products may serve as important factors for ‘ lagging behind ’ regions to meet a smart, sustainable and inclusive growth.


| INTRODUCTION
The latest report of the World Travel & Tourism Council (WTTC, 2019) highlights the contribution of tourism to the GDP growth (+3.9% in 2018) and the formation of additional jobs (319 million). This latter figure represents the 10% of global employment and 1/5 of all global net jobs created in the past 5 years. Tourism is therefore recognised as a key driver for socio-economic growth.
Since the pioneering work by Buhalis and Law (2008), an increasing interest in the studies of Information and Communication Technology (ICT) and tourism management has emerged by international scholars (Navío-Marco et al., 2018;Pourfakhimi et al., 2018;Ukpabi & Karjaluoto, 2017). ICT plays a crucial role in accelerating economic growth, particularly through the generation of new competitiveness in the tourist management, destinations and organisations (Benckendorff et al., 2019;Fern andez et al., 2020). The process of digitalization of the economy is re-shaping the tourism industry in terms of market competition, cost optimization, marketing and distributional chain in view of a more sustainably-oriented society (Assaf and Tsionas, 2018). The same process is also affecting the personalization of the touristic offer by emphasising the tourist's motivation, his/her own desires and personal experiences (Kotiloglu et al., 2017;Sigala, 2012). For the above reasons, an increasing attention of the tourism literature is now being paid to the demand-side. First, the evolution of the concept of 'value co-creation' in tourism (Casais et al., 2020;Lin et al., 2017;Rihova et al., 2015) which contributes to the understanding of the social sphere of the sustainability paradigm (Font et al., 2021;Heinonen et al., 2010;Liang & Hui, 2016).
Second, the relevance of tourists' attitudes in relation to several tourist activities (Crouch et al., 2016;Szopi nski & Staniewski, 2016). Third, the residents' attitudes and behaviours in viewing tourism as a driver of sustainable development (Blasco L opez et al., 2018;De Lucia et al., 2020;Hsu et al., 2020). In consideration of this latter aspect, the following elements allow us to argue why it is relevant an understanding of the residents' attitudes to consider tourism, driven by ICT use, a relevant factor for a sustainable regional development.
First, the limitations from low growth countries to enhance tourism and ICT developments with sustainability goals. On this aspect, Gössling and Hall (2019) emphasise the potential of the sharing economy, which, through the tourism (i.e., accommodation) industry, would contribute to fulfil the sustainable development goals (SDGs). Nonetheless, the authors argue that this potential could encounter drawbacks in terms of product transformation, ICT infrastructures, and ecological trade-offs particularly relevant in low growth countries.
Second, the issue of ICT and smart tourism specialisation in the context of regional sustainable development (Boschma, 2016;Encalada et al., 2017;Jucan & Baier, 2012). In particular, Jucan and Baier (2012) consider the contribution of ICT an important factor for inclusion in corporate social responsibility practises by tourism businesses in new emerging markets. A particular aspect is highlighted in the study by Boschma (2016). The author argues about the importance of the smart specialisation by regions endowed with diverse technologically related industries and diverse natural resources. This diversity would induce an increase in the degree of networking and creation of new capabilities in the tourism sector and would help the renovation of the local economy. This view integrates the role of ICT with a sustainable development approach. Furthermore, the smart specialisation in the tourism context also provides the basis of a knowledge-based innovation strategy, which in turn, promotes sustainable development processes at a regional level (Encalada et al., 2017).
Third, the social acceptance of ICT in tourism studies (Chunqing et al., 2019;Ukpabi & Karjaluoto, 2017). Based on the findings of Ukpabi and Karjaluoto (2017), three major study domains can be classified under this topic: (a) consumer adoption of web-based services in tourism. This field considers the services offered by various tourism websites to consumers (Kim et al., 2013;Kucukusta et al., 2015); (b) consumer adoption of social media in tourism. This domain takes into account how social media affects the travellers' choice on the their trips (Wand et al., 2014;Chen et al., 2015); and (c) consumer adoption of mobile transformation systems in tourism. This latter field of studies focuses on the way how increasing incremental innovation in tablets and mobile phones affects the consumer's adoption of these devices for tourism purposes (Lai, 2015;Lu et al., 2015).
Given the above considerations, the issue of residents' attitudes to tourism as a driver of sustainable regional growth lacks of a clear understanding of the resident perception in low growth regions in the EU. Our work intends to cover this gap and offers new insights for bottom up regional tourism and ICT policies to achieve sustainability goals in lagging behind regions of the EU. To do this, the present study uses a resident attitude approach to investigate the existence of the argumentations offered by the following papers: The work of Romão and Neuts (2017) argues three main points: the first one, in contrast with Benckendorff et al. (2019) and partially aligned with the considerations of Gössling and Hall (2019) highlighted earlier, argues that tourism offers a poor contribution to the achievements of the SDGs; the second one, highlights that regions endowed with natural resources are highly specialised in tourism. This high specialisation includes the concept of 'smart tourism' (Boes et al., 2016), where the term 'smart' implies the use of digital technologies in the tourism sector; and the third one, argues that the same regions experience, in contrast, high levels of unemployment and low growth. As a consequence, the tourism sector in these regions would not drive towards development paths.
In terms of methodological approach, our study adopts a survey analysis at regional level as key source of data and, based on the work by Brida et al. (2020), it employs a generalised ordered logistic model.
The use of this model overcomes the limitation of the classical ordered logistic regression which assumes the rigid proportional odds assumptions between the categories of a categorical variable. In addition, based on the recent insights of Henseler et al. (2015), Voorhees et al. (2015, and Rönkkö and Cho (2020) and the suggestions offered by Lindell and Whitney (2001), our work investigates the issues of divergent validity and misspecification of data measurement arising from behavioural attitudes.
Our contribution to the above mentioned debate provides a focus on the following research questions: RQ1. To what extent residents perceive tourism as a driver of regional development? What are the potential factors affecting this perception?
RQ2. To what extent residents recognise the ICT potential in the tourism sector and how this contributes to perceive tourism as a driver of regional development?
RQ3. How the residents' perception in RQ1 and RQ2 can contribute to the advancements and adoption of sustainable regional development policies?
Inspired by the issues arising in the above papers and to respond to our research questions, we aim to deepen the understanding on ICT use as an important determinant of the resident's perception of tourism as a driver of sustainable regional development in lagging behind regions. To do this we infer whether potential linkages between the residents' perception of ICT use and tourism exist in regions with high levels of unemployment and low growth. The study considers a quali-and quantitative approach carried out in the region of Puglia, in the south of Italy. This region is one of the 'lagging behind' regions of the EU with low growth performances and it is one of Italy's most visited tourist destinations for its natural resources, beaches, cultural and local traditions. In 2014, it also appeared in the National Geographic's best trips list (https://www. nationalgeographic.com/travel/best-trips-2014/). The choice of this region is also supported by the study of Romão and Neuts (2017) when they argue that regions endowed with natural resources are highly specialised in tourism. The present study provides quantitative evidence on the perception of tourism as driver of sustainable regional development caused, mostly, by the contribution of ICT use in planning tourism and recreational activities, other than other socio-economic determinants.
The present work is structured as follows: Section 2 illustrates a brief review of the literature; section 3 defines the methodology of the inferential analysis; section 4 describes an overview of the case study, the survey analysis, and the investigation of data measurement; section 5 presents the results of the stochastic model and the discussion; and finally, section 6 concludes the work.

