Consumer Brand Preferences

Three Brands of Perfume

Table of Contents

Consumer Brand Preferences

Three Brands of Perfume

Executive Summary

The purpose for conducting this research over three brands of perfume was to use copy testing as a means to examine brand preferences among consumers. The study was conducted to show how consumers initially reacted to three brands, the common attributes associated with those brands, the effect the print ads had on the consumer’s brand preference, and which of the three brands had the highest overall preference.

The data for this research was obtained through an online questionnaire completed by 75 respondents, which was then used for a variety of statistical analyses. The Beauitful brand became more favorable after the print ads were shown, which Lovely and Hypnose’s ratings went down after the print ads were shown. This tells us that the print ads did not have a strong affect to consumer’s brand preference, regardless if a respondent preferred one brand over another.

In conclusion, the respondents had a higher preference of Beautiful over Lovely and Hypnose. For a more comprehensive analysis and for further discussion, please refer to the following pages of this report. 

Consumer Brand Preferences

Three Brands of Perfume

Introduction

The purpose of this research was to get an in depth analysis of consumer brand preferences and the effect the print ads had on the consumer’s brand preference. Three comparable brands of perfume (Beautiful, by Estee Lauder; Lovely, by Sarah Jessica Parker; and Hypnose, by Lancome Paris) were selected for the purpose of this analysis. The online questionnaire was used to gather data on consumer preferences for the three brands, common attributes associated with those brands, the effect the print ads had on consumer’s brand preferences, and which brand had the overall highest brand preference. Copy testing was used for the analysis of the data collected because copy testing evaluates and diagnoses the communication power of an ad. Copy testing is always of necessity in advertising and this report contains the results of copy testing for three brands of perfume. In the online questionnaire, respondents were asked to rate three brands of perfume. Next they were shown a print ad for each brand and were asked to look at each brand for 30 seconds. The respondents were then asked the same questions they were asked before the ad exposure in order to determine if their attitudes changed from before they had seen the print ads. The results from the data acquired from the online questionnaire were then examped for statistical significants in order to determine if the results could be projected in 85 or more samples out of every 100 samples taken from the same population as these 75 respondents.

This report will give a description of the methodology used to conduct the research, an analysis of the statistical tests run on the data, conclusions that were drawn from the statistical tests, and an overall summary of the findings. The report also includes the questionnaire that was distributed to respondents, as well as the frequency output for each question drawn from the respondents answers.

Consumer Brand Preferences

Three Brands of Perfume

Methodology

Online Questionnaire
The online questionnaire included the following sections:
Introduction
Pre-ad Exposure
Advertising Requirements
Post-ad Exposure
Brand Attitudes
Advertisement Attitudes
Demographics

Introduction
Before the questionnaire begins, respondents are shown a short message which explains the educational purpose behind the research being conducted and the strict confidentiality of the respondent’s answers.

Pre-ad Exposure
A constant sum scale was used in order to obtain respondent’s likelihood to purchase the three different brands of perfume. The respondents were asked to divide 10 points among the three brands of perfume according to their preferences of the brands by entering the numbers into a form field.  

Advertising Exposure
Next the respondents were shown a full-color magazine ad for each brand. They were asked to study each ad for 30 seconds each before continuing on to the next section.

Post-ad Exposure
After the respondents studied each print ad for thirty seconds each, they were then shown the same constant sum scale as they saw before the print ads were shown. Again, they were asked to distribute 10 points among the three brands of perfume according to their preferences of the brands by entering the numbers into a form field. Next, the respondents were given a series of “yes” or “no” questions about their attitudes toward each of the brands after seeing the print ads.

Brand Attitudes
Ten Likert items for each brand were presented to the respondents in this section in order to get a better understanding of how strongly they agreed or disagreed with a statement. They were asked to select a radio button ranging from “Strongly Agree” down to “Strongly Disagree.”

Advertising Attitudes
For this section respondents were shown a set of “yes” or “no” questions about items that related directly to the print ads. Respondents were shown a statement and asked to choose “yes” or “no” depending on whether or not they agreed or disagreed with the statement.  

Demographics
In the last section of the survey, respondents were asked demographic questions about themselves including age, sex, income, education, and perfume tendencies.

 
Design
The design of this research study is a pre and post advertisement experiment that was used to measure the effect of the print ads on respondents’ brand preference. For this experiment there was no control group used to measure responses because of time constraints.

Sampling
For the purpose of this study a convenience sample was used due to resource and time constraints. A minimum of 60 responses to the online survey were required to meet course objectives and to ensure a significant sample size for the statistical analysis. For this study, 75 responses were gathered and analyzed.

