Customer Satisfaction Pilot Studies and Analysis
Dispersion of Responses (Continued)
Table 2: Standard Deviations for Participant ACSISAT Questions and Index
:
State
|
# of Responses
|
Satisfaction
|
Met expectations
|
Ideal
|
ACSISAT
|
State A
|
405
|
7.86 (2.71)*
|
7.45 (2.96)
|
7.51 (2.80)
|
73.21 (29.38)
|
State B
|
454
|
8.03 (2.46)
|
7.44 (2.69)
|
7.30 (2.77)
|
73.23 (26.73)
|
State C
|
417
|
7.70 (2.54)
|
7.43 (2.77)
|
7.27 (2.70)
|
71.88 (27.47)
|
State D
|
444
|
7.41 (2.54)
|
6.94 (2.9)
|
6.76 (2.79)
|
67.0 (28.42)
|
State E
|
456
|
7.69 (2.54)
|
7.41 (2.79)
|
7.33 (2.74)
|
72.08 (27.5)
|
State F
|
257
|
8.08 (2.12)
|
7.68 (2.53)
|
7.30 (2.53)
|
74.37 (24.0)
|
*Mean with standard deviation in parenthesis
State
|
# of Responses
|
Satisfaction
|
Met expectations
|
Ideal
|
ACSISAT
|
State B
|
503
|
7.00(2.47)*
|
6.28(2.70)
|
6.09 (2.59)
|
60.67 (27.01)
|
State D
|
575
|
7.51(2.22)
|
6.85(2.68)
|
6.62(2.51)
|
66.41(25.42)
|
State E
|
497
|
7.49(2.05)
|
7.13(2.41)
|
6.80(2.48)
|
68.28(23.50)
|
State F
|
557
|
7.31(2.06)
|
6.91(2.42)
|
6.42(2.35)
|
65.23(23.32)
|
*Mean with standard deviation in parenthesis
Table 2 and 3 Comments
The standard deviations of the ACSISAT for the six States indicate that
the scores are widely dispersed around the mean. This dispersion, in combination
with a graph of the data as in Figure 1, would alert management and staff
that not all customers perceive service in the same way and that additional
analysis is required to determine why responses vary widely.
Not only are standard deviations affected by the actual variability of
responses, they are also affected by the number of responses provided.
Up to a certain point, as the number of responses in a sample increases,
the standard deviation becomes smaller. After that point, the standard
deviation changes very little approximating the standard deviation of
all possible respondents. The table above raises an issue in this regard.
These standard deviations are quite large compared to the total possible
range of responses (1-100). With standard deviations approximately one
quarter the size of the possible response range, the question arises whether
these pilot studies had a sufficient sample size or whether there is simply
a wide range of opinions about services from these customers. Surveys
with larger numbers of responses would have to be considered to determine
if a larger number of responses would reduce the size of the standard
deviation.
One important consequence of this large standard deviation is that a
significant difference between a State's ACSISAT and their performance
requires an approximately 2.5 point difference between the obtained level
and the negotiated level. Stated another way, if the obtained score is
70, then 95 percent of the time you can be confident that the score from
successive samples would fall between 67.5 and 72.5.6
|