| BACKGROUND
The understanding of residents' attitudes is of great importance for local governments and policy makers to achieve the goals of the Europe 2020 strategy of a smart, sustainable and inclusive growth (European Commission, 2010b). Under a theoretical point of view, the most commonly accepted framework considering resident's attitudes for tourism systems is that of the Social Exchange Theory (SET) (Blau, 1964;Emerson, 1976;Homans, 1961) which, according to Jonason and Middleton (2015), can be thought as the result of a 'cost-benefit analysis' that people do with their daily behaviours.
When referred to tourism, residents generally compare the positive effects with the negative consequences of it. If the net effect produces positive impacts, then residents are likely to support tourism developments and all the potential relationships that are generated (Lee, 2013).
Recent literature on the perception of the linkages between ICT, tourism and regional development can be considered the modern view of the Social Exchange Theory applied to tourism as well as the pioneering Bjorklund and Philbrick's (1972) host attitudinal model (HAM), in which the social sphere is essential to understand the needs of the local communities towards tourism systems. The literature can be classified under the following categories: (i) conceptual and theoretical issues, including the perceived needs, drivers and obstacles of innovation; (ii) perceived determinants and usefulness of innovation for tourism and systems; (iii) Tourism systems, ICT and regional development.
The first piece of the recent literature can be related to the studies of Najda-Janoszka and Kopera (2014) and Pikkemaat et al. (2018).
The former work discusses the main perceived barriers by the local residents (i.e., stakeholders, governments, and communities) to the development of tourism industry such as asymmetries in markets, low capacity to innovation and high transaction costs, volatility of touristic firm ownerships, vulnerability of demand and incapacity to adequate touristic supply, lack of trust between entrepreneurs and the policy maker in the tourism sector, free-rider behaviour which can affect future perspectives of regional development. The latter work refers to the perceived needs, drivers and innovation in tourism systems by Alpine communities. The authors use a survey analysis to define the above terms. Main results illustrate the perceived needs in terms of quality and destination improvements; the perceived drivers such as market observation, internal key players, external best practices, networking; and the perceived innovation factors to overcome knowledge gaps, employees' engagement, and cooperation among the actors of the local community.
The second category of literature refers to the works by Hall and Williams (2008), Orfila-Sintes and Mattsson (2009) and Qi et al. (2010). Hall and Williams (2008) offer the first attempt at a comprehensive review of innovation in tourism (including resident perception); and, at the same time, considers tourism as a driver of innovation. Orfila-Sintes and Mattsson (2009) focus on innovation perceived by the local host as driver of the hospitality sector. The authors emphasise, through a stochastic model, the incidence of additional services, booking through tour operators, the brand and owner-type of the hospitality infrastructure on the innovation decisions of the local host. The study by Qi et al. (2010) offer useful insights on the residents' perception of web applications to shop for recreational travels.
The authors use a survey analysis delivered to Hong Kong residents. Their main findings suggest that experienced residents are keen to use web technologies to plain their trips and appear more sensitive to price, payment security and website's reputation variations.
The third category of literature deals with the works of Nunkoo and Ramkissoon (2012), Palafox-Muñoz et al. (2013) and Garau-Vadell et al. (2018). Nunkoo and Ramkissoon (2012) support the idea that 'residents' trust in government actors is a significant predictor of their support for tourism […]' (p. 1015), which assumes an important role to boost the regional economy. Palafox-Muñoz et al. (2013) emphasise the community perception on the quality and sustainability of tourism considered as a key factor for regional development. Also, the authors underline that the interrelationships and attitudes between the local resident and the tourist helps to understand the growth of a place in view of tourism development initiatives in the regional context. Garau-Vadell Similar approaches to the SET and HAM theories above, are the well-known contributions provided by the Theory of Reasoned Action (TRA) and the Technology Acceptance Model (TAM) (Taherdoost, 2018;Ukpabi & Karjaluoto, 2017). The former is based on the pioneering work by Fishbein and Ajzen (1975) in the fields of Sociology and Psychology and subsequently used in tourism studies (Kim et al., 2011;Otieno et al., 2016); the latter is an extension of the TRA including three main user's features such as the perceived usefulness of a technology, the user's attitude towards its use, and the degree of easiness to use the technology (Amaro & Duarte, 2015;Nunkoo & Ramkisson, 2013;Sahli & Legoherel, 2015). The TRA model is based on three main components of the human cognitive sphere such as attitudes, social norms and behaviours, and the individual's intentions. As such, we consider it, together with the SET and HAM theoretical models, as foundations providing a thorough vision of the resident attitude towards ICT, tourism and regional sustainability. On the other Social acceptance to tourism and regional development development and the perception of the ICT as a contributor to this view in lagging behind regions of the EU.