Data Collection
Data was collected by means of an online questionnaire that is available to view in this report under the section titled Appendix B. An email was sent out to friends and family with a link to the survey, which they were asked to forward on to their friends and family in order to obtain the minimum 60 respondents necessary to run statistical analysis. Once the respondents filled out the questionnaire and hit the submit button, their responses were sent to a Cold fusion database, which was later uploaded into an Access file. The statistical analysis were run in the program SPSS.

Consumer Brand Preferences

Three Brands of Perfume

Analysis

Consumer Brand Preferences

Three Brands of Perfume

Basic Statistics

1. Correlated (paired) t-tests

In table 1a, the mean brand index score was calculated for each of the three brands based off of the respondent’s responses to the ten Likert questions in the consumer brand preference survey. This calculation is done in order to better understand respondent’s attitudes towards the three brands.

Table 1a. Mean Brand Index Scores for Brands A, B, and C.

Table 1b. Comparison of Brand Index Scores

*P≤0.15

In table 1b, the first correlated t-test calculation is between Beautiful and Lovely. Beautiful was preferred to Lovely based on a t-ratio of .97. The second t-test calculation is between Lovely and Hypnose, with Lovely being preferred over Hypnose with a t-ratio of 1.54. In the final t-ratio calculation we can see that Beautiful was preferred over Hypnose with a t-ratio of 1.06.

2. Between Group t-test

Using a sample of 75 respondents, two between groups t-tests were created in order to determine respondents who moved up pre-to-post ad exposure and those who moved down.

Table 2a. Changes for Brand Index Score for Beautiful, by Estee Lauder

*p≤.15

Table 2b. Changes in Ad Index Score for Beautiful, by Estee Lauder

*p≤.15

After seeing the ads for Beautiful, by Estee Lauder, 45 respondents felt more favorably toward the brand than before they saw the ad, while 30 of the respondents felt less favorably towards the brand post ad exposure. Because of there is no statistical significance, we can not expect that 85 or more samples out of every 100 drawn from the same population as this sample would result in the same way.

3. Chi-Squared Significance Test: Beautiful, by Estee Lauder
   

A chi-squared test was calculated in order to determine if there is a significant relationship between up, same, and down pre/post movers on Beautiful perfume after being exposed to the ad.

Table 3a.

p≤.15*

From table 3a we can see that from the respondents that moved up in pre-post testing, 36 were above the brand index score median, while 39 were below the brand index score median. From the respondents who stayed the same during pre/post testing, 7 were above the brand index score median testing, while 6 were below the median. Looking at the end of table 3a we can see that of the respondents who moved down during pre/post testing, 7 of them had brand index scores above the median and 16 had brand index scores below the median.

4. Frequency Test for Changed Scores

Table 4a shows a frequency table for all three brands in order to analyze pre-to-post exposure changes in respondent’s attitudes for each brand advertisements.

Table 4a.

5. In order to indicate the respondent’s attitudes towards each brand, a brand index score was developed shown in table 5a. This table allows us to establish respondent’s preference of Beautiful perfume to Lovely perfume. The results show that 48 respondents prefer Lovely perfume to Beautiful perfume. 

Table 5a.

6. Simple Correlation Coefficient

In order to determine if there is any relationship between two brands, a correlation coefficient test was conducted in table 6a of Beautiful and Lovely brand index scores.

Table 6a.

*p≤.15

7. Subset Correlation Coefficient of females only

Table 7a.

*p≤.15

With a subset of only female respondents, a correlation coefficient test between Beautiful and Lovely’s brand index scores was used to determine if any relationship exists. From the table we can see that with a correlation coefficient of .169 that there is little to no correlation between Beautiful and Lovely’s brand index scores of female respondents.

Back to Analysis Home

Consumer Brand Preferences

Three Brands of Perfume

Regression Analysis

In order to determine if there was a relationship between how the respondents rated the ten Likert items on the survey and the ad change score, three linear regression analysis were conducted for the three brands of perfume. The regression analysis helps us to answer the question, “To what extent are the independent variables related to the dependent variable, and which independent variables are most important?” In each of the analysis the brand was the dependent variable and the ten Likert brand attributes were used as the independent variables.

Beautiful, by Estee Lauder

*p≤.15

The first regression analysis performed on Beautiful perfume determines the relationship between how respondents rated Beautiful perfume from the Likert items on the survey and the changes in their pre and post ad scores. In the table above we can see that with R-Squared being at 17.5%, there is little to no relationship between how respondents rated the perfume and the pre and post ad exposure. In other words, the ten Likert items are responsible for 17.5% of the variance in the pre-post ad exposure change scores. R-Squared represents the percent of variance in the dependent variable explained by the independent variables. With the F-ratio being at 1.2 it is not significant, meaning that in 85 or more samples of every 100 drawn from the same population as in this situation, we would not expect the same results.

