Exploratory factor analysis was undertaken given the study’s exploratory nature. An Eigenvalue of greater than one (1.000) was deemed to confirm the scale item as being an appropriate measure of the construct being measured. Cronbach’s alpha will also be presented as a measure of the scales’ reliability to measure a particular construct. Cronbach’s alpha values greater than α=0.60 will be considered as reliable (Hair et al. 2006).
6.7.1 Total Trust
Total trust (in the relationship manager) from the customer’s perspective was defined as being comprised of benevolence, competence and integrity. The tables in this subsection present the results of the factor analysis and reliability tests of the various dimensions and sub-dimensions of the total trust construct.
6.7.1.1 Benevolence – Interpersonal
On the interpersonal level of benevolence (i.e. customer/relationship manager level), one component was extracted with λ=3.696 having been recorded and accounts for 74 per cent of the variability (see Table I.1). All except for the last item were represented fairly equally as presented in Table I.2. The last item’s loading was 0.461 suggesting that it is not correlated with the sub-construct.
6.7.1.2 Benevolence – Organisational
Similarly, one component was extracted for the organisational benevolence sub-construct (i.e. customer/bank level) with λ=2.693 being recorded (see Table I.3). This component accounted for 67 per cent of the variability.
All except for the last item were represented almost equally as presented in Table I.4. The last item’s loading was 0.320 suggesting that it is not correlated with the sub-construct.
6.7.1.3 Emotional Intelligence – Empathy
Emotional intelligence (a sub-dimension of benevolence) was found to have two
components which both scored λ=1.991 and λ=1.199 respectively (see Table I.5). The first component was accountable for 40 per cent of the variability while the second 24 per cent. Table I.6 would suggest that the scale is only marginally multi-dimensional due to one measure which looks to have been more biased towards the second component.
6.7.1.4 Competence
One clear component was extracted for the competence construct with λ=2.472 recorded (see Table I.7). It accounted for 62 per cent of the variability and three of the four items were almost equally represented except for the last one which did not appear to be related (see Table I.8).
6.7.1.5 Integrity – Interpersonal Credibility
One clear component was extracted for this sub-construct which scored λ=3.565 and was accountable for 59 per cent of the variability as per Table I.9. Again, three of the four items were almost equally represented except for the last one, which did not appear to be related (see Table I.10).
6.7.1.6 Integrity – Organisational Credibility
However, for the organisational dimension of integrity, two components were extracted and scored λ=3.177 (53.0% variance) and λ=1.080 (18.0% variance) respectively (see Table I.11). However, the second component was only marginally greater than λ=1.000 and arguably not strong enough a result to be considered notwithstanding its extraction by SPSS. The last item in Table I.12 did not appear to be related to the first component as it appeared to be highly correlated to the second.
6.7.1.7 Reliability of the Total Trust Construct
The reliability of the total trust construct was good, α=0.881 (see Table J.1). This result was reinforced by the fact that would any of the sub-constructs (i.e. benevolence, competence
or integrity) have been removed, Cronbach’s alpha would have diminished (see Table J.1). 6.7.2 Quality Relationship
Quality relationship from the customer’s perspective was defined as being comprised of
commitment, satisfaction and trustworthiness that is trust in the relationship manager (in the interpersonal dimension) and bank (in the organisational dimension). The tables in this subsection present the results of the factor analysis and reliability tests of the various dimensions and sub-dimensions of the quality relationship construct.
6.7.2.1 Commitment – Interpersonal
One component was extracted here which resulted in λ=2.244 (See Table I.13). This component accounted for 75 per cent of the variability and the components are almost equally represented (see Table I.14).
6.7.2.2 Commitment – Organisational
Similarly, within the organisational dimension, only one component was extracted here with λ=3.559 being achieved accounting for 71 per cent of the variability (see Table I.15). All but the first item presented in Table I.16 are almost equally represented.
