CONSUMER PREFERENCE ANALYSIS
THREE BRANDS OF SNACK/GRANOLA BARS
The following is a comprehensive consumer preference analysis which includes copy testing research for Quaker, Nature Valley and Kudos snack/granola bars. Several statistical techniques were used for the analysis of the survey results and will be discussed in detail on the following pages.
This web site is a course requirement of Professor Leckenby's Summer 2007 Advertising 380J graduate seminar Quantitative and Qualitative Research at the University of Texas at Austin.
From the 86 respondents, it appears that in terms of brand index score, or consumer preference, all three brands are close in favorability. However, we would expect to find a more favorable brand perception of Quaker when compared with Kudos and Nature Valley and Nature Valley with Kudos. This would indicate that while respondents may slightly prefer Quaker, the do not show any particular loyalty or preference to either brand, indicating that consumer brand switching may easily take place.
Furthermore, a relationship does exist between the up and down-movers and the ten brand characteristics for Kudos. Therefore, if we know the attitudes someone in this population has toward Kudos (as defined by the Likert items on the questionnaire), we can predict with 91.9% accuracy whether or not they will like the ad (as indicated by a positive pre-post ad exposure change score). This allows ad developers to create ads that will appeal to consumers and therefore create more persuasive advertisements, increasing consumer appeal and preference for Kudos snack/granola bars. A more detailed report of the results is available in the Analysis section.
Several research techniques have been developed to predict or reveal the success of an ad both before and after its production. One such technique, copy testing, serves as a means of measuring the effectiveness of advertising messages. Copy testing is a common method of testing the effectiveness of an advertisement before it is released to the general public.
Copy testing involves the analysis of potential and specific effects an advertisement might have on consumer behaviors and/or brand preferences. In this process, researchers seek out general consumers and show them the advertisements. After viewing the ads, respondents are often questioned about their feelings towards the product.
Copy testing can reduce the risk associated with producing new advertising by providing quantitative measures of consumer response to many aspects of the brand and advertising message. The research can reveal whether or not it is actually effective in persuading a sample of that population and ultimately the population as a whole.
This study aims to determine whether a sample of 86 respondents, presumed over the age of 18, liked or disliked three brands of snack/granola bars and their particular print ads, which were then used to test the effects of print advertisements on the respondents' brand preferences. The brands used in this study are: Quaker, Nature Valley and Kudos. The three print ads for each brand were found in various popular lifestyle magazines, including "O" and "Shape" magazines.
Data was collected through a web-based questionnaire that was fully completed by 86 individuals. Once all surveys were returned, the data was imported into analytical software SPSS, through which various statistical analyses were performed. These statistical techniques were completed in order to determine initial preferences for the three snack/granola bars and to aid in assessing the effect each brand’s print ad had on consumer preference and perception.
The results from the questionnaire and the statistical approaches conducted on the data will help examine common attributes associated with the three brands and compare which brand had the most favorable attitudes and which gained the preference of the majority. The results will be examined for statistical significance, revealing whether or not the finding can be projected onto the general population based on this sample of 86 respondents.
Following this introduction, sections dedicated to methodology, analyses of all the statistical tests, conclusions drawn from these statistical analyses, and a summary of the findings. Two appendices are included to provide examples of the actual questionnaire (Appendix B), and a sample of the questionnaire exhibiting summary statistics associated with each question (Appendix A).
Basic Structure of the Questionnaire
The questionnaire developed for this study was created for the purpose of analyzing three snack/granola brands: Quaker, Nature Valley and Kudos. This survey was created with the Dreamweaver Software and the layout was designed specifically for online data collection. Through clickable links, the questionnaire was created to appear as a multi-page layout, but in actuality was only one page. Arranged in eight separate sections, some with multiple parts, the questionnaire featured a total of 68 composed of different response elements: text fields, text areas, radio buttons, list options or drop-down menus, and check boxes.
Welcome Page: Respondents are greeted with an initial welcome page. The welcome page is designed to introduce the survey and motivate participants to complete it. It also states the survey’s educational intent and states that all responses will be held strictly confidential.
Pre-ad Exposure:
Section One: The purpose of this section is to measure purchase intent change after ad exposure on pre-post constant sum scale. The respondents were asked to name a brand they recently purchased in the following three categories: candy bars, chips and snack/granola bars. This question attempts to determine which brands are "top of mind" to the respondent and if the respondent demonstrates loyalty to any of them. The questionnaire is yet reveal which brands the survey will focus on in attempts to minimize biases when collecting data.