*p≤.15

From the t score in the table above we can see that the constant term and the brand attributes Good Brand, High Quality and Attractive Bottle were all found to be significant, meaning that they help explain the variance in pre and post ad exposure score changes. This means that in 85 or more of every 100 samples drawn from the same population as this sample, we would expect to find about the same result for the constant term and the brand attributes Good brand, High Quality, and Attractive Bottle.

Lovely, by Sarah Jessica Parker

*p≤.15

The next regression analysis performed on Lovely perfume determines the relationship between how respondents rated Lovely perfume from the Likert items on the survey and the changes in their pre and post ad scores. In the table above we can see that with R-Squared being at 28.1%, there is a slight relationship between how respondents rated the perfume and the pre and post ad exposure. R-Squared represents the percent of variance in the dependent variable explained by the independent variables. With the F-ratio being at 1.2 it is not significant, meaning that in 85 or more samples of every 100 drawn from the same population as in this situation, we would not expect the same results.

*p≤.15

From the t score in the table above we can see that the brand attributes Makes Me Feel Pretty, High Quality and Attractive Bottle were all found to be significant, meaning that they help explain the variance in pre and post ad exposure score changes. This means that in 85 or more of every 100 samples drawn from the same population as this sample, we would expect to find about the same result for the constant term and the brand attributes Makes Me Feel Pretty, High Quality, and Attractive Bottle.

Hypnose, by Lancôme Paris

*p≤.15

The last regression analysis was performed on Hypnoses perfume determines the relationship between how respondents rated Hypnoses perfume from the Likert items on the survey and the changes in their pre and post ad scores. In the table above we can see that with R-Squared being at 35.4%, there is a small relationship between how respondents rated the perfume and the pre and post ad exposure. R-Squared represents the percent of variance in the dependent variable explained by the independent variables. With the F-ratio being at 1.2 it is not significant, meaning that in 85 or more samples of every 100 drawn from the same population as in this situation, we would not expect the same results.

*p≤.15

From the t score in the table above we can see that the constant term and the brand attribute Attractive Bottle were found to be significant, meaning that they help explain the variance in pre and post ad exposure score changes. This means that in 85 or more of every 100 samples drawn from the same population as this sample, we would expect to find about the same result for the constant term and the brand attribute Attractive Bottle.

Back to Analysis Home

Consumer Brand Preferences

Three Brands of Perfume

Discriminant Analysis

A discriminant analysis was conducted for Beautiful in order to determine if there was a relationship between how respondents rated ten brand attributes and if their preference for the brand moved up or down after seeing the ad. If the respondent’s score went up after seeing the ad they were classified as up movers, and if their score went down after seeing the ad they were classified as down movers. For the purpose of the discriminant analysis the changed score of the respondents was used as the dependent variable and the ten brand attributes were used as the independent variables.

From the table above we can see that the mean scores for the up and down movers are very similar. This tells us that there is no significant connection between how the respondents felt about the brand without consideration of how they felt about the ad.

Group Centroids

*p≤.15

In the group centroids table above for the two groups we can that neither the Wilks’ Lambda Score nor the Chi Squared score are considered significant, meaning that there is little relationship between how respondents perceived the brand and whether or not they liked the ad. These results mean that in 85 or more samples out of every 100 drawn from the same population as the sample, we would not expect to find the same result. The insignificant Wilks’ Lambda score also tells us that the mean of the discriminant function score are also not significant and can also not be projected to the population.

Classification Matrix

*p≤.15

79% of the group was correctly classified.

t0 = (.789 -.500)/Ö{((.789)(1-.789)/57) + ((.5)(.5)/57)}
t0 = 3.36

The classification table above tells us that 79% of the actual group cases were predicted correctly. This tells us that there is a moderately high accuracy rate when compared with the random probability of 50%. The observed t-ratio of 3.36 is statistically significant when compared with critical t value of 1.44. Since the t-ratio is significant, in 85 or more samples out of 100 drawn from the same population as this sample, we would expect to find an accuracy rate of the same magnitude.

Back to Analysis Home

Consumer Brand Preferences

Three Brands of Perfume

ANOVA/MANOVA

A two-way factorial ANOVA and multitvariant analysis, MANOVA, were conducted on the data from the 75 respondents in order to see if there were any statistically significant relationships between respondent’s gender, their brand perception changed score, as well as the interaction between the change score and their gender. For the ANOVA analysis the brand attribute “Hypnose is a good brand” was used as the dependent variable, and the two factors affecting it were the respondent’s gender and their changed score. In the MANOVA analysis the 10 Likert items were used as the dependent variables, while the respondent’s gender and changed score were the independent variables.