| METHODOLOGY OF THE INFERENTIAL ANALYSIS
The stochastic analysis conducted in this work is based on the use of a generalised ordered logistic regression model. The choice of this model is not random. Likert-scale ordinal type categorical dependent variables (e.g., 1 = strongly disagree; 4 = strongly agree) are generally modelled with ordered logistic regressions (Williams, 2006). These models assume that the order of responses of the dependent variable is known but the distance between categories is not. For this reason, this model is also known as Proportional odds/Parallel Line Model.
The Brant test is generally used to check that the parallel lines assumption holds. The null hypothesis tests the hypothesis that the ordered logistic regression model is valid. In our case, the null hypothesis is rejected at 95% confidence interval. Based on this result, 1 we relax the assumption of the parallel lines. This means that both the intercepts and the slopes of the regression lines are allowed to vary across the categories of the dependent variable. We employ the generalised ordered logistic regression model as described by Williams (2016). The model considers the following equation: where j represents the number of the categories of the ordinal dependent variable Y, α j is the constant term of each category, X i s are the explanatory variables with i = 1…N, and β jthe slopesare the estimated coefficients of the explanatory variables for each category j.
To estimate the generalised ordered logistic regression model we use STATA v.16 and the command gologit2 written by Williams (2016).
How can the signs of the estimated coefficients of the explanatory variables can be interpreted? Positive β coefficients increase the likelihood that higher values of the explanatory variable are associated with higher categories of the dependent variable than the category under consideration. On the other hand, negative β coefficients indicate that higher values of the explanatory variable increase the likelihood of being in the current or a lower category (Williams, 2006).

