1 Key drivers of satisfaction affecting attitudinal and behavioral loyalty: Combining quantitative and qualitative research methodologies Key Words: Satisfaction, Loyalty, attitudes, behavior, Structural Equation Models, Interviews Abstract The paper researches the link between satisfaction and loyalty in a B2B setting in the healthcare industry, in particular the clinical pathology area via a longitudinal study design. It suggests that attribute satisfaction predicts global satisfaction, which in turn predicts behavioral loyalty. Based on recent recommendations, the researcher operationalized attitudinal loyalty with a single-item scale in order to focus on a specific behavioral aspect of loyalty, the continuation of the usage of the service provider. Furthermore, two key moderating variables were included in the study to test the validity of the structural equation model used to investigate the satisfaction loyalty relationship. While the moderator variables were not significant, the research concludes that future researchers should consider moderator variables when conducting satisfaction loyalty research in B2B settings. In addition, the authors followed up with interviews to better understand why physicians switched to other Service providers.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Key drivers of satisfaction affecting attitudinal and behavioral loyalty: Combining
quantitative and qualitative research methodologies
Behavioral loyalty is measured by reviewing if the physician truly continued to
process with the service provider and was still engaged in an ongoing relationship with
the service provider 12 months after filling out the survey. A special staff reviewed the
ordering volume for each respondent for a period of 18 months following the receipt of
the survey response using a Shewart control chart. A Shewart control chart determines
variability in test ordering and downward trends in the sending of specimens.
The researcher observed if the volume had ceased after 12 months. The researcher
used two methods to determine if a physician had truly switched to a competitor.
First, the researcher reviewed the quality chart data for an additional 6 months to
determine if the physician had truly defected. Some of the volume is seasonally
determined and a six-month period was deemed sufficient to determine if a significant
drop in the volume, ordering pattern was due to seasonality.
Furthermore, sales people reviewed the records of the physician and made on-site
visits after 18 months to determine if and why the physician had seized to process with
the service provider. The additional, qualitative validity check ensured that the
quantitative correlation between attitudinal and absolute loyalty was valid and gave
additional actionable information to the service provider about the reasons for defection.
In regards to the timing of the measurement of behavioral loyalty, no clear standards
have evolved in the literature. For this study, and for the lifecycle of this organization, the
researcher chose the period of one year, to observe the absolute defection. The sales force
11
and the management team had direct input into this decision. By asking the physician if
he/she intended to continue processing with the service provider for another year and by
reviewing the actual records, this research could establish a clear quantitative link
between attitudinal and behavioral loyalty.
In addition, the institution instituted a process of content validity to determine if the
observed behavior loyalty/disloyalty truly reflected the measurement findings. The
information obtained by the content validation also gave valuable, actionable information
about the reasons underlying the switching to a competitor. It also allowed for identifying
why intentions did not translate into actual behavior. The research hereby addressed one
of Grisaffe’s “dozen problems with applied customer measurement” namely “believing
that customer satisfaction measurement is doing” (Grisaffe, 2004, p. 5).
Sample
The company sent 1200 surveys to the physicians’ offices. The potential for non-
response bias using the time stamp of the response could not be tested because the
surveys were sent to the company’s headquarters and time stamps of early and late
responders were not captured.
Two segmentation criteria were used to evaluate if there is potential bias of
survey responses: Physician function, and county in which physician practices.
The top seven categories of practitioner function responding to the survey are almost
the same as in the main population. No significant difference between the population
frame and the survey respondents could be found (Chi-square = 1.54, d.f. = 6, p = 0.957),
12
Research Model
As previously mentioned, structural equation modeling was used to investigate
the extent to which differences in observed satisfaction ratings translate into attitudinal
loyalty and behavioral loyalty. Structural Equation Modeling allows testing the
hypotheses simultaneously and at the same time determines to what degree the
hypothesized data reflect the moments of the actual data. In addition, it allows for the
assessment of the reliability in the more stringent form of a confirmatory factor analysis.
The output of the final model is shows in Figure 1.
13
Figure 1: Structural Equation Model
.97
TAT1
.96
TAT2.89
SO.97
CSO
.94
CSA
.97
SP
.93
ST
e3 e4 e5
Turnaround Time Specimen HandlingClient Services
.96.98.94
.57
Sat. Overall
.42
AttitudinalLoyalty
.00
BehavioralLoyalty
.62
.65.67
e8
e9
e10
.19 .25.42
.65
.07
Structural ModelChi Square = 46.012
P =.205D.F. =39
GFI =.958AGFI =.930NFI =.982TLI =.996
e1 e2 e6 e7
.97.98.98.98
.79
TAT 4
.89
e11
Legend:TAT1 = TAT Overall; TAT2 = TAT Standard Tests; TAT4 = TAT Esoteric Tests; SO = Special Handling Overall; SP = Protection of Specimen; ST = Timing of Special Handling; CSO = Client Services Overall; CSA = Client Services Accessible.
14
Overall Goodness-of-fit of the hypothesized model
Overall, the model fit the data structure well. The chi-square of the model is 46.012 with
39 degrees of freedom and p = .215. All critical fit indexes show values above the
recommended critical value of .90 (GFI = .958, AGFI = .930, NFI = .982, TLI = .996).
The individual hypotheses could therefore be evaluated within the framework of the
Structural Equation Model
Major Findings
The results of the hypothesis tests are as follows. Table 1 summarizes the unstandardized
and standardized regression coefficients, the critical ratios, and p-values of the variables
included in hypotheses one through five
1. HA1: The standardized regression path leading from turnaround time to overall
satisfaction is positive and significant (r = .186, d.f. = 39, p < .008). The null
hypothesis is rejected and thus confirming the alternative hypothesis
2. HA2: The standardized regression path leading from specimen handling to overall
satisfaction is positive and significant (r = .245, d.f. = 39, p < .001). The null
hypothesis is rejected and thus confirming the alternative hypothesis
3. HA3: The standardized regression path leading from client services to overall
satisfaction is positive and significant (r = .423, d.f. = 39, p < .001). The null
hypothesis is thus rejected and thus confirming the alternative hypothesis
4. HA4: The standardized regression path leading from overall satisfaction to
attitudinal loyalty is positive and significant (r = .65, d.f. = 39, p < .001). The null
hypothesis is thus rejected and thus confirming the alternative hypothesis.
15
5. HA5: The standardized regression path leading from attitudinal loyalty to
behavioral loyalty is not significant (r = 0.069, d.f. = 39, p > .05). The null
hypothesis failed to be rejected and thus the alternative hypothesis could not be
confirmed.
The results show that turnaround time, special handling and customer service are
attributes that are critical in determining overall satisfaction with the services provided.
These three constructs explain 57% of the variation in overall satisfaction. It is
noteworthy that the standard regression coefficient for client services is twice as large as
that of turnaround time. This is an indication that satisfaction with service quality is
highly critical and almost more important than the satisfaction with the tangible product
itself, i.e. the time to receive the final test result. Furthermore, overall satisfaction is
significantly correlated with attitudinal loyalty. The model explains 42% of the variation
in attitudinal loyalty. Based on Cohen’s tables (1985) the effect size of the relationship is
very large. Finally, the relationship between attitudinal loyalty and behavioral loyalty is
much smaller than anticipated.
Table 1: Tests of hypotheses 1 through 5: Unstandardized and Standardized coefficients
Dependent VariableIndependent Variable Estimate S.E. C.R. P Label
A Qualitative Post Mortem – Comments on the Findings
The findings of the “post mortem” discussions with the sales people were particularly
interesting. Of the total number of service switchers, 11.6% complained about service or
pricing issues. By contrast, 88.3% of the service switchers mentioned structural reasons.
In the clinical laboratory industry, the sending of specimens is often dictated by the
insurance of the patient. Patients with PPO have the choice of a clinical pathology
laboratory service provider, while the insurance company contracted by the insurance
provider binds HMO patient. Thus, it may not so much the service failure on the part of
the service provider, but the pressure of the underlying industrial organizational structure
that may have caused the weak link between attitudinal and behavioral loyalty. Future
research on the link between satisfaction and loyalty should take this finding into
account. This finding also puts into question the relatively strong link of cross-sectional
studies as opposed to longitudinal studies in regards to the strength of the satisfaction –
loyalty link. More studies should be conducted to investigate the method-method
contamination in cross-sectional study designs particularly as they relate to the link of
satisfaction and loyalty in B2B settings.
24
Limitations of the study
This study has several limitations. As mentioned above, it is a study in a specific
industry, i.e. in the medical services industry. External generalization is therefore difficult
to justify. The study is also limited in that it reviews only one key player in the industry.
This makes external generalization even more difficult. However, it does enhance the
internal validity of the finding.
In addition to the limitations in research design, the study has shown that the
original sample size may be too low for the effect size of the relationship between
attitudinal and behavioral loyalty.
Finally, the measurement instrument of satisfaction has shown some weaknesses.
Most importantly, the factorial structure changed depending on the type of statistical
analysis used and the sample size. While the exploratory factor analysis yielded that
reporting and turnaround are part of the same construct, the structural equation model
showed consistently that reporting does not load on the same factor. In addition,
turnaround time is not a stable construct across different situations. This has partly to do
with the fact that not all business units service the clients in an identical way. For
example, some business units process esoteric tests in-house, while others send them to a
referral lab. Consequently, the factorial structure of the measurement instrument was less
stable than anticipated. The instability of the factorial structure raises the question if it is
advantageous to use a standardized service quality scale rather than an industry-specific
satisfaction scale.
Any replication of this research needs to take the shortcomings of the
measurement system into consideration.
25
Implications for researchers
The findings of this study offer several implications for researchers.
First, the study supports previous research that suggests that overall satisfaction is
driven by service quality (measured as client satisfaction with problem handling) and
product quality (the turnaround time of result reporting). This study showed that the
impact of satisfaction with problem resolution provided by the call center operations on
overall satisfaction is stronger than the impact of satisfaction with the actual product.
This indicates that customers perceive value in the ability of a company to provide
effective problem resolution. The direct effect on attitudinal loyalty also indicates that the
way that companies handle problems may actually increase attitudinal loyalty.
Second, the analysis makes clear that the link between attribute satisfaction, overall
satisfaction and attitudinal satisfaction is real. Satisfaction matters in that it is an
important antecedent of attitudinal loyalty.
Third, the link between attitudinal loyalty and behavioral loyalty is weaker than
anticipated given the extensive literature research. While Hennig-Thurau and Klee (1997,
p. 739) assume a correlation between .18 and .26, and Bolton (1998) finds a correlation
that is around .30, the current study estimates the relationship to be less than .10.
As mentioned above, given the findings of meta-analysis such a divergence of
estimates is not unusual given the sampling error of correlation coefficients from study to
study (Hunter and Schmidt, 1990). There is some justification for the conclusion that
future research needs to ensure that the research design has sufficient power to detect the
expected relationship. This study suggests that samples sizes for a replication should not
be lower than 400 assuming that the effect size is small in the sense of Cohen (1988).
26
This implies sample sizes may need to be much larger than are typical in this research
field. Rather than using sample sizes between 50 and 200, the link between attitudinal
and behavioral loyalty may need investigation with sample sizes that exceed 400.
In addition, the current research opens up the question why the correlation between
attitudinal loyalty (intention) and behavioral loyalty is so low. One possibility is that
psychological factors of satisfaction play a secondary role in explaining behavioral
loyalty and defection (Reichheld 1996).
Another possibility is that additional psychological factors influence behavioral
loyalty. The marketing relationship literature, for example, puts heavy emphasis on trust
and commitment as two variables that are important in explaining attitudinal and
behavioral loyalty. The service quality literature emphasizes the importance of service
quality. Thus, future research may include satisfaction, service quality, trust and
commitment as factors that determine attitudinal and behavioral loyalty.
Alternatively, researchers need to take more closely into consideration the type of
exchange that is prevalent in their research environment. The loyalty concept was
adapted from the consumer behavior literature, which historically analyzed true
transactional exchanges. Business-to-business transactions occur via relationships that are
more formalized. Thus, the concept of loyalty may need to take into consideration the
type of discretionary behavior that the formalized relationship allows. Clients have the
option to either end the relationship altogether such as switching providers, or if they
have multiple providers, shift volume from one provider to another. Behavioral loyalty
may thus have an absolute and a relative aspect. Absolute loyalty would mean that a
relationship exists at all. In case where a company has several service providers, relative
27
loyalty means that businesses can switch volumes from one provider to another.
Anecdotal evidence in the clinical laboratory industry shows that physician tend to switch
volumes if a service provider loses a specimen. Research has not yet sufficiently analyzed
this possibility.
Management Implications
It is also possible that the strength of the correlation between attitudinal and
behavioral loyalty is a function of the level of satisfaction itself. Follow-up studies
showed that the bulk of the defection was due to external factors rather than service
failures.
The level of performance of the organization itself can be a reason why defection may
be high or low. Thus, the performance of the company in the eye of the customer itself
may moderate the link between attitudinal and behavioral loyalty.
In other words, companies that perform at a low level have two problems: On the one
hand, they have to manage defection due to external factors, and on the other hand, they
need to manage disloyalty due to service failure and dissatisfaction.
By contrast, companies that perform at a high level are able to keep those customers
that otherwise willingly defect to a competitor. Therein lays an important implication for
managers. Satisfaction may act as an “insurance policy”. While the company cannot
hedge against external, structural factors such as contractual arrangements that make
customers switch and that are out of the control of the organization, high levels of
satisfaction may ensure that existing customers do not voluntarily switch to a competitor.
28
Suggestions for Future Research
This study opens up new avenues for reviewing the attitudinal and behavioral
consequences of satisfaction:
1. Future studies should include other antecedent factors such as service quality,
trust, and commitment.
2. There is an opportunity to replicate this study in different industries such as the
financial services industry in order to assess what the correlation between
attitudinal and behavioral loyalty is. Care should be taken to measure the
impact of both absolute and relative defection and loyalty.
3. The impact of the performance of the organization on the strength of the
correlation between attitudinal and behavioral loyalty needs to be assessed. As
mentioned above, performance in the eye of the customer may be a moderating
factor.
4. This longitudinal study showed a weaker link between satisfaction and loyalty
than cross-sectional studies. The link may be strongly moderated by the
underlying structure of the industrial organization. As a result, more studies
should compare the impact of the legal and industrial environment of the
organization in their design to investigate the satisfaction-loyalty link.
5. Pressure of the underlying industrial organizational structure may cause to
weaken the link between attitudinal and behavioral loyalty. Future research on
the link between satisfaction and loyalty should take this into account.
29
Limitations and Extensions
A key limitation of this study is that this is a firm-specific study, and that additional
studies need to show if a generalization to other firms and industries are valid. As pointed
out earlier, functional relationships other than a linear relationship may explain the low
correlation between attitudinal and absolute behavioral loyalty. However, given the low
level of service related issues and the impact of external, situational factors that drive
behavior in this study, a different function will not yield a higher explained variance.
Another key limitation is the fact that relative behavioral loyalty was measured via a
survey item, which arouses the question if method variance accounts for the significant
difference between the measure of absolute and relative behavioral loyalty. In the
healthcare industry, firms will have to set up specific tracking systems to separate and
differentiate between HMO and PPO specimen volumes.
The findings appear to be contrasting Keaveny’s (1995, p. 78) findings that only 6%
of switching behavior is explained by factors “beyond the control of either the customer
or the service provider”. However, Keaveny’s research was conducted in a B2C setting,
which in its ideal form is characterized by consumers with full decision-making power
who can freely decide about individual transactions or contractual obligations.
The research suggests that future research should be conducted to identify the degree
to which external factors determine the strength of the link between satisfaction and
switching behavior. This research suggests that the type of relationship (contractual vs.
other types of relationships) and the decision-making power within the relationship are
important factors that moderate the strength of the link between satisfaction and loyalty.
In addition, this research suggests that loyal behavior can be viewed as two-dimensional:
30
Absolute behavioral loyalty occurs when a customer or client switches suppliers
altogether. Relative behavioral loyalty occurs when a customer or client moves volumes
from one supplier to another without abandoning the relationship altogether.
Thus, the link between satisfaction and loyalty needs to take into consideration an
internal and an external dimension.
The internal dimension analyses the nature of the relationship between satisfaction
and loyalty. In this context, questions about the linearity vs. non-linearity of the
satisfaction – loyalty link, thresholds of satisfaction, or customer segmentations play a
key role.
