Cluster Analysis
Two cluster analyses, a two-group and a three-group, were conducted for the Dasani brand using the ten Likert items as variables to find groups of respondents that answered survey questions similarly. Descriptive statistics, F-ratios (to determine the significance of the means), and Chi-Squared/crosstab results were reported for each cluster.
TWO CLUSTERS:
Table 1A: Descriptive Statistics
Brand Attribute (Likert Item) |
Cluster Number of Case |
|||
|
Cluster 1 (n=41) |
Cluster 2 (n=41) |
||
|
Mean |
Std. Dev. |
Mean |
Std. Dev. |
…good brand … |
3.6 |
0.8 |
3.8 |
0.7 |
…makes me feel healthy. |
3.1 |
1.0 |
3.3 |
0.8 |
…tastes great. |
3.3 |
0.8 |
3.4 |
0.9 |
…brand I trust. |
3.5 |
0.9 |
3.7 |
0.9 |
…fits my lifestyle. |
3.3 |
0.8 |
3.4 |
0.8 |
…is too expensive. |
2.7 |
0.9 |
2.6 |
0.8 |
…is just like any other… |
3.2 |
1.0 |
3.1 |
0.9 |
…would recommend to my friends. |
3.0 |
0.7 |
3.2 |
0.8 |
…prefer over regular water. |
2.6 |
1.0 |
2.9 |
1.0 |
…doesn't taste artificial. |
3.2 |
0.9 |
2.8 |
0.7 |
Total sample size = 82
Table 1A displays the means and standard deviations of all ten Likert items when respondents are divided into two groups. As the table shows, the number of respondents is evenly distributed between cluster 1 (41) and cluster 2 (41). The mean scores reveal that overall Cluster 2 rated, and liked, the brand better.
Table 1B: ANOVA/F-ratios
Brand Attribute (Likert Item) |
F-ratio |
…good brand |
1.24 |
…makes me feel healthy. |
1.47 |
…tastes great. |
0.16 |
…brand I trust. |
0.59 |
…fits my lifestyle. |
0.99 |
…is too expensive. |
0.30 |
…is just like any other… |
0.00 |
…would recommend to my friends. |
1.67 |
…prefer over regular water. |
1.75 |
…doesn't taste artificial. |
4.91* |
* p≤0.15 |
|
Table 1B shows that only one Likert item (…doesn't taste artificial) is statistically significant. This Likert item differs significantly between groups, which means that the results can be projected to a larger population.
Table 1C: Chi-Squared Significance Test
Cluster Number |
Gender |
||
|
|
Male |
Female |
1 |
Count |
15 |
26 |
Row % |
53.6% |
48.1% |
|
Column % |
36.6% |
63.4% |
|
Total % |
18.3% |
31.7% |
|
2 |
Count |
13 |
28 |
Row % |
46.4% |
51.9% |
|
Column % |
31.7% |
68.3% |
|
Total % |
15.9% |
34.1% |
|
Chi-Squared = 0.22 (* p≤0.15)
Sample size = 82
A Chi-Squared Significance Test was conducted to determine the relationship between cluster membership and gender for the Dasani brand. Table 1C shows the number of males and females in each cluster with Cluster 1and 2 being clearly female-oriented. These results are not statistically significant and cannot be projected to a larger population. In 85 (or more) of 100 samples drawn from the same population as this sample, we would not expect to find the same results.
THREE CLUSTERS:
Table 2A: Descriptive Statistics
Brand Attribute (Likert Item) |
Cluster Number of Case |
|||||
|
Cluster 1 (n=27) |
Cluster 2 (n=28) |
Cluster 3 (n=27) |
|||
|
Mean |
Std. Dev. |
Mean |
Std. Dev. |
Mean |
Std. Dev. |
…good brand |
3.6 |
0.9 |
3.7 |
0.8 |
3.9 |
0.7 |
…makes me feel healthy. |
3.0 |
1.2 |
3.2 |
0.8 |
3.4 |
0.7 |
…tastes great. |
3.3 |
0.8 |
3.5 |
0.7 |
3.4 |
1.0 |
…brand I trust. |
3.5 |
0.9 |
3.7 |
0.8 |
3.6 |
0.8 |
…fits my lifestyle. |
3.2 |
0.8 |
3.4 |
0.7 |
3.4 |
0.8 |
…is too expensive. |
2.7 |
0.8 |
2.6 |
0.8 |
2.6 |
0.8 |
…is just like any other… |
3.1 |
1.0 |
3.1 |
0.9 |
3.2 |
0.9 |
…would recommend to my friends. |
2.9 |
0.6 |
3.3 |
0.9 |
3.1 |
0.8 |
…prefer over regular water. |
2.6 |
1.0 |
3.0 |
1.1 |
2.6 |
1.0 |
…doesn't taste artificial. |
3.2 |
0.9 |
2.8 |
0.7 |
2.9 |
0.8 |
Total sample size = 82
Note: A 3-group cluster analysis would actually not be necessary anymore, since the results of the 2-group cluster analysis already showed a perfect distribution of respondents between cluster 1 and 2.
Table 2A displays the means and standard deviations of all ten Likert items when respondents are divided into three groups. Overall cluster 1 had the lowest mean scores and cluster 2 and 3 are very similar with the mean scores varying slightly for certain attributes.
Table 2B: ANOVA/F-ratios
Brand Attribute (Likert Item) |
F-ratio |
…good brand |
0.94 |
…makes me feel healthy. |
1.65 |
…tastes great. |
0.58 |
…brand I trust. |
0.36 |
…fits my lifestyle. |
0.60 |
…is too expensive. |
0.35 |
…is just like any other… |
0.14 |
…would recommend to my friends. |
1.22 |
…prefer over regular water. |
1.57 |
…doesn't taste artificial. |
1.93 |
* p≤0.15 |
|
As table 2B shows, none of the ten Likert items is statistically significant. The results do not differ significantly between groups and thus cannot be projected to a larger population.
Table 2C: Chi-Squared Significance Test
Cluster Number |
Gender |
||
|
Male |
Female |
|
1 |
Count |
10 |
17 |
Row % |
35.7% |
31.5% |
|
Column % |
37.0% |
63.0% |
|
Total % |
12.2% |
20.7% |
|
2 |
Count |
10 |
18 |
Row % |
35.7% |
33.3% |
|
Column % |
35.7% |
64.3% |
|
Total % |
12.2% |
22.0% |
|
3 |
Count |
8 |
19 |
Row % |
28.6% |
35.2% |
|
Column % |
29.6% |
70.4% |
|
Total % |
9.8% |
23.2% |
|
Chi-Squared = 0.38 (* p≤0.15), Sample size = 82
A Chi-Squared Significance Test was conducted to determine the relationship between cluster membership and gender for the Dasani brand. Table 2C shows the number of males and females in each cluster. All 3 clusters are clearly female-oriented. The findings are not statistically significant and cannot be projected to a larger population.