| CASE STUDY
This section illustrates the case study of our investigation. After providing a geo-location of the region of Puglia, the main economic information particularly related to its tourism sector are provided. The section also shows the descriptive statistics of the survey.
The region of Puglia is located in the South-East of Italy

| Survey design and sample descriptive statistics
Our analysis is based on the use of a set of information gathered through a questionnaire structured into three sections. The first section of the questionnaire is designed to elicit the demographic information of the respondents. The second section captures a general knowledge of tourism perception; and the third section refers to information on the general knowledge of ICT, tourism and regional development issues.   In terms of regional distribution of the residency, 48% of the sample is from the northern part of the Puglia region (20% from the province of Foggia and 18% from the province of Barletta-Andria-Trani), 29% from the centre of the region (17% from the province of Taranto and 13% from the capital of the region Bari), and 33% from the south (16% and 15% from the provinces of Brindisi and Lecce, respectively).

| Data measurement
The present section deals with data measurement and reliability of constructs from multi-item scales of Likert-type variables of our survey (Bacharach, 1989). To this end, the above issues will be tested on our sub-set of 11 psychometric variables related to the attitudinal aspects towards: • the touristic products of the region of Puglia such as 'Heritage', 'Landscape', 'Infrastructures', 'Food and wine', 'Recreational activities', 'Touristic events', 'Hospitality' and 'Services'; • the ICT issue such as: 'ICT satisfaction for holiday planning' and 'ICT promotes tourism', and • the resident attitude's variable to perceive tourism as a driver of development: 'Tourism promotes sustainable regional development'.
The reliability of the above multi-scale constructs is provided through the use of the recent insights offered by the disattenuated correlation using parallel reliability i.e., heterotrait-monotrait ratio of the correlations (Henseler et al., 2015). Potential data mis-specification is investigated through the use of the Confirmatory Factor Analysis with Marker Variable method illustrated by Lindell and Whitney (2001).

