CC BY
87
Section 11. Economics of recreation and tourism
D OI: http://dx.doi.org/10.20534/EJEMS-17-1-71-77
Musina Zanna,
University of Science and Technology Beijing (USTB), Beijing, P. R.C. PhD candidate, Donlink School of Economics and Management, Department of Management Science and Engineering E-mail: zanna-musina@yahoo.com Gao Xuedong,
University of Science and Technology Beijing (USTB), Beijing, P. R.C. Professor and PhD supervisor, Donlink School of Economics and Management, Department of Management Science and Engineering E-mail: gaoxuedong@manage.ustb.edu.cn
Improvement of External factor dimension's structure for the assessment of persuasiveness of tourism destination websites using Factor analysis
Abstract: Website is a complex system, which can be divided to several dimensions. This research is a part of research aimed to create Destination Management/Marketing Organization (DMO) tourism websites success evaluation instrument. External variables are cognitive and psychological factors, e. g. knowledge and interest towards destination, destination's popularity and perceived credibility of website. These variables affect individual's attitudes and beliefs and are influencing behavioral intention to visit a destination. Aim of this research is to explore the structure of the external variables and reduce quantity of variables for usage in subsequent research. Exploratory factor analysis discovered different structure from initially proposed basing on theory review. Four new factors were extracted, total quantity of variables were reduced. Factors were not found to be strongly reliable, and fourth factor found to be least reliable.
Keywords: Factor analysis, External factors, Website persuasiveness, External Factor dimension, Tourism website evaluation.
Introduction
In the era of digital information, when it became common to consult internet sources, many scientists are focusing on the topics related to the website management improvement aspects. Website is a platform for disseminating information, providing services, doing business. In tourism sector, website is important marketing instrument, well-designed and well-managed website may raise destination's popularity and increase visitation. Scientists concentrated much attention towards the question of the website quality and the influence of
website elements on the success of website and behavioral intentions of users [1; 2], including topics related to persuasiveness of communication [3; 4; 5; 6].
Apart of website elements and service elements provided on website, individual's knowledge, experience, motivation and constraints are forming prospective tourist's attitude towards destination and influence website's persuasiveness, in this research such factors are called — External factors.
The aim of this research is to explore the structure of the data related to external factor dimension, reduce
number of variables to achieve reliable structure which can be used in the further research.
1. Theory review
1.1. External variables and website persuasiveness
This research is a part of overall research working with Destination management/marketing organization (DMO) website's success evaluation model [7] which is based on the DeLone and McLean Information System Success Model [8] and the Technology Acceptance Model (TAM) [9]. Present research focuses on the re-
lationship between External and Other factors (further called External factors) and persuasive power of tourism website. (This relationship is marked with doted arrow in Fig. 1.)
Website's persuasive power can be defined as an influence of website on individual's intention to visit recommended destination. Persuasive power of communication is affected by source of the message [3], elements ofwebsite [4; 5], users' knowledge and interest towards destination [6] and other external factors.
Figure 1. DMO website's success evaluation model [7, 69]
There are variety of external factors which can influence individual's perception and behavior. Different researchers use set ofdifferent external factors according to the specifics of their research topic [10, 166; 11, 40; 12, 29-30] The external factors or external variables are used in the TAM introduced by Davis (1986) as an initial element in the model. External factors are influencing beliefs and attitudes and the latter are influencing user behavioral intentions [9, 985] External variables may be regarded as any external (device, technology, help etc.) or internal factors (psychological, cognitive etc.) which can stimulate behavioral intention, and influence individual's believes and attitudes.
The TAM was inspired by the Theory of Reasoned Action (TRA) developed by Fishbein and Ajzen
(1975). In the TRARA behavioral intention is influenced by individual's attitude towards behavior and by subjective norm. Subjective norm reflects a pattern when other people's opinion towards behavior influence individual's behavioral intention [9, 984]. It was suggested in other studies to research more about attitude towards destination and prior knowledge of it as mediating variable in the process of destination selection. (Perdue, 2001 cited in 5, 78).
In this research external variables' set includes tourist's prior knowledge of tourism destination, destination's reputation, website's reputation sub-dimensions. (see table 1.) [7] Prior knowledge of tourism destination and website's reputation sub-dimension cover individual's
attitude towards destination and website, and includes cognition degree of the destination.
Different users may have different involvement levels with the website or destination. Individual's involvement is described as a state of motivation, arousal, interest and so on which is directly connected to the perceived value, risk, importance etc. associated with destination or website [6, 41]. As suggested by Tang et al. (2012) destination website communication strategies should consider individuals' involvement levels, and develop communication routs' persuasion strategies to address more effectively those different groups.
Subjective norm is reflected in destination's reputation sub-dimension. Other people's opinion often influ-
ences individual attitude towards behavior or object. Website's reputation sub-dimension also covers the area of trust towards a mediator, information provider. Credibility of the message is influencing behavioral intention to visit a destination [5, 77] More trusted and more qualitative website would potentially have greater persuasive power and will more successfully influence potential tourist's intention to visit tourism destination advertised on the website. For example, it is known that users more often are consulting government tourism websites. (Boyne & Hall, 2004 cited in [13, 75]) Government tourism websites are perceived as more qualitative and abundant source of information comparing to private websites [14, 71].
Table 1. - External variables
External variable sub-dimension Variable name Measurement items
Tourist's prior knowledge of tourism destination frsttm You did not know before tourism destination country presented on the website.
plnvisit You've already made plans to go to the tourism destination country presented on the website, but haven't been there yet.
intrst only You feel interest towards the tourism destination country presented on the website, however, you do not have plans to visit it.
visited d You have already visited tourism destination country presented on the website.
Destination's reputation famous d The tourism destination country represented on the website is famous.
ntfamous d The tourism destination country represented on the website is not famous.
popular d The tourism destination country represented on the website is popular, everyone wants to visit it.
ntpopular d The tourism destination country represented on the website is not popular and few people want to visit it.
safe d The tourism destination country represented on the website is safe.
ntsafe d The tourism destination country represented on the website is not safe.
Website's reputation wbs rel Tourism destination website's reliability is high.
wbs ntrel Tourism destination website's reliability is average.
1.2. Research aim and questions
The goal of the research is to explore chosen variables and external variable's sub-dimensions, eliminate least reliable variables from the framework. New variable framework will be used in the further research.
Research questions:
1) What is the structure of external variables subdimension and the relationships between variables?
2) If some of the measurement items should be omitted from the further research?
2. Methodology
2.1. Data collection and results
Data for this research was collected from Chinese respondents in the period from March till June, 2015. Respondents were asked on the streets of Beijing (P. R.C.) to fill in self-administered questionnaires; partially responses were obtained through Internet website www. wenjuan.com. All together were collected 116 questionnaires in Chinese language. Seven point scale is used for the measurement, where 7 stands for highest influence on travel intention and 1-no influence at all on travel in-
tention. Missing data accounted per variable not more than 10%, therefore, it is acceptable to ignore missing values, or employ any imputation technique to substitute missing values [15, 47-48].
2.2. Factor analysis
Factor analysis is technique which does not require dependent variable. This analysis is used for defying data structure, understanding relationships between variables and reduction of variables, for example by elimination of variables which have least correlation with factor, and creating uncorrelated factors [15, 91-151]. In this research R-type exploratory analysis is used to define relationships among variables and group them into latent factors [15, 93].
To use factor analysis researcher is required to check data for the appropriateness. Data measurements, data sample size, correlation among the variables [15, 103105; 16, 2].
Normality is not always considered necessary. For example, principal component and principal factor methods can be used when data is not normally distributed [16, 2] But normality is considered important if significance of the factor is statistically tested [15, 103].
Data should be measured at least at ordinal level, however coded (1-2) data can be used as well [17, 240] The general rule of thumb is to have minimum absolute sample size of 50 observations and 5 observations or more per variable [15, 102; 16, 2] In this research there are 12 variables, therefore sample size of 116 observations is adequate.
Data correlation is checked through Bartlett's test of sphericity, there sig. < 0.05 indicates that correlation is sufficient among variables; if after visual check there are discovered substantial number of correlations greater than 0.30, this indicates that the sample can be used in factor analysis; inspection of anti-image correlation matrix helps to determine variables with high partial correlation, larger anti-image correlations may signal that matrix is not suitable for factor analysis; measures of sampling adequacy (MSA) which lay on diagonal of anti-image matrix must exceed 0.5, if value is less than 0.5 variable should be eliminated from the analysis; and Kaiser-Mayer-Olkin measure (KMO) greater than 0.5 shows that sample is adequate for factor analysis [15, 103-105; 16, 2].
After the sample is tested for the adequacy, following steps are performed:
1) Extracting initial factors;
2) Specifying number of factors;
3) Rotating factors;
4) Interpreting rotated factor matrix;
5) Terminal solution on factor structure;
6) Validating new factors [15].
3. Analysis and results
3.1. Profile of respondents
Gender distribution of respondents is equal, 50% of males and 50% females filled in the questionnaires. Young people from 18 to 35 years old accounted for 88% of responses, elderly people from 36 to 60 years old -10%. More than half of the respondents — 58% were students, 26% — companies' employees, 4%- people working in governmental sector and 12% of another occupation. Most of the people have never travelled before, but have plans for travelling in future- 88% and 86% accordingly. Most of the respondents have experience using Internet and often are using it for tourism related information search, among them 10% always are using Internet for tourism information search, 47% are often using Internet, but there is also quite high percentage of those who rarely consult Internet for this purpose- 42%.
3.2. Results of Factor analysis
In this research for the factor extraction was used one of the most used methods — Principal Components Analysis (PCA). This method is used for summarizing most of the variance of variables in the minimum number of uncorrelated factors. Principal Component considers all of the variance (unique and common) in the variables. PCA can be used with data which is not normally distributed [15, 107; 16, 2-5].
Criteria for decision on number of factors which are commonly followed are: 1.Kaiser's criteria- Eigen Value > 1; 2. Kettell's criteria- Examination of Scree plot keeping all the factors before the breaking-point (K). It is suggested to check solutions for number of factors K, K+1 and K-1.; 3. Percentage of variance explained should be at least 50%; 4. Hypothesized number of factors also should be taken into account, factors should have sense [16, 4; 18, 259-260; 15, 108-111].
Factors were rotated with orthogonal rotation Varimax. Orthogonal rotation produces uncorrelated factors [19, 3]. Varimax rotation is maximizing the sum of the squared factor loading across the columns, it tends to give clearer separation offactors, where variable tend to have maximum loadings on as few factors as possible [15, 115; 16, 5].
3.2.1. Factor solution
First factor extraction approved that sample can be used in Factor analysis (Bartlett's test sig. = .000, KMO= = .749, MSA values > .6). However it also revealed that communality ofvariable "famous_d" is low (.477) and it cross-loads with other factors, hence extraction of fac-
tors was repeated without this variable. Second extraction with rotation showed that variable "visited_d" had significant cross-loading with another factor, and would complicate interpretation of factor solution; therefore it was eliminated from the further analysis. New extraction and rotation proved to be most successful among other possible variants, as it allowed simplest and easy to interpret factor structure.
Factor extraction gave solution of four factors with eigenvalues greater than 1 end explained 71% of total variance in factors, which is satisfying. (see table 2.) The results of Bartlett's test of sphericity statistics showed sig. = .000, KMO value = .704, correlations greater than 0.3 accounted for 29% of cases (13 out of 45 cor-
Factor 1 contains 3 variables: ntfamous_d (loading .841), ntpopular_d (.767), intrst_only (.718). According to the nature of variables this factor can be named as "Not prestigious destination".
Factor 2 contains 3 variables: wbs_rel (loading .848), famous_d (.712), safe_d (.693). It can be named as "Reliable place".
Factor 3 contains 2 variables: wbs_ntrel (.891) and ntsafe_d (.853). It can be named as "Unreliable place".
Factor 4 contains 2 variables: plnvisit (.813) and frsttm (.792). This factor is unified by the concept of involvement. "plnvisit" variable is describing situation when individual already knows about the destination and has a plan to visit it. However variable "frsttm" describes situation when individual has no prior knowledge of the destination, and no prior intention to discover destination. Therefore possible name can be "Involvement".
3.2.2. Validity
Results of factor analysis show that the factor structure is not very stable as there are two variables which
relations) at sig. = .000, MSA values exceeded value of .5 and anti-image matrix did not contain large correlations — all these indicates that sample matrix is suitable for further analysis.
All factors contain variables with indicative loadings (> .7), which show that the factor structure is well-defined. [15; 116-117]. Result are statistically significant, as reported for sample size of 100 observations significant loading is 0.512 (Stevens 2003, P. 294 cited in (16), 6 to 0.55 [15, 117]). No variables have significant cross-loadings. Particularly, difference between highest loading of variable and loading on other factor (same variable) does not exceed 0.2. [18, 271] Rotation results are displayed in the table 2.
have difference between two loadings lower than .05. [18, 271-272] Particularly, for the variable "intrst_only" difference is .388 (.718 (factor 1)- .330 (factor 2)), and for variable "frsttm" difference is .433 (.792 (factor 4) -.359 (factor 1). (See table 2.) Generally accepted that more reliable factors contain 3 and more variables with significant loadings, the value of loading can be smaller if number of observations is bigger. [16, 3-6; 19, 5] Therefore first two factors are more reliable. It is reported that the Chronbach's alpha is influenced by the number of items in the tested scale. It is generally agreed that Chronbach's alpha > .7 indicates that items in scale are reliable [15, 125], however, values > .6 are also acceptable for exploratory research (Robinson at al. (1991) cited in ibid.). The results of Chronbach's alpha test prove that factor solution is reliable. For Factor 1 Cronbach's alpha is .740, Factor 2 it is.680, Factor 3 it is .735 and Factor 4 it is .599. Therefore 4th factor is least reliable along with other findings this leads us to decision that this factor can be excluded from the next research framework.
Table 2. - Rotated Component Matrix
Component Communality
1 2 3 4 5
ntfamous d 0.841 -0.080 0.155 0.229 0.791
ntpopular d 0.767 0.252 0.203 0.114 0.706
intrst only 0.718 0.330 -0.037 -0.040 0.628
wbs rel 0.096 0.848 0.098 -0.017 0.738
famous d 0.187 0.712 -0.106 0.259 0.620
safe d 0.111 0.693 0.053 0.136 0.513
wbs ntrel 0.073 0.022 0.891 0.116 0.814
ntsafe d 0.161 0.037 0.853 -0.134 0.772
plnvisit -0.035 0.284 -0.180 0.813 0.776
frsttm 0.359 0.051 0.191 0.792 0.795
Rotation Sums of Squared Loading s Total
% ofVariance 20.29 16.69 16.81 14.74 71.53
Conclusion Factors "Reliable place" and "Unreliable place" both
The results of the Exploratory Factor analysis suggest characterize credibility of objects. These two factors that the initial theoretical structure of the variables should are reflecting positive and negative sides of credibility. be changed from free to four dimensions. New three fac- Both factors contain items which cover destination and tors were named as "Not prestigious destination", "Unre- website reliability. Additionally, factor "Reliable place" liable place" and "Reliable place", fourth factor "Involve- contains concepts which are based on reputation and ment" is found out to be least reliable, therefore can be prestige of destination, and partially reflecting subjec-omitted in the consequent research. Quantity of variables tive norm and individual's knowledge. was reduced from 12 to 10. Variables "popular_d" and Factor "Involvement" is least reliable factor. Variable
"visited_d" were excluded from Factor solution. "plnvisit", reflecting formed intention to visit destination.
Factor "Not prestigious destination" reflects concept However variable "frsttm" is reflecting situation when of subjective norm, or other people's opinion. From one potential tourist has no prior intention to visit destina-side, this factor emphasizes that places which are rela- tion, and no knowledge of destination. In both cases the tively infamous among other people have influence on level of the involvement with destination was different. the degree of website persuasiveness. However it also This research is based on Chinese language ques-
contains individual's attitude towards destination. Un- tionnaire; therefore there can be some ambiguity, which discovered destination, destination which does not have comes from the language conceptual differences. Represtige is correlating positively with the destination search is of exploratory nature and is due to confirma-which is interesting for user, but this interest is not high tion by other studies. enough to stimulate person to form travel intention.
References:
1. Bai B., Law R., Wen I. The impact of website quality on customer satisfaction and purchase intentions: Evidence from Chinese online visitors. International Journal of Hospitality Management. - 2008; - 27 (3): -P. 391-402.
2. Chung N., Lee H., Lee S.J., Koo C. The influence of tourism website on tourists' behavior to determine destination selection: A case study of creative economy in Korea. Technological Forecasting and Social Change. - 2015; -96: - P. 130-143.
3. Sparks B. A., Perkins H. E., Buckley R. Online travel reviews as persuasive communication: The effects of content type, source, and certification logos on consumer. Tourism Management. - 2013; - 39: - P. 1-9.
4. Lee W., Gretzel U. Designing persuasive destination websites: A mental imagery processing perspective. Tourism Management. - 2012; - 33 (5): - P. 1270-1280.
5. Loda M. D., Teichmann K., Zins A. H. Destination websites' persuasiveness. International Journal of Culture, Tourism and Hospitality Research. - 2009; - 3 (1): - P. 70-80.
6. Tang L. R., Jang S. S., Morrison A. Dual-route communication of destination websites. Tourism Management. -2012; - 33 (1): - P. 38-49.
7. Musina Z., Gao X. DMO tourism website's success evaluationmodel and framework. European Journal of Economics and Management Sciences. - 2016; - (4): - P. 61-76.
8. DeLone W. H., McLean E. R. The DeLone and McLean Model of Information Systems Success: A Ten-Year Update. Journal of Management Information Systems. - 2003; - 19 (4): - P. 9-30.
9. Davis F. D., Bagozzi R. P., Warshaw P. R. User Acceptance of Computer Technology: A Comparison of Two Theoretical Models. Management Science. - 1989; - 35 (8): - P. 982-1003.
10. Monzavi T., Zarei B., Ghapanchi A. H. Investigating the Impact of External Factors on User Perceptions: A Case Study of Software Adoption in Middle East Actual System Use. The International Technology Management Review. - 2013; - 3 (3): - P. 160-174.
11. Sternad S., Bobek S. Impacts of TAM-based External Factors on ERP Acceptance. Procedia Technology. - 2013; -9 p. - 33-42.
12. Sternad-Zabukovsek S., Bobek S. TAM-based external factors related to ERP solutions acceptance in organizations. International Journal of Information Systems and Project Management. - 2013; - 1 (4): - P. 25-38.
13. Horng J. S., Tsai C. T. Government websites for promoting East Asian culinary tourism: A cross-national analysis. Tourism Management. - 2010; - 31 p. - 74-85.
14. Dion C. P., Woodside A. G. Usefulness of Government and Private Destination Websites. Tourism-Marketing Performance Metrics and Usefulness Auditing of Destination Websites Advances in Culture, Tourism and Hospitality Research. - 2010; - 4 p. - 69-137.
15. Hair J. F., Black W. C., Babin B.J., Anderson R. E. 7th ed. ЗЬЖ: ШМ1
- 2011.
16. Habing B. Department of Statistics. Art & Science. University of South Carolina. [Online].; - 2003 [cited 11 11 2016. Available from: URL: http://www.stat.sc.edu/~habing/courses/530EFA.pdf.
17. Ho R. Handbook of Univariate and Multivariate Data Analysis with IBM SPSS. Second Edition ed. Boca Raton: CRC Press Taylor and Francis Group, an Inform business; - 2014.
18. Наследов А. Д. Математические методы психологического исследования. Анализ и интерпретация данных. Учебное пособие СПб.: Речь; - 2004.
19. Costello A. B., Osborne J. W. Best Practices in Exploratory Factor Analysis: Four Recommendations for Getting the Most From Your Analysis. Practical Assessment, Research & Evaluation. - 2005. - July; - 10 (7): p. - 1-9.
