## Computation of Tourist Satisfaction Index

By including formative measures, a component-based approach known as partial least square was used to estimate the sectoral-level models using the SmartPLS software program (Ringle, Wende, & Will, 2005). The tourist satisfaction index at the sectoral level is first computed using the model-implied factor loadings ωη31, ωη32 and ωη33, representing the weights of the three tourist satisfaction indicators: overall satisfaction (y31), comparison with expectations (y32) and comparison with ideal (y33), respectively (Song, van der Veen, Li, & Chen, 2012). The formula for calculating the sectoral-level tourist satisfaction indices is as follows:

The tourist satisfaction index of a particular service sector equals the weighted average of its three tourist satisfaction indicators’ mean values multiplied by a scaling constant of 10. Thus, each tourist satisfaction index is scaled on a comparable range of 0–100. Essentially, the higher the tourists’ average scoring on the satisfaction indicators, the higher the sectoral-level tourist satisfaction index. Subsequently the overall tourist satisfaction index is aggregated based on the six sectoral-level tourist satisfaction indices ( TSI1, TSI2, TSI3, TSI4, TSI5 and TSI6 ) and the six factor loadings (γ1, γ2, γ3, γ4, γ5 and γ6 ) which are derived from the aggregation model. The overall tourist satisfaction index is calculated by using the formula as follows:

## Computation of Tourism Service Quality Index

The approach for computing the tourism service quality index is to aggregate the means of the two indicators of tourism service quality weighted by the factor loadings of the indicators. This approach, which was originally proposed by Chan et al. (2003), has been used for calculating the PolyU Tourist Satisfaction Index since 2009 (Song et al., 2012). The computation follows that PLS estimates the value of each construct by the weighted aggregate values of its indicators, where the weights are the factor loadings for reflective indicators and regression coefficients for formative indicators after rescaling (Chin, 1998; Fornell & Cha, 1994). Hence, the two estimated unstandardized weights (if unstandardized measurements are used)ω'η31andω'η32of the two service quality indicators, namely service that satisfies my need ( y'31 ) and excellent service provision ( y'32 ) are used to estimate the sectoral-level tourism service quality index. The formula is as follows:

The tourism service quality index for a particular service sector equals the weighted average of the mean values of the two service quality indicators multiplied by a scaling constant of 10. Thus, each tourism service quality index is scaled on a comparable range of 0–100 (Song et al., 2012). The tourism service quality indices are computed at three levels, namely the market-sectoral level, the market/sectoral level and the destination level. The market-sectoral level remains the most basic on which the indices of the other two levels are derived. It incorporates 42 sub-indices, each representing the index of one of the six service sectors evaluated by one of the seven source markets. The market/sectoral level index incorporates seven overall indices at the market level and six overall indices at the sectoral level.

The market-level index is based on the six market-sectoral level indices. The weighting scheme of the six sectors is determined by a second-order confirmatory measurement model. The factor loadings indicate the contributions of the sectoral service quality to the overall and, hence, are adopted as the weights for obtaining the overall service quality index. Given the objective weights obtained from the second-order confirmatory factor analysis, the aggregation has a strong scientific basis, which in turn guarantees the robustness of overall service quality estimation. The overall index is computed based on sectoral-level indices using a weighting scheme that is determined through tourists’ own evaluation. Thus, the overall tourism service quality index is aggregated based on the six sectoral-level service quality indices ( TSQI1, TSQI2, TSQI3, TSQI4, TSQI5 and TSQI6 ) weighted by the corresponding factor loadings (γ'1, γ'2, γ'3, γ'4, γ'5 and γ'6 ) derived from the aggregation model:

The market-level index is based on the six market-sectoral level indices. The weighting scheme of the six sectors is determined by a second-order confirmatory measurement model. The factor loadings indicate the contributions of the sectoral service quality to the overall and, hence, are adopted as the weights for obtaining the overall service quality index. Given the objective weights obtained from the second-order confirmatory factor analysis, the aggregation has a strong scientific basis, which in turn guarantees the robustness of overall service quality estimation. The overall index is computed based on sectoral-level indices using a weighting scheme that is determined through tourists’ own evaluation. Thus, the overall tourism service quality index is aggregated based on the six sectoral-level service quality indices ( TSQI1, TSQI2, TSQI3, TSQI4, TSQI5 and TSQI6 ) weighted by the corresponding factor loadings (γ'1, γ'2, γ'3, γ'4, γ'5 and γ'6 ) derived from the aggregation model: