Chapter 4 Results
4.2. Results of Survey Study
4.2.1 Factor analysis of the questionnaire
In this section the study results are presented. Firstly, factor analysis was used to create summed scales and their reliability was consequently tested. The variables´
correlation between scale items follow. Hypotheses were tested by techniques for exploring relationships among variables, by MANOVA and T-test. When possible, same methods were used while comparing values cross-nationally or alter locally, Mann Whitney U-Test was conducted as the non-parametric test option.
Factor analysis is performed in order to find similarities between set of variables. It helps to identify correlating factors and their shared variances. In this study the factor analysis was conducted in order to determine number of items forming dimensions for data analysis. It was preliminary recommended that four following dimensions help explaining the data variance. Nevertheless, conducting factor analysis was necessary to verify this supposal. Additionally, since some dimension included rather high number of items (i.e. 5 statements in case of Opinion on the AGENT´s competence and behavior), the purpose of the analysis was to determine whether smaller number of items may explain the data variance.
At first, data needed to be assessed for the factor analysis suitability. In this regard, Bartlett’s test of sphericity and Kaiser-Meyer-Olkin measure of sampling adequacy, both generated by PASW program, are those determining the appropriateness of factor analysis. For data being considered as suitable, Bartlett’s test of sphericity should be significant (p < 0.05) and Kaiser-Meyer-Olkin measure ranging from 0 to 1 should reach value of .6 in minimum. (Pallant, 2005). In this study Bartlett’s test of sphericity was of P= .000 significance and Kaiser-Meyer-Olkin measure reached .810
54
suggesting the factor analysis appropriateness.
In total, 15 items were selected for the principal component analysis (PCA). PCA recommended the suitability of 4 items all exceeding Eigenvalue of 1 explaining 62 % of variance. Eigenvalues reached values of 4.46; 2.35; 1.44 and 1.08. Next, the Varimax rotation was performed in order to enable their interpretation. Consequently, Oblimin rotation was run to determine whether another alterlocal to the component grouping exists. Oblimin rotation offered fairly same solution as the Varimax. Finally, 12 variables in total were divided into 3 components. 3 items (Other AGENT would serve me better - S8, wonder why the AGENT was employed - S9 and call center located abroad - S14) were deleted due to two reasons – loading in multiple factors and additionally their contribution to variance explanation was relatively low (less than .300). The summated scales as recommended by factor analysis are presented in the table 4.
The factor analysis recommended to group components slightly differently than it was proposed in previous chapters (Perceived service quality - 4 items, opinion on the AGENT´s competence and behavior - 5 items, opinion on the AGENT´s helpfulness and politeness - 3 items, opinion on the AGENT´s trustworthiness - 2 items). The analysis grouped together 2 items (opinion on AGENT´s politeness - S6, AGENT did her best to help me - S7) originally intended to measure helpfulness and 2 items supposed to determine competence (satisfaction with AGENT´s behavior - S4, perception of AGENT´s competence - S5). The scale was renamed to Attitude since it included four items forming attitude towards customer (behavior, qualification for the task, politeness and helpfulness). The next cluster contained statements related to competence and ability to handle call. The third component was identical to the
55
originally recommended 4 items for service quality perception measuring; thus, it has confirmed that statements are well-connected and form a solid group. Therefore, scales were reorganized as follows:
Table 4. Results of factor analysis of 12 items
Attitude Competence Service Quality Perception S6 In my opinion Customer Care person was
polite .865
S4 I am satisfied with behavior of Customer Care
person towards me .812
S5 I think that Customer Care person was
competent .755
S7 I wonder why call center employed exactly
this person .676
S10 I felt annoyed by need to repeat my sentences .806 S12 I wonder why Customer Care person
complained about poor lines when I heard her well
.689 S13 I think that poor phone connection was just
an excuse for her not really understanding me .682 S11 There was a high probability of
misunderstanding during this call .625
S2 The solution Customer Care person offered
was customer-friendly .795
S1 I am satisfied with the way my complaint was
handled .751
S15 My overall impression from the service is
good .654
S3 I will purchase from this company also in the
future .542
56
(1) Attitude
I am satisfied with behavior of Customer Care person towards me (S4) I think that Customer Care person was competent (S5)
In my opinion Customer Care person was polite (S6)
I believe that Customer Care person did her best to help me (S7)
(2) Competence
I felt annoyed by need to repeat my sentences (S10)
There was a high probability of misunderstanding during this call (S11)
I wonder why Customer Care person complained about poor lines when I heard her well (S12)
I think that poor phone connection was just an excuse for her not really understanding me (S13)
(3) Service quality perception
I am satisfied with the way my complaint was handled (S1)
The solution Customer Care person offered was customer-friendly (S2) I will purchase from this company also in the future (S3)
My overall impression from the service is good (S15)
Based on the results from Factor analysis items were grouped into clusters. The internal consistency of scales is one of the most important aspects of the study´s reliability (Pallant, 2005). In this regard, Cronbach’s Alpha is the most suitable indicator. For scale to be considered as reliable Cronbach’s Alpha should be above .7 (Hair et al., 2010)
57
Table 5 presents overview of reliability statistics for each scale. All scales reached values above .7. They are considered as internally consistent and therefore, according to Cronbach´s Alpha, reliable. The four items of Attitude have received the highest value of .838. This suggests good internal consistency of items. Nevertheless, values for Competence and Perceived service quality components .742 and .763 respectively suggest a good consistency as well.
Table 5. Correlation coefficients and reliability of factors
Attitude Competence Perceived Service Quality
Attitude .838
Competence -.193* .742
Perceived Service Quality
.312** -0.13 .763
Value in the diagonal is Cronbach’s
* p < 0.05 ** p < 0.01 (2-tailed)
4.2.2 The correlation between factors ‘components
The factor analysis has identified 3 groups within variables used in the research. Next, the analysis was performed to further determine relationships and correlations between components. The value of Pearson correlation (r), varying between 0 and 1, is important indicator of strength of variables´ relationship (Pallant, 2005).
Firstly, the analysis was conducted for the factors of Attitude and Service quality perception. The test showed positive small and medium correlation. All correlation
coefficients are significant, suggesting that the items are suitably grouped. This implies that items of these two clusters are positively related, thus, the positive value for variable in one cluster suggests that as a consequence also variables in the second
58
cluster are positive. Additionally, the relationships between all items in two clusters are significant, thus their relationship may be considered as rather strong.
Table 6. Correlations between Attitude and Service quality perception
S1 S2 S3 S15
S4 .409** .212** .299** .426**
S5 .436** .231** .206** .398**
S6 .313** .203** .185** .334**
S7 .355** .276** .315** .399**
** p < 0.01 (2-tailed)
Next, the correlation between Competence and Attitude was examined. However, with the one exception (S6 and S11) negative correlations may be seen from the table 7.
Negative correlations mentions to the case when high results of one variable links with low results of the other (Pallant, 2005). This implies that items do not measure the same aspect. As for instance high values on satisfaction with AGENT´s behavior (S4) shows that this item is not positively connected to AGENT complaining about poor phone connection (S13).
Table 7. Correlations between Attitude and Competence
S10 S11 S12 S13
S4 -.228** -.053 -.335** -.267**
S5 -.125 -.079 -.298** -.212**
S6 -.126 .032 -.302** -.184**
S7 -.224** -.127 -.321** -.235**
** p < 0.01 level (2-tailed)
59
Lastly, the correlation scores between Service quality perception (satisfaction with complaint handling - S1, solution was customer-friendly - S2, likeliness of future purchase - S3, overall good service impression - S15) and Competence (need to repeat sentences - S10, high probability of misunderstanding - S11, complaining about poor connection - S12, poor connection as excuse - S13) are presented. In case of all variables correlation reached negative values, thus variables correlate negatively (high scores for one variable imply low scores for the other one and vice versa). Similarly to table 15, also these values point out on the negative relationship among variables in two categories. High values for S13 (poor connection as excuse) correlate negatively with the items of Service quality perception (likeliness of future purchase - S3 and overall good service impression - S15). To conclude, the fact that AGENT complains about the poor quality of phone connection indicates low values for service quality satisfaction. The output of the correlation analysis is presented in the table 8.
Table 8. Correlations between Competence and Service quality perception S1 S2 S3 S15
S10 -.117 -.035 -.299** -.205**
S11 -.174* -.062 -.123 -.102
S12 -.114 -.076 -.181** -.198**
S13 -.079 -.023 -.140* -.144*
* p < 0.05 ** p<.01 (2-tailed)
The most significantly correlating items may be found between the groups of Attitude and Service quality perception where all correlations are significant at the 0.01.
Instead, Competence and Service quality perception link somewhat negatively and plus Pearson correlation is small (-0.10 to- 0.29).
60
4.2.3 The hypotheses testing
It was recommended that when “interacting with non-local speaker, customers are more likely to suppose that call center is located outside of country” (H1). This hypothesis has been tested by t-test, because it enables to examine means of two different groups (Pallant, 2005). The answer of the item S14 (I think that call center is located outside of the country) were examined. The t-test recommended rejecting null hypothesis (for both local and non-local speaker customers equally suppose that service is located abroad). Therefore, when interacting with non-local speaker customers (mean=4.42) more likely supposed that call center is located outside of the country (mean=3.82). There was a significant difference (p<.05) between the location of the call center when interacting with local and non-local speaker (p=0.009).
Next, the testing of trustworthiness perception level for local and non-local speaker was conducted. H2 was proposed as follows: Customers perceive non-local Customer Care representative as less trustworthy compared to local speaker. Nevertheless, the null H2 was tested (Customers perceive non-local Customer Care representative on the same level of trustworthiness compared to local speaker). Hypothesis was tested by t-test measuring the differences in means for the reversed item S13 “I think that poor phone connection was just an excuse for the not really understanding me”.
According to the means, local speaker was considered as more trustworthy. People were more likely to trust local speaker (mean = 4.23) that she does not hear what they say whilst when interacting with non-local speaker (mean = 3.44) customers predominantly supposed that she does not understand them but at the same time does not admit it loudly. There was a significant difference (p<.05) between trustworthiness when interacting with local and non-local speaker (p=0.003). Therefore, the null
61
hypothesis was rejected and H2 was supported.
It was proposed that customers perceive lower service quality when interacting with non-local Customer Care representative (H3). For the purpose of analysis null hypothesis was recommended “customers do not perceive different service quality when interacting with non-local Customer Care representative”. Therefore, the influence of service quality perception factors (satisfaction with complaint handling - S1, solution was customer friendly - S2, likeliness of future purchase - S3, overall good service impression - S15) on the local or non-local speaker version was examined. Hypothesis was tested with MANOVA. There was no significant difference in the service quality perception of encounter provided by local or non-local speaker.
Tests of between subjects’ effects confirmed non-significant values for items.
Therefore, the null hypothesis has not been rejected. The summary of the results can be seen in the table 9. S1 - satisfaction with complaint handling
S2 - solution was customer-friendly S3 - likeliness of future purchase S15 - overall good service impression
62
It has been proposed that non-local Customer Care representative is perceived as less competent compared to local speaker (H4). For the purpose of analysis, null hypothesis supposing that there are no differences in perceived competence was tested.
The questionnaire included the item designed for measuring the perceived level of competence – S5 “I think that Customer Care person was competent”. There was a significant difference (p<.05) between perception of local and non-local speaker´s competence (p=0.039). According to the means, local speaker (mean = 4.42) was considered as more competent than non-local speaker (mean = 3.97). Thus, null hypothesis was rejected.
T-test was performed to assess 5th hypothesis (Customers perceive higher probability of misunderstanding when interacting with non-local speakers). At first, null hypothesis (Customers do not perceive higher probability of misunderstanding when interacting with non-local speakers) was recommended. T-test compared means for item S11 “There was a high probability of misunderstanding during this call.” There was no significant difference (p<.05) between perception of probability of misunderstanding when interacting with local (mean = 3.51) and non-local speaker (mean = 3.17) (p=0.127). Based on this, null hypothesis was retained.
