TM_2016_Authenticity and place attachment of major visitor attractions

張貼日期：2016/4/12 上午 02:57:54

Authenticity and place attachment of major visitor attractions 

Tourism Management  Volume 52, February 2016, Pages 110–122 

http://www.sciencedirect.com/science/article/pii/S026151771500134X

The indicators reliabilities were found as acceptable as all indicators, but one, met the preferred level of 0.7 or higher (Wong, 2013). However, in exploratory stages a loading of 0.4 or higher is acceptable (Hulland, 1999). The Composite Reliability (CR) measurements demonstrated a high level of internal consistency in both latent variables, exceeding the recommended threshold of 0.6 (Chin, 1998 and Wong, 2013). The evaluation of Average Variance Extracted (AVE) confirmed the convergent validity, when found as higher than 0.5 in both latent variables (Chin, 1998 and Wong, 2013). Two approaches were used to assess the discriminate validity of the measurement model. First, an examination of the indicators' cross loading confirmed that no indicator loads higher on the opposing construct (Chin, 1998). Second, the Fornell and Larcker (1981)criterion was applied by confirming that the square root of AVE in each latent variable is larger than other correlation values among the latent variables.

The descriptive statistics (means and standard deviations) of the study's variables of place attachment and authenticity are presented together with their F scores (Table 1). Results show significant differences between attractions in most items. In order to test the study hypotheses, two mean scores were calculated, representing measures of place attachment (Mean = 4.277, SD = 1.149) and authenticity (Mean = 5.010, SD = 1.155). In both cases, F scores were significant and indicated on significant differences between the attractions with regard to place attachment (F(3,368) = 3.820, p < 0.01) and authenticity (F(3,367) = 44.744, p < 0.01).

Evaluation of  the Measurement Model

In addition, an examination of the indicators' cross loading revealed that no indicator loads higher on an opposing construct (Hair et al., 2011)