analysis

basic statistics

Paired t-Tests

Table 1: Descriptive Statistics of Brand Index Scores for each Brand

Table 1 shows the mean Brand Index Scores for the three brands under consideration.  The Brand Index Score is the total of ten judgments made about each brand measured on a Likert scale.

Table 2: Paired t-test for Mean Brand Index Scores

Three Paired t-Tests were run to compare the mean Brand Index Scores of the brands. Pairs one and three were proven to be statistically significant at the alpha level of 0.15. In 85 or more samples out of every 100 samples drawn from the same population as this sample, the mean scores for Stolichnaya and Belvedere and for Belvedere and Ketel One would be about what they are in this sample. The means from these two pairings are statistically different and their values can be accurately projected to the population.

Between Group t-Test

Table 3: Means of the Advertising and Brand Index Scores for Stolichnaya vs. the Respondents Change in Purchase Intention after Viewing the Ad

Respondents were asked to score their likelihood of purchasing each of the brands on a constant sum scale both before and after viewing an ad for each brand. Those respondents who recorded an increased purchase intention post-viewing are labeled as ‘up movers’ those who recorded a lower likelihood of purchase are ‘down movers.’ As Table 3 shows there were 19 ‘up movers’ and 26 ‘down movers’ for Stolichnaya. Advertising Index Scores were calculated from the number of positive attributes out of a possible 12 were selected by the respondent. The mean of which was 7.3 for ‘up movers’ and 4.5 for ‘down movers.’

Table 4: Between Group t-Tests for Stolichnaya Brand and Advertising Index Score

Between Group t-Tests were used to find out if the differences between the mean Advertising Index Score and Brand Index Score for ‘up movers’ and ‘down movers’ were statistically significant. It was found that in a one-tailed test at an alpha level of 0.15 the results were significant. Therefore, in 85 out of 100 samples drawn from the same population as this sample was drawn it would be expected that the mean Brand and Advertising Index Scores for that population would be approximately what they are for this sample, as shown in Table 3.

Chi Squared Test

Table 5: Relationship between Stolichnaya Purchase Intent and Brand Index Score

Table 6: Chi Squared Significance Test

A Chi-Squared Test was used to determine if there is a statistically significant relationship between whether a respondent’s intention to purchase Stolichnaya went up, went down or stayed the same and whether the same respondent’s Brand Index Score for Stolichnaya was above or below the median Stolichnaya Brand Index Score of 31.  No statistically significant relationship was found at a 0.15 alpha level, therefore the sample results cannot be effectively projected to the population.

Frequency Count

Table 7: Purchase Intention Change by Brand

The respondent’s purchase intention was measured both before and after exposure to the ad for each brand. The number of respondents whose’ scores moved up, down or stayed the same for each brand are recorded in Table 7. If a respondent recorded that they were more likely to buy Stolichnaya after seeing the ad then they were recorded in the ‘up’ group for Stolichnaya, 19 respondents fell in this category. Belvedere Vodka was the only brand for which more of the respondents’ purchase intentions went up. While Stolichnaya and Ketel One both had more respondents score move down after viewing their respective ads.

Brand Preference Comparison

Table 8: Stolichnaya and Belvedere’s Brand Index Scores Relative to Each Other

26 out of the 68 total respondents, or 38.2%, gave Stolichnaya a higher Brand Index Score than they gave to Belvedere. This was calculated by subtracting the Brand Index Score of Belvedere from that of Stolichnaya. This result is in line the information found in the previous question, which showed that more respondent’s changed their purchase intentions changed positively for Belvedere than for Stolichnaya.

Pearson Correlation Test

Table 9: Correlation between Brand Index Scores for Stolichnaya and Belvedere

A Pearson Correlation Test was run to identify the presence of a connection between the Brand Index Scores of Stolichnaya and Belvedere. The result is statistically significant at the alpha level of 0.15 and at -0.3 is a low inverse correlation. This means that Stolichnaya’s Brand Index Score has a weak, inverse effect on the Belvedere score; as the Stolichnaya score rises the Belvedere score falls.

Brand Preference Comparison Among Female Respondents

Table 10:Stolichnaya and Belvedere’s Brand Index Scores Relative to Each Other Among Females Only

11 respondents out of a total of 43, or 25.6%, gave Stolichnaya a higher Brand Index Score than they gave to Belvedere. These results make it appear that fewer females in this sample prefer Stolichnaya to Belvedere than the total sample, which was analyzed in question five. This would also imply that male respondents score Stolichnaya more positively on the Brand Index Score than both females and the total sample.

next

multiple regression

Linear Regression

Table 1: Multiple Linear Regression for Stolichnaya, Belvedere and Ketel One

Table 2: Stolichnaya Brand Attributes

The brand attributes of Stolichnaya as taken from the 10 Likert questions account for a very low amount (12.5%) of the variance in the pre and post ad-viewing scores, this finding cannot be predicted to the population. Only one attribute - Stolichnaya has a nice bottle - was found to be significant and in 85 out of 100 samples taken from the same population as this sample the coefficient is expected to be about what they are here.

Table 3: Belvedere Brand Attributes

The brand attributes of Belvedere Vodka account for a very small amount (15.0%) of the variance in the pre and post ad-viewing scores, but this cannot be predicted to the population. The two brand attributes, Belvedere has and nice bottle and Belvedere is sophisticated, are statistically significant therefore, in 85 out of 100 samples taken from the same population as this sample it would be expected that these two coefficients would be the same as they are here.

