relationship management (independent variables) and the outcome should be considerably weaker than the main effects of predictor and criterion variables. For mediation analysis, in step 1, we entered control variable (qualification). In second step, the mediator customer trust was entered. In the third step, customer relationship management was entered in equation and was regressed on switching intention. Hypothesis 4: customer trust is mediating the effect of customer relation management on switching intention. Customer trust is mediating the effect of customer relation management on switching intention.. Hypothesis 5: customer trust is mediating the effect of customer relation management on switching intention. To test the mediating effect of customer trust, we regressed switching intention, customer trust and customer relationship management together as supported by the study. As shown in table 3, results of multiple regression revealed significant reduction in variances (from ß =-.13* to -.08 and ∆R² = .02*, to ∆R² = .00). These result confirmed full mediation condition according to the research study providing support of our hypothesis 4, hypothesis5 which states that customer trust mediates the relationship between customer relationship management and switching intention. TABLE 3 RESULTS OF MEDIATOR REGRESSION ANALYSIS SWITCHING INTENTION PREDICTORS Β R² ∆R² MODEL 1 STEP 1

17 QUALIFICATION .00 STEP 2 CUSTOMER RELATIONSHIP MANAGEMENT -.13* .02 .02* MODEL 2 MEDIATION OF CUSTOMER TRUST STEP 1 QUALIFICATION -.045 .00 STEP 2 CUSTOMER TRUST -.10 02 .02 STEP 3 CUSTOMER RELATIONSHIP MANAGEMENT -.08 .027 .00 N = 354 (QUALIFICATION WAS USED AS CONTOL VALUABLE) ***p< .001, **p< .005, *p< .05 M ODERATION A NALYSIS Earlier in the study we predicted that word of mouth moderates the relationship between customer relationship management and customer trust. According to Baron and Kenny (1986), moderation can be established with three regression tests. First customer relationship management (independent variable) should be related to customer trust (dependent variable in this case). Second, customer relationship management and moderator (word of mouth) should be related. Third when customer relationship management (independent variables) and word of mouth (moderator) are concurrently incorporated in regression equation, then the relationship between customer relationship management (independent variables) and the outcome should be considerably weaker than the main effects of predictor and criterion variables.

18 For moderation analysis, in step 1, we entered control variable (qualification). In second step, the moderator word of mouth along with independent variable customer relationship management was entered. In the third step, customer relationship management and word of mouth were linked and an interaction term was entered in equation and was regressed on customer trust.