Examining the Relationship between Psychological Empowerment, Relationship Quality and Online Word of Mouth in Brand Facebook Page

2.1 Psychological empowerment

Consumer empowerment refers to the notion of transferring the balance of power from businesses that are normally in control to the habitually powerless consumers (Hoffman, Novak and Peralta, 1999). In the current study, psychological empowerment is defined as perceived individual control of expressing feeling and/or thought in the online environment. Thomas and Velthouse (1990) proposed that individual psychological empowerment is exhibited through four cognitions namely meaning, competence, self-determination and impact. Meaning refers to the value of one’s work objective or drive. Competence is one’s belief in his/her ability to perform the work. Self-determination is an individual’s autonomy in having choice in regulating his/her work, behaviors and processes. Impact is the level to which an individual thinks his/her actions can influence work results or other people behaviors. The advancement of social networking sites has contributed to an increase in consumer empowerment (Constantinides, 2007), specifically the power to express their feeling and thought. For example, the SNS would allow consumers to give opinion and share their experiences with others. This gives them meaningful purpose of helping others within the online community. Also, it would be easy for them to speak their mind using SNS, as they have the ability and confidence to use such technology. They perceived they have the autonomy to express their feelings without any major restrictions. By having such power, they believe they could exert influence to organization and other people.

2.2 Relationship quality

Fournier (1998) defined brand relationship quality as an overall evaluation of the depth and strength between consumer and brand, a construct that is multi-aspect in nature. Although the indicators differ depending on each study’s particular setting, researchers commonly conceptualize brand relationship quality as a construct composed of three key components namely satisfaction, trust, and commitment (e.g. Baker, Simpson & Siguaw, 1999; Dorsch, Swanson & Kelly, 1998; Garbarino & Johnson, 1999; Smith, 1998).

2.3 Online word of mouth

WOM is increasingly crucial for brand marketers due to its effect on purchase decision by consumers (Gajendra et al., 2012; Soares et al., 2012). Park and Lee (2008) referred online word of mouth as any positive or negative statement made by potential, actual, or former customers about a product or business, which is made available to a multitude of people via the Internet. It is an informal communication between individuals related to the usage or characteristics of particular goods and services, or their sellers that is uniquely different from communications through mass media Goldsmith (2006). This study focuses on positive online word of mouth, which is defined as the act of spreading favorable statements that relate to a brand’s product or service.

Conceptual framework

Based on the suggestion put forward by Ulaga and Eggert (2006), the current study applied Fishbein and Ajzen’s (1975) cognition, affection and conation theory. This provides a theoretical guidance in examining whether members’ cognitive perceptions of psychological empowerment are related to their emotional states of satisfaction, trust, and commitment that eventually lead to their word of mouth behaviors.

3.2 Hypotheses development

Sampling and data collection

Brand Facebook page was selected as the appropriate representative of brand social networking sites due to its popularity worldwide. In particular, 10 most popular hospitality brands in Malaysia Top 200 brand Facebook page were chosen as the research setting. Currently, there are 18 brands related to the hospitality industry in the Malaysia Top 200 brand Facebook page with estimated total members of 10,228,884. The study did not include the remaining 8 hospitality brands, as from our observation they were considered inactive. Based on the definitions of hospitality (see Ditmer, 2002; Lashley & Lee-Ross, 2008), the hospitality brand was categorized into 4 sectors: foodservice, transportation, entertainment, and lodging.

Based on a confidence interval of 95%, and an error margin of +5% the minimum sample size was established at 384 (Krejcie & Morgan, 1970). However, Bryman’s (2008) suggested that increasing the sample size could reduce sampling error. Thus, a sample size of 500 was considered appropriate for this study. Based on the total number of fans, the four categories of hospitality brand Facebook pages are proportionally stratify to determine the number of samples for each category. As a result of systematic random sampling, the sample size of 500 for the study is represented by 238 members for foodservice, 170 for transportation, 62 for entertainment, and 30 for lodging.

A web survey on eSurv.org was used to collect the data. Three waves of invitation link were sent to the selected samples via their Facebook. After one week, a total of 271 responses were received out of the 500 sent to the target respondents. However, 36 questionnaires were invalid and were excluded from further data statistical analysis. Thus, total valid questionnaires were 235, representing 47 percent response rate. Based on the sample size calculation using G*Power analysis (Faul et al., 2009), 235 respondents were adequate to achieve statistical test power of more than 0.80 for this study. This was grounded on 0.05 significance level and effect size of 0.15.

Measures

All items were rated on a 5-point agreement scale anchored by 1 = strongly disagree and 5 = strongly agree except online word of mouth, which was rated on a 7-point frequency scale (1 = strongly disagree and 7 = strongly agree). Psychological empowerment was assessed with a 4-item scale adapted from (Spreitzer, 1995). Relationship satisfaction was measured using a 4-item scale derived from (Lin, Wu & Tsai, 2005). Relationship commitment was measured using a 4-item scale adapted from (Turri, Smith & Kemp, 2013). Relationship trust was assessed with a 4-item scale derived from (Johnson & Grayson, 2005). Online word of mouth, the consequences of relationship quality was measured using a 3-item scale adapted from (Jahn & Kunz, 2012). All items were slightly modified to suit the context of the current study. Additionally, 7-item social desirability by Fisher and Fick (1993) was included to address common method variance.

4.3 Data analysis

Partial Least squares (PLS) a second-generation multivariate technique was used to test the developed research model (Fornell and Cha, 1994), which can simultaneously evaluate the measurement model (the relationships between constructs and their corresponding indicators), and the structural model with the aim of minimizing the error variance (Chin, 1998; Gil-Garcia, 2008). Smart PLS Version 3.0 (Ringle et al., 2005) was used to analyze the data. Also following the suggestions of (Chin, 1998; Gil-Garcia, 2008), bootstrapping method (5000 resampling) was applied to determine the significance levels for loadings, weights, and path coefficients.

5.1 Descriptive statistics

The descriptive analysis indicated that 48.5 per cent of the respondents were male and the remaining female. In terms of age, majority was in the age between 18 – 24 years (75.70%). Most of the respondents surveyed were Malay (67.2%), followed by Chinese (17.4%), Indian (2.1%), and others (13.2%). In relation to education level, a majority of the respondents (77.9%) have either Diploma or Bachelor’s degree. Most of the respondents are active members of the hospitality brand Facebook page with 41.3 per cent of them visited or browsed the hospitality brand Facebook page at least twice a month.

5.2 Testing the measurement model

As suggested by Hair et al. (2010), factor loadings, composite reliability and average variance extracted to assess convergence validity. The recommended values for loadings are set at > 0.7, the average variance extracted (AVE) should be > 0.5 and the composite reliability (CR) should be > 0.7. Based on previous study (e.g. Zhang & Bartol, 2010), psychological empowerment is measured as a second order reflective constructs. Thus we followed the method suggested in the literature in PLS which is the repeated indicator approach to model the second order factors in the PLS analysis. From table 1 it can be seen that the results of the measurement model exceeded the recommended values thus indicating sufficient convergence validity. Next, discriminant validity is assessed using the Fornell and Larcker (1981) method. The criterion used to assess this is by comparing the AVE with the squared correlations or the square root of the AVE with correlations (refer Table 2 ), The criteria is that if the square root of the AVE, shown in the diagonals are greater than the values in the row and columns on that particular construct than it is concluded that the measures have discriminant validity. From Table 2 , it can be seen that the values in the diagonals are greater than the values in their respective row and column thus indicating the measures used in this study are distinct. Thus the results presented in Tables 1 and 2 demonstrate adequate discriminant and convergent validity for the measurement model.

5.3 Testing the structural model

Using a bootstrapping technique with a re-sampling of 5000, the path estimates and t-statistics were calculated for the hypothesized relationships. It was found that psychological empowerment benefit ( = 0.125, t = 2.093, p < 0.05) was positively related to relationship satisfaction. In terms of the interrelationship among the dimensions of relationship quality, it was discovered that relationship satisfaction has a positive significance effect on both relationship commitment ( = 0.223, t = 2.417, p < 0.01) and relationship trust ( = 0.651, t = 11.483, p < 0.01). Also, it was found that relationship trust has a significant effect on relationship commitment ( = 0.448, t = 4.338, p < 0.01). Focusing on relationship commitment as an endogenous construct, the standardized beta () value indicated that relationship trust ( = 0.448) has more impact on the endogenous construct as compared to the relationship satisfaction ( = 0.223). Based on the results of R^2, both relationship satisfaction and relationship trust explained 38 per cent variance for relationship commitment. Additionally, Q² was satisfactory at 0.293. In relations to the correlation between the dimensions of relationship quality and online word of mouth, it is found that relationship satisfaction ( = 0.166, t = 2.12, p < 0.05), relationship trust ( = 0.163, t = 1.863, p < 0.05), and relationship commitment ( = 0.257, t = 3.349, p < 0.01) all have a positive significant effect on online word of mouth. Thus, based on the two-stage PLS analysis, all 7 hypotheses (H1 – H7) were supported (refer Table 3 ).