63
4.2.4 Cross – national comparison
One of the aims of this study was to compare the service quality perception in one concrete situation between two nations. For this purposes, Taiwanese formed group 1 while those marking Tagalog as their mother tongue were grouped into cluster called Filipinos. MANOVA was performed to determine the influence of nationality on the service quality perception (satisfaction with complaint handling - S1, solution was customer-friendly - S2, likeliness of future purchase - S3, overall good service impression - S15). The test showed that there is a significant difference of the service quality perception for 3 out of 4 items. The only item being of non-significant values was S2 (the solution Customer Care person offered was customer-friendly).
MANOVA pointed out on the significant differences in service quality perception between nations. The results are presented in table 10.
Table 10. Cross-national comparison in service quality perception
Nationality Mean p-Value S1 - satisfaction with complaint handling
S2 - solution was customer-friendly S3 - likeliness of future purchase S15 - overall good service impression
Next, nationalities were compared in the regard of their opinion of AGENT´s attitude (satisfaction with AGENT´s behavior - S4, perception of AGENT´s competence - S5, opinion on AGENT´s politeness - S6, AGENT did her best to help me - S7).
64
MANOVA was run to find whether there is a significant difference in perception of AGENT´s attitude. The test´s output is presented below. As seen from the table 11 the significant variance was found only in case of one item, S7 (AGENT did her best to help me). All other items retained null hypothesis.
Item S7 was proposed as follows: I believe that Customer Care person did her best to help me. Comparing the means, it can be deducted that difference is noticeable.
Filipinos considered AGENT as helpful while Taiwanese believed she could have done more in order to provide higher service quality (Means were 5.47 for Filipinos and 4.69 for Taiwanese).
Table 11. Cross-national comparison for opinion about AGENT’s attitude
Item Null Hypothesis p-value Result
S4 Taiwanese and Filipinos perceive the same level of satisfaction with AGENT’s behavior
.340 Retained H0
S5 Taiwanese and Filipinos perceive the same level of satisfaction with AGENT’s competence
.961 Retained H0
S6 Taiwanese and Filipinos perceive the same level of satisfaction with AGENT’s politeness
.223 Retained H0
S7 Taiwanese and Filipinos believe that AGENT did her best to help them
.000 Rejected H0
S4 - satisfaction with AGENT´s behavior S5 - perception of AGENT´s competence S6 - opinion on AGENT´s politeness S7 - AGENT did her best to help me
65
Table 12. Cross-national comparison for opinion about AGENT’s competence
Item Null Hypothesis Significance Result
S10 Taiwanese and Filipinos perceive the need for repeating sentences as equally annoying
.000 Rejected H0
S11 Taiwanese and Filipinos perceive equally high probability of misunderstanding
.093 Retained H0
S12 Taiwanese and Filipinos equally
wondered why AGENT
complained about poor connection
.894 Retained H0
S13 Taiwanese and Filipinos equally perceived poor connection as an excuse for not understanding
.000 Rejected H0
S10 - need to repeat sentences
S11 - high probability of misunderstanding S12 - complaining about poor connection S13 - poor connection as excuse
Finally, nationalities were compared in the regards of their opinion of AGENT´s competence (need to repeat sentences - S10, high probability of misunderstanding - S11, complaining about poor connection - S12, poor connection as excuse - S13).
MANOVA was run to find whether there is a significant difference in perception of AGENT´s attitude.
66
Table 13. Hypotheses testing summary
Hypothesis Results
H1-Interacting with non-local speaker, customers are more likely to suppose that call center is located outside of country
Supported
H2-Customers perceive non-local Customer Care representative as less trustworthy compared to local speaker
Supported
Comparing the means for items S10 (I felt annoyed by need to repeat my sentences) and S13 (I think that poor phone connection was just an excuse for her not really understanding me) revealed that Taiwanese perceived as very annoying that they had to repeat sentences (means 5.01 and 4.23 for Taiwanese and Filipinos respectively).
Additionally, Taiwanese were more likely to believe that the true reason for need to repeat a sentence was that AGENT did not understand and not the poor phone connection as she stated (means 4.61 and 3.67 for Taiwanese and Filipinos respectively).
In this section descriptive statistics together with data analysis was presented. In Table 13 the testing for hypotheses is summarized. The data served as the basis for comparison between two nations, Taiwanese and Filipinos. The comprehensive discussion of results is to be found in the next chapter.
67