Table 3: Ketel One Brand Attributes

The brand attributes of Ketel One account for only a small amount (33.1%) of the variance in pre and post ad-viewing scores, the results can be predicted to the population. Four of the brand attributes for Ketel One –has a nice bottle, is preferred over other vodkas, makes good mixed drinks, and is not a brand that I would buy- have coefficients which are statistically significant. In 85 out of 100 samples taken from the same population as this sample, it would be expected that these four coefficients would be about what they are here.

Ketel One Equation

Change Score Ketel One=(0.0 + 0.1)+(0.6 + 1.5)+(0.1 + 0.0)+(0.8 + 0.3)+(0.6 + 0.3)+(0.9 + 0.3)+(0.9 + 0.4)+(0.3 + 0.1)+(0.5 + 0.2)+(0.4 + 0.1)

next

discriminant analysis

Table 1: The Mean Score for each Brand Attribute within the Up or Down Groups

Table 2: Wilks Lambda and Chi Square

The Wilks Lambda test, when converted to a Chi Squared test found that he Group Centroids, listed in Table Three, were statisiticaly significant at the alpha level of 0.15. The value of the Group Centroids can be projected to the population; meaning that in 85 out of every 100 samples taken from the same population as this sample the value would be expected to be about what it is here.

Table 3: Group Centroids

Table 4: Unstandardized Discriminant Function Coefficients

The Discrimant Function Coefficients indicate the relative importance of each independent variable, in this case each LIkert item to the grouping variable, in this case the up or down movers. The most important attributes have been bolded in the table. Belvedere’s “bottle” and its perceived “high quality” have the strongest coefficients at 1.0. Belvedere “fits my lifestyle”, ‘is preferred”, and “is recommended” are also influential.

Table  5: Classification Matrix

Classification Accuracy is 70.6%

               
t ratio =  2.50*

t critical = 1.04
*µ= 0.15

The t test proves the classification accuracy of 70.6% to be statistically significant. For 85 out of 100 samples taken from the same population as this sample it is expected that about 70% of the respondents would be grouped correctly. The result can therefore be predicted to the population.

next

ANOVA/MANOVA

Table 1: Descriptive Statistics

Table 2: F-test Results

The F Test resulted in a significant coefficient meaning that in 85 out of 100 samples taken from the same population as this sample it is expected that the results would be about equal to what they are here. The differences in means between gender and change score movement (up, same, down) are significant for each independent variable. The interaction between the independent variables also produces a significant result, therefore the means can be predicted to the population.

MANOVA
Table 3: MANOVA Descriptive Statistics

Table 4: MANOVA Multivariate Statistics

For the two independent variables acting individually the means are significant and can be predicted to the population. In 85 out of 100 samples, taken from the same population as this sample, the results are expected to be approximately what they are here. The interaction between the gender and change score does not have a significant effect on the means and cannot be predicted to the population.

next

factor analysis

Stolichnaya

Table 1: Communalities of each Likert Questions

Table 2: Eigen Values

Table 3: Extraction

Factor one explains almost 46% of the variance of the ten Likert items regarding Stolichnaya vodka. Factor two explains just under 12% of the variance, together they account for 57% of the variance in the responses.

Table 4: Factor Matrix

Table 5: Varimax Rotated Factor Matrix

The bolded component scores are the important ones, and show which factor the variable ‘loads on.’ The un-rotated matrix was used for calculating the attitude score as it is less ambiguous than the rotated version. The initial Likert item which stated “Stolichnaya is a good brand of vodka” is used as the identifying variable. In this case 8 additional variables load on the same factor as the ‘good’ Likert item.

Belvedere

Table 6: Communalities of each Likert Questions

Table 7: Eigen Values

Table 8: Extraction

Factor one explains almost 41% of the variance of the ten Likert items regarding Belvedere vodka. Factor two explains just under 16% of the variance, together they account for 56% of the variance in the responses.

Table 9: Factor Matrix

Table 10: Varimax Rotated Factor Matrix

The bolded component scores are the important ones, and show which factor the variable ‘loads on.’ The Varimax rotated matrix was used for calculating the attitude score as it is less ambiguous than the rotated version. Five variables ‘load’ on variable one, the same variable as the initial ‘good’ Likert item.

Ketel One

Table 11: Communalities of each Likert Questions

Table 12: Eigen Values

Table 13: Extraction

Factor one explains almost 51% of the variance of the ten Likert items regarding Ketel One vodka. Factor two explains just under 13% of the variance, together they account for 64% of the variance in the responses.

Table 14: Factor Matrix

Table 15: Varimax Rotated Factor Matrix

The bolded component scores are the important ones, and show which factor the variable ‘loads on.’ The un-rotated matrix was used for calculating the attitude score as it is less ambiguous than the rotated version. A total of eight variables load on factor one, the factor which the ‘good’ Likert item ‘loads on’.

The factors which were determined to be a positive attitude toward the brand were used to compute an attitude score for each brand which were then compared to the others using a paired t-test.

Table 16: Descriptive Statistics for Attitude Scores

Table 17: Two Sample Paired t-test

The differences between the mean attitude scores of Stolichnaya and Belvedere, and Stolichnaya and Ketel One are statistically significant, therefore in 85 out of 100 samples taken from the same population as this sample the difference would be approximately what it is here. The results from these two pairs can be predicted to the population. The difference between the attitude scores of Belvedere and Ketel One are not statistically significant at the alpha level 0.15 and the means cannot be predicted to the population.

next

cluster analysis

Table 1: Subject Distribution between Clusters

Table 2: ANOVA

The 68 total subjects are split between the cluster 1 and 2 fairly evenly, with 39 and 29 subjects respectively.  The ANOVA resulted in significant differences between cluster 1 and 2 for all LIkert items, excluding “Stolichnaya is just like any other vodka”

Table 3: Descriptive Statistics for each LIkert item by Cluster