| Discriminant validity of multi-scale items
The heterotrait-monotrait ratio ( To determine the validity of the attitudinal constructs in our study, the above method is based on factor loadings in factor analysis.  Figure 3 illustrates the Horn's Parallel analysis (Horn, 1965) to identify the number of factors (if greater than 1) in our sub-sample.
We notice that the maximum number of factors to extract to carry out the factor analysis is 2. We run the factor analysis with the oblique rotation method, since the factor correlation matrix in Table 4 presents values larger than the 0.30 threshold (Hinkle et al., 2003).
Finally, Tables 5 and 6 show the factor loading matrix and the HTMT ratio computations. In particular, the correlation squared value between components in  Table 7, 7 we proceed with the use of the classical one factor model in CFA.
From Table 8, the CFI indicates a 0.90 value. The closer the CFI is to 1, the better the fit of the model. In addition, the Root Mean Square Error of Approximation (RMSEA), which measures the degree of misspecification, has a value of 0.036. Since it is lower than 0.05 (the cutoff point for close-fit models), we can argue that our psychometrictype model is not biased. Table 9 illustrates the results obtained from the generalised ordered logistic regression model. The dependent variable of our inferential model is the resident attitude to perceive 'tourism as a driver of sustainable regional development'. This is a psychometric Likert-scale (1 = totally disagree 5 = totally agree) categorical variable.

| GENERALISED ORDERED LOGISTIC MODEL RESULTS
The output of this model is analogous to a sequence of binary  Table 9 may look odd; nonetheless, a positive coefficient indicates a higher value of the independent variable and therefore it is more likely that the respondent's attitude will favour higher categories of the dependent variable than lower ones.
We can argue that the panel of regression 1, 2, 3 & 4 run against 5 (Panel 4), presents the highest number of estimated factor variables with statistically significant coefficients using 95% (*** and **) of confidence interval. Among these, we find two socio-demographic variables such as 'Education', and 'Job'; three variables representing touristic products of the region such as 'Landscape', 'Food and wine' and 'Events'; and three variables considered ICT proxies such as: 'ICT tools for touristic events', 'ICT satisfaction for holiday planning' and 'ICT promotes tourism'.
Given the above, it is clearly evident a general positive attitude (i.e., categories 4 = agree and 5 = totally agree) that tourism is a driver of sustainable regional development. Based on this result, we next present the analysis of predicted probabilities for category 5 = totally agree. Table 10 illustrates the results of the analysis of predicted probabilities. The analysis of margins is computed for the predictors with statistically significant coefficients using 95% (*** and **) of confidence interval.

| Analysis of predicted probabilities
As for the socio-demographic variable 'Education', we notice that respondents with higher education (41%) and secondary school degree (42%) mostly affect the probability to perceive tourism as a driver of sustainable regional development. Surprisingly, in terms of 'Job', housewives increase the above probability by +54%.
As for the touristic products 'Food and wine' and 'Events', the higher the importance of these two variables (=Very important) for the resident, the higher the probability to perceive tourism as a factor promoting sustainable regional development (+43% and + 49%, respectively predicted probability (+59% and 42%, respectively) to have a strong perception of the relationship between tourism and sustainable regional development.

| Discussion
Our findings provide evidence to argue against the debate that tourism offers a poor contribution to the achievements of the SDGs particularly in low growth regions endowed with natural resources and highly specialised in tourism activities.
The discussion is developed in light of the results obtained from the analysis of predicted probabilities and is structured according to theoretical contributions, managerial implications and limitations.

| Theoretical contributions
In a society more oriented to adopt sustainable behaviours and economies, the benevolence of host communities and the investigation of

General contribution to the SET theory
The suggested results of our analysis point us towards a general positive attitude of the resident to perceive tourism as a driver of sustainable regional development. We can summarise the main factors affecting this general perception as socio-demographic, with particular reference to education and job, touristic products of the Puglia region and ICT. Although the latter two contribute to support the SET theory, we leave our reflections for these two main factors as specific contribution to the HAM, TRA, and TAM theories and concentrate our attention to the socio-demographic factors.
In terms of education, we believe that an increasing attention to     Ribeiro et al. (2013)