ANOVA TABLE

*p≤.15

In order to see if there was significance in gender, the changed score post ad exposure, and the interaction between gender and the changed score an ANOVA analysis was conducted (see table above).

MANOVA TABLE

In the table below, the Wilks' Lambda score shows that there is no significance from the respondent’s changed score, gender, or the interaction of gender and changed score. This means that in 85 or more samples out of every 100 drawn from the same population as these 75 samples, we would not expect to find mean scores of the same magnitude among respondent’s changed score. The analysis also means that we could not expect to project the mean scores to the population among gender. This would also mean that in 85 or more samples out of 100 drawn from the same population that we would not expect to find mean scores of the same magnitude among the changed sore and gender.

*p≤.15

From this analysis and looking at each brand attribute with each independent variable (gender and changed score), as well as the interaction between the two, there are no significant relationships.

Back to Analysis Home

Consumer Brand Preferences

Three Brands of Perfume

Factor Analysis

In order to place brand attitudes in to independent factors for the calculation of the brand attitude score, a factor analysis was conducted for each brand of perfume. The ten brand attributes from the survey (Likert items) were used as the dependent variables. In order to find potential statistical significance for the means between the brand attitude scores of the three brands of perfume a paired t-test was also conducted.

Table 1: Total Variance Explained

Table 2: Factor and Varimax Rotated Matrix (Beautiful)

In Table 2 above, a Factor Matrix and a Varimax Rotation Matrix were created for Beautiful perfume. There were two variable attributes in the Factor Matrix, which means that either there was a variable attribute with an absolute value greater than or equal to 0.5 in both Factor I & II, or neither Factor I or II had an absolute value greater than or equal to 0.5.

Table 3: Factor and Varimax Rotated Matrix (Lovely)

A Factor Matrix and a Varimax Rotation Matrix were created for Lovely perfume in Table 3 above. There were no variable attributes in the Factor Matrix and in the Varimax Rotated Matrix. For Lovely perfume, the good factor (“Lovely is a Good Brand”) fell on factor one.

Table 4: Factor and Varimax Rotated Matrix (Hypnose)

A Factor Matrix and a Varimax Rotation Matrix were created for Hypnose perfume, shown in Table 4 above. There were two variable attributes in the Factor Matrix, meaning that either there was a variable attribute with an absolute value greater than or equal to 0.5 in both Factors I & II or neither Factor I or II had an absolute value greater than or equal to 0.5.

Table 5: Brand Mean Attitude Scores

The mean attitude score for Beautiful was 3.5 indicating that it was the preferred brand of perfume. Lovely perfume was in a close second as the next favorite brand of perfume with a mean score of 3.4. Hypnose came in as the least preferred brand of perfume with a mean score of 3.2.

Table 6: Paired Sample t-test for Attitude Scores

*p≤.15

In Table 6 above, a paired sample t-test was conducted in order to determine if any brand attitude score was found to be significant. From the mean attitude score between Beautiful and Lovely, the calculation shows that it was found to be significant. This means that in 85 or more out of 100 samples drawn from the same population as this one, we would expect to find the same magnitude of difference between Beautiful and Lovely attitude scores.

Back to Analysis Home

Consumer Brand Preferences

Three Brands of Perfume

Conclusions

Basic Statistics

Paired t-tests:

The mean is calculated from a sample size of 75 respondents to the consumer brand preference survey. The highest possible brand index score from this calculation is 50 with the minimum being 10. From table 1a we can see that Lovely, by Sarah Jessica Parker is the most favored brand among respondents with a brand index score of 33.6. Beautiful, by Estee Lauder and Hypnose, by Lancôme Paris came in a close second and third, respectively. Next, the three brand index scores were compared using a correlated t-test to determine if there was a significant correlation.

Between Groups t-tests:

Each of the t-tests are above .15, therefore found to be significant. Because of the significance level for the three pairs, it can be concluded that, in 85 or more samples out of every 100 drawn from the same population as this sample, it would be expected that the mean brand index scores would be about the same as they are in this sample.

Chi-squared Significance Test: Beautiful:

The results of the chi-squared show a significant relationship between the movements in pre/post testing above or below the brand index score median. This means that in 85 or more samples out of every 100 samples drawn from the same population as this sample, we would expect to find the scores for Beautiful to be about the same as in this sample.

Frequency Test for Changed Scores:

The table shows that Beautiful, by Estee Lauder, had the most positive results with 36 respondents feeling more favorably towards the brand after being exposed to the ad. 23 of the respondents had less favorable feelings, and 16 respondents