6.7.2.3 Commitment – Affective
As discussed in the literature review (Chapter Two) affective commitment (as distinguished from the above interpersonal/organisational dimensions) refers to the measurement of one’s emotional bonding to a bank and the relationship manager as well as their sense of belonging and identification with the same (Johnson et al. 2008). Here, one component was extracted with λ=3.501 and a variance of 70 per cent (see Table I.17). Table I.18 shows that the components are largely similar and highly correlated to the sub- construct.
6.7.2.4 Commitment – Relationship Continuity
This construct aimed to gauge the propensity of the customer to persist with the
relationship into the future. One clear component was extracted which achieved λ=2.850 and was accountable for 71 per cent of the variability (see Table I.19). All but the first item in Table I.20 are similarly loaded, however all showed good correlation to the sub- construct.
6.7.2.5 Commitment – Word of Mouth
A factor analysis was not conducted for this sub-construct has it had only two measures and therefore a factor analysis would have been of little value.
6.7.2.6 Satisfaction – Interpersonal
Two components were extracted for this dimension of satisfaction. Eigenvalues of
λ=4.820 (69.0% variance) and λ=1.012 (14.0% variance) respectively were recorded (See Table I.21). It could be argued that the second component was only marginally greater than one and that perhaps it should not be considered notwithstanding its extraction by SPSS. The component matrix (see Table I.22) would suggest that there is only one item that is not correlated to the first component.
6.7.2.7 Satisfaction – Organisational
Again, two components were extracted here with λ=6.613 and λ=1.116 respectively (see Table I.23). The first component was accountable for 60 per cent of the variability however the second only ten per cent and therefore perhaps not a strong enough
component to be considered in its own right notwithstanding its extraction by SPSS. This is further reinforced by the fact that there was only one item that stood out as not being correlated to the first component (see Table I.24).
6.7.2.8 Trustworthiness – Interpersonal
A factor analysis was not conducted for this sub-construct has it had only two measures and therefore a factor analysis would have been of little value.
6.7.2.9 Trustworthiness – Organisational
The trustworthiness sub-construct (in the organisational dimension) of commitment only had one component extracted, with λ=1.986, which accounted for 66 per cent of the variability (see Table I.25). Table I.26 shows that two of the measures loaded quite similarly while one loaded a little heavier then the others.
6.7.2.10 Reliability of the Quality Relationship Construct
Cronbach’s alpha for the quality relationship construct was recorded as α=0.774 (see Table J.2). However, if the trustworthiness sub-construct were removed, a marked improvement would be evident as α=0.916 (see Table J.2).
6.7.3 Sustainability
Three components were extracted for this construct with λ=3.027, λ=1.035 and λ=1.017 respectively (see Table I.27). However, the initial component at λ=3.027 was quite overwhelmingly the standout accounting for 43 per cent of the variability. The other two extracted components were only nominally greater than λ=1.000 so arguably they perhaps should not be considered as components in their own right notwithstanding their extraction by SPSS.
6.7.3.1 Reliability of the Sustainability Construct
Overall, this construct was reliable, α=0.742 (see Table J.3). However, its reliability could have been improved with the deletion of the following scales as identified in Table J.3: • How important is a long-term relationship as a reason for choosing and judging
financial institutions? – if deleted, α=0.778
• How important is competitive pricing when choosing your main bank? – if deleted,
α=0.758.
6.7.4 Crucial Stages – The Stages within the Relationship Life Cycle The crucial stagesof a relationship were identified as being exploration, expansion, maturity and dissolution, as developed by Dwyer et al. (1987), and as adapted by Hsieh et al. (2008) and Jap and Ganesan (2000) with the addition of a further stage being
recovery (as proposed by the researcher).
The factor analysis and reliability results for this construct are presented in the tables in Appendices I and J.
6.7.4.1 Exploration
For this stage, two components were extracted with λ=4.202 (53.0% variance) and
λ=2.203 (28.0% variance) respectively having been achieved (see Table I.29).
6.7.4.2 Expansion
Here, only one component was extracted, recording λ=2.576, which accounted for 86 per cent of the variance (see Table I.31). The items were almost equally represented (see Table I.32).
6.7.4.3 Maturity
A factor analysis was not conducted for this sub-construct has it had only two measures and therefore a factor analysis would have been of little value.
6.7.4.4 Dissolution
Only one component was extracted for this stage-of the relationship life cycle (λ=3.116), which accounted for 78 per cent of the variability (see Table I.33). Again all the items were almost equally represented (see Table I.34).
6.7.4.5 Recovery
This stage of the relationship life cycle was proposed as also being a crucial stage of the relationship life cycle. For this stage, one component was extracted in the factor analysis, which scored λ=3.176 and accounted for 79 per cent of the variability (see Table I.35). Table I.36 shows that the items are almost equally represented.
6.7.4.6 Reliability of the Crucial Stages Construct
The overall reliability of this construct appeared to be good (α=0.719, see Table J.4). However, would the exploration and recovery stages have been removed, Cronbach’s alpha would have improved slightly to α=0.754 and α=0.789 respectively (see Table J.4). 6.7.5 Identification of Value Accounts
The identification of value accounts, as per the conceptual model – see Figure 6.2, entails first defining what a value account is from a business-banking context, then identifying
the value accounts within the portfolio of customers, defining the role of the relationship manager and then implementation of the value account management strategy. The tables discussed in the subsections that follow represent the factor analyses conducted on the aforementioned constructs.
6.7.5.1 Defining Value Accounts from a Business-banking Context
Three components were extracted for this sub-construct scoring λ=5.482 (46.0%
variance), λ=2.248 (19.0% variance) and λ=1.596 (13.0% variance) respectively (see Table I.37). Table I.38 identifies that cross-loadings appear to be evident.
6.7.5.2 Role of the Relationship Manager
Here there were two components extracted from the factor analysis recording λ=6.471 (54.0% variance) and λ=1.210 (10.0% variance) respectively (see Table I.39). However, given that the second component was only marginally greater than λ=1.000 and only accounted for ten per cent of the variability, arguably it should not be considered notwithstanding its extraction by SPSS. Furthermore, only one item stands out in the second component in Table I.40 as potentially not being related to the initial component.
6.7.5.3 Reliability of the Identification of Value Accounts Construct
Due to there only being the two sub-constructs, if deleted, Cronbach’s alpha could not be determined. Therefore, the sub-constructs’ measures were all tested together collectively and this resulted in α=0.869 (see Table J.5).
There were only two items out of the 24 that if deleted would have improved Cronbach’s alpha (see Table J.5):
• Please select all the roles that you think best describe the role of the relationship manager:
• Sells you the Bank’s products/services – if deleted, α=0.875 • All of the above – if deleted, α=0.877
6.7.6 Co-Creation of Value
Due to there only being one scale item for this construct, neither a factor analysis nor a reliability test could be conducted.
6.7.7 Long-term Value-adding Relationship
Two components were extracted for this construct and recorded λ=6.098 (61.0% variance) and λ=1.076 (11.0% variance) respectively (see Table I.41). Given that the second
component is only marginally greater than λ=1.000, it is questionable whether it should be considered in its own right not withstanding its extraction by SPSS. Furthermore, only the first item in Table I.42 appears not to be related to the initial component, whereas all the others do.
6.7.7.1 Reliability of Long-term Value-adding Relationship Construct
The reliability of this construct scored favourably with α=0.925 being recorded (see Table J.6). However, there was one item (out of the 10) which if deleted would have improved Cronbach’s alpha to α=0.939 (see Table J.6):
• A long-term relationship that adds value to your business is one that is there for the long-run, and continues to be a source of value to those in the relationship
6.8 Regressions and Correlations