The external dimension takes into consideration the type of relationship between the
customer and supplier. This external condition sets the foundation for the internal
dimension and determines if and to what degree the internal dynamics of the satisfaction
– loyalty work. It is suggested that part of the lack of the empirical validation of the
satisfaction – loyalty in B2B settings can be explained by the fact that not enough
attention has been paid to the interplay between external and internal dimensions of that
link.
31
References
Anderson, E. and Sullivan, W. (1993). The Antecedents and Consequences of CustomerSatisfaction for Firms. Marketing Science, 12 (2), 125 – 143.
Bolton, R. N. (1998). A Dynamic Model of the Duration for the Customer’sRelationship with a Continuous Service Provider: The Role of Satisfaction. Marketing Science, 17 (1), 45 – 65.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates. Hillsdale, New Jersey.
Cronin, Joseph J. and Brady, M.K. (2000). Assessing the Effects of Quality, Value andCustomer Satisfaction on Consumer Behavioral Intentions in Service Environments. Journal of Retailing, Vol. 76, Issue 2. 193 – 218.
Fornell. C., Johnson, M. ., Anderson, E.M., Cha, J. and Bryant, B.E. (1996), The American Customer Satisfaction Index: Nature, Purposes and Findings, Journal of Marketing, 58, 1 – 19.
Grisaffe, D. (2001). Loyalty – Attitude, Behavior, and Good Science: A Third Take OnThe Neal-Brandt Debate. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 14, 55 - 59
Grisaffe, D. (2004). A Dozen Problems with Applied Customer Measurement.Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior.
17, 1 – 15
Iseline, S. (2007) Re: Amendments to 114.3CMR 20.00: Clinical Services.http://www.clinicalabs.org/documents/ACLACommentstoMAMedicaid011007.pdf. (Retrieved on 12/10/08).
Hennig-Thurau, Th. and Klee, A. (1997). The Impact of Customer Satisfaction andRelationship Quality on Customer Retention: A Critical Reassessment and Model Development. Psychology and Marketing, 14 (8), 737 – 764.
Hoyle, R. H. (1995). Structural Equation Modeling: Concepts, Issues and Applications. Sage Publications: Thousand Oaks.
Hunter, J.E. and Schmidt, F.L. (1990). Methods of Meta-Analysis: Correcting Error andBias in Research Findings. Thousand Oaks: Sage Publication.
Keaveny, S.M. (1995). Customer Switching Behavior in Service Industries: AnExploratory Study. Journal of Marketing. Vol. 59 (April 1995). 71 – 82.
32
Lawrence, R.J. (1969). Patterns of Buyer Behavior: Time for a New Approach? Journal of Marketing Research, 7, 137 – 144.
Mittal, V. and Wagner, A.K. (2001): Satisfaction, Repurchase Intent, and RepurchaseBehavior: Investigating the Moderating Effect of Customer Characteristics. Journal of Marketing Research. Vol. XXXVIII (February 2001). 131 – 142.
Pessemier, E.A. (1959). A New Way to Determine Buying Decisions. Journal of Marketing, 23, 41 – 46.
Reichheld, F. F. (1996): The Loyalty Effect: The Hidden Forces Behind Growth, ProfitsAnd Lasting Value. Boston, MA: Harvard Business School Press. Bain Company, Inc.
Reichheld, F. P. and Sasser, W.E. (1990), “Zero Defections: Quality Comes to Service”, Harvard Business Review, 68, September-October), 105 – 111.
Rigdon, E. (1995). “A necessary and sufficient identification rule for structural equation models estimated in practice” Multivariate Behavioral Research, 30, 359-383
The Institute of the Future (2003). Health and Healthcare Care 2010: The Forecast, the Challenge. Princeton, N.J.: The Robert Wood Johnson Foundation.
Tucker, W.T. (1964). The Development of Brand Loyalty. Journal of Marketing Research, 1, 32 – 35.
Zeithaml, V. A., Berry, Leonard L., and Parasuraman A. (1996): The BehavioralConsequences of Service Quality. Journal of Marketing, 60, April, 31 – 46.