Specific contribution to the HAM theory
We argue that the positive outcome of the estimated coefficients for 'Landscape', 'Food and wine' and 'Event' to the probability of the resident attitude to tourism can be seen as a contribution in support of the HAM model. As already mentioned in the previous sections, these variables are extensively investigated as touristic products of a place (Fytopoulou et al., 2021;Lee, 2013Lee, , 2020Stylidis et al., 2014;Vukovic et al., 2019). In the context of our findings, we share the view that the concept of landscape from the resident point of view is more conceived as a dwelling place than the classical 'land expert' definition (Lee, 2020). Of course our reflections go beyond the diverse perspectives, definitions, and treaties on the meaning of landscape. We point out that the contribution of 'Landscape' to the HAM theory, is that other than its traditional morphological concept, the resident recognises his/her own identity with the experiences and relationships of the dwelling places (i.e., the landscapes) in his/her daily life (Jung & Han, 2015). As a result, the landscape is perceived by its hosts as tangible and intangible assets in which human and eco-systems interact, exchange and experience their social and cultural dimensions (Halling, 2011). Arguably, a similar view can be expressed for the resident attitude to gastronomy and events. 'Food and wine' or gastronomy in general, provides the opportunity for the host to present the richness of the regional culinary traditions to the visitor (Avieli, 2013; Brulotte & Di Giovine, 2014; Kim et al., 2013;Pérez G alvez et al., 2017;Salgueiro Rachão et al., 2021). Manniche and Larsen (2013) recognises the importance of the culinary product as an element of tourism attractiveness which should be conceived from a knowledge perspective. This view looks into the 'experience' of the visitor based on the socio-cultural heritage of the community to visit.
The knowledge perspective through the 'experience' of the community is seen as an innovative factor which contributes to create relevant exchange of cultural values across the visitor and the resident as well as to co-design new opportunities for growth, particularly in marginal and rural areas. As a result, the culinary tradition becomes a link between the resident and the visitor, a way to share and exchange cultures and traditions, and enhance the social dimension of the visit (Bruwer & Rueger-Muck, 2019;Campos et al., 2018;Chien et al., 2018;Wassler et al., 2019). Similarly, a touristic event can be considered as the contribution to both hosts and tourists to an unforgettable and pleasing experience, an intense interaction of emotions and shared behaviours (Ouyang et al., 2019). In addition, the perceived positive impacts for traditional events, such as exhibitions and festivals, can have significant beneficial effects on the resident attitudes and on his/her quality of life in the long term (Gursoy et al., 2017;Gursoy & Kendall, 2006;Schnitzer et al., 2020).

Specific contribution to the TRA and TAM theories
In our analysis we find a positive effect of 'ICT tools for touristic events', 'ICT satisfaction for holiday planning', and 'ICT promotes tourism' to the resident attitude. We argue that these findings complement the research and strengthen the existing results supported by the TRA and TAM theoretical models. According to one of the founding work by Schiffman and Kanuk (1987), human's behaviours are formed by the intentions to act. This is true particularly when ICT tools are used to get information (e.g., search, contents, transactions) on touristic events or to plan for holidays or recreational activities.
Therefore, the residents' attitudes to use ICT tools and its satisfaction can be seen as a function of the quality and speed of the circulation of information provided by the same tools (i.e., Web and Social) (Jiménez-Barreto & Campo-Martínez, 2018;Lin et al., 2019;Shafiee et al., 2013), and the quality of services offered for recreational activities (Gretzel, 2011;Yoo et al., 2017  As for this latter aspect, the literature identifies at least three types of tourist's generations: the first one is the tourist which is eager to have some free time to dedicate to his/her leisure activities (Cohen, 1979); the second one, is the mass tourist, which identifies the touristic destination as rooted in his/her own identity (Cohen, 1972); and the third one, defined as the post-modern tourist (Sohrabi et al., 2020;Xu et al., 2019), uses modern technology tools to participate, share and conform his/her own needs to demand the touristic destination. The degree of diffusion of new ICT tools offers, in the actual society, new opportunities for the socio-economic growth of a region and appears particularly relevant in terms of decision support systems for the local policy maker (Bonzanigo et al., 2016;Shafiee et al., 2019;Lazoglou & Angelides, 2020). New digital techniques, such as Geographic Information System (GIS), are powerful resources that contribute to the development of adequate tourism policies by making a proper use of all the resources (i.e., cultural, natural, human, economic) that a region is endowed with.
GIS techniques display the retrieved information in a clear way to support the policy maker and provide the identification of the spatial relationships between socio-economic and environmental ecosystems (Jeong et al., 2016;Wolf et al., 2018). In a recent paper by Sevilla- Based on this approach and the positive residents' attitudes to use ICT for tourism purposes, we can argue the usefulness of the latest technology to enhance tourism in particular 'lagging behind' regions of the EU such as the region of Puglia, and to meet the need of environmental protection. In the region of Puglia, for example, ecotourism activities can be particularly strengthened in the rural areas and meet the needs of marginalised residents. The recent Rural Development Programme (RDP, Regione Puglia, 2019) recognises the rural tourism as a driving force to contribute to the sustainable development of the local communities. Also, the RDP considers rural tourism as a mean to enhance specific regional features. In particular, the implementation of detailed actions targeted to develop cooperation among stakeholders and network creation for rural tourism would aim at integrating biodiversity into the different landscapes across the region, and at preserving traditional agricultural products.
A further important issue is that of smart specialisation. Romão and Neuts (2017) explore the case of ICT (among other factors) as element that can contribute to the sustainable development of EU regions; in particular the authors also suggest the creation of positive synergies between smart development (i.e., digital tourism experience) and smart tourism specialisation which would affect long-term economic growth.
This issue assumes particular relevance for the region of Puglia, in terms of strategic role for future growth (Asheim, 2019;Biagi et al., 2020;Polido et al., 2019). In recent years, the regional institution of Puglia is adopting an ad-hoc policy namely Puglia365. 8 Puglia365 is a strategic plan of tourism for the period 2016-2025. This strategic plan has been conceived as a participatory approach across various stakeholders to determine future guidelines of the regional tourism and support sustainable tourism entrepreneurs and managers. The regional tourism strategy focuses on six main priority areas such as touristic products, marketing, innovation, infrastructures, hospitality, and continuing education. More than two meetings, per each priority area, are regularly organised per year throughout the region. The participatory approach, thus, determines specific marketing and advertising actions in order to increase networking activity across stakeholders and competitiveness towards neighbouring areas and regions. This will help entrepreneurs and young start-ups to innovate and the policy practitioner to find adequate solutions to promote sustainable growth beyond 2020 (European Commission, 2019) through adequate regional ICT and tourism policies in the years to come.

| Limitations
The present work is not without limitations. First, although generalised ordered logistic regression models allow to relax the rigid assumption of the parallel lines for Likert scale variables, they are more difficult to interpret than the parallel lines ordered logistic regression model (Williams, 2006(Williams, , 2016.
Second, the analysis is developed in a low growth region in the south of Italy and lacks of comparisons of using a resident attitude approach with other regional contexts.

| CONCLUDING REMARKS
The present work explored residents' attitudes to consider tourism as a driver of sustainable regional development and the perceived role of ICT to emphasise this view.
We performed a case study in the region of Puglia in the south of Italy, one of the 'lagging behind' regions of the EU which suffers from low growth. Nonetheless, tourism performances of this region has shown increasing trends recorded during the 2008-2018 economic recession period. We applied both qualitative and quantitative analyses. The former one is carried out through a survey analysis of the regional resident investigating socio-demographic, touristic products, and ICT use. The latter is performed through an inferential analysis employing a generalised ordered logistic regression model. This type of model is currently receiving international attention due to the relaxation to the parallel lines assumption generally used for Likert-scale type categorical dependent variable models. In addition, we performed divergent validity and confirmatory factor analysis with marker variable method to our psychometric-type variables to test for data measurement and validity. These additional analyses provided to argue for the validity and existence of unbiased psychometric-type variables in our dataset.
Main findings from the generalised ordered logistic model suggested that ICT mostly contributes to perceive tourism as a driver of development by the residents of a EU region which is particularly endowed with traditions and natural resources but with low income growth and high unemployment rates. As a result, the regional governance may foresee challenging socio-economic implications driven by the development of the digital technologies to implement future sustainable development policies.