Section Two: The respondents are then asked to indicate purchase intent for three specific brands in each category by dividing 10 points between the brands according to their likelihood of purchase. They were restricted to using only whole numbers so as to discourage them from using decimals or fractions. The same questions were asked of the two other product categories used in the preceding question in order to familiarize respondents with the constant-sum scale format. The values from this scale are eventually used to analyze pre-ad and post-ad exposure changes.
Advertisement Exposure:
Section Three: Respondents are then asked to view three print ads, asking that the respondent should only take 15 seconds to review each of the three ads. The respondent is asked to continue on with the survey without clicking back to the ads, noting that it is not necessary for the respondent to remember the ads exactly. The short timing method of ad exposure is an attempt to duplicate the same viewing environment as if flipping through a magazine.
Post-ad Exposure:
Section Four: After viewing the ads, the respondents are asked to once again indicate purchase intent for each of the three snack/granola bars (Quaker, Nature Valley and Kudos) for which they saw an ad. Following the same instructions featured in previous sections, respondents were asked to divide 10 points between the brands according to their likelihood of purchase. This "post-exposure" question allows for comparison of pre-ad and post-exposure scores in order to obtain a change score for each snack/granola bar brand.
Section Five: This section has four questions, three that are optional based on the responses to the first one. The first question determines if viewing the ad made the respondent feel differently toward any of the three brands. If there is no difference, the respondent is asked to skip questions 5b – 5d. If the respondent marks yes, he or she is then asked which brand he or she felt most differently towards, whether the impact was favorable or unfavorable, and to describe the attributes of the ad that made he or she feel that way.
Brand Attitudes:
Section Six (Questions 6a – 6c): Divided into three sections of questions, one per brand, a Likert scale is used to measure attitudes, preferences, and subjective reactions. The questions reflect common attributes (both negative and positive) associated with each snack/granola bar brand on an individual basis. The respondent is asked to choose only one of the following answers for each attribute: Strongly Agree, Agree, Neither Agree nor Disagree, Disagree, or Strongly Disagree. Each radio button was coded as a numeric value for later statistical analyses.
Advertisement Attitudes:
Section Seven: This section asks the respondent to agree or disagree with 10 descriptive adjectives (both negative and positive associations) related to the print advertisements they viewed. The same ten descriptive adjectives were posed for each of the three snack/granola bar brands. By selecting a checkbox, respondents indicate an "agree” response, while an unchecked box indicates a "disagree" response. These responses are numerically coded for quantitative analysis.
Demographics:
Section Eight: The last section is to examine if any demographic and lifestyle information is relevant to how the respondents react to the ads. The final questions gather basic demographic (the respondents' gender, age group, and highest level of completed education) and relevant lifestyle information (how often they purchase snack/granola bars).
After all questions were completed, participants are then asked to click the “Submit” button at the bottom of the page to securely and officially submit their questionnaire.
Design
This study can be considered a survey. Due to lack of a control group or environment this research should be classified as conducting a pseudo-experiment. Instead of having a control group, comparison is made between before and after advertising exposure responses, using the sample of 86 respondents. The effects of the print advertising will be measured by subtracting Pre-Ad Exposure Purchase Intention from Post-Ad Exposure Purchase Intention (the method for calculating the change score). Movement on the constant-sum scale will be used to determine any differences in brand preferences.
Sampling Description
Due to time and monetary constraints, this study opted for a non-random, convenient sample. It was decided that a sample size of at least 60 respondents would be sufficient enough to perform statistical analysis, which also permits enough representation so that if statistical significance is found, projection can be made in 85 or more samples of 100 samples from the same population as the sample used in the questionnaire.
Respondents were recruited through an email among family and friends of the researcher. A “snowball effect” occurred, with recipients of the email being asked to forward it on to any individuals they believed would be willing to complete the questionnaire. The e-mail contained the link to the questionnaire, requesting participation in the research project. All respondents provided agreement to participate by clicking the "Submit" button. The required 60 respondents completed this survey.
Data Collection Process
All the data in this study was collected electronically from June 13, 2007 to June 18, 2007. When the questionnaire was created using the Dreamweaver Software, each answer was given a field name and a coded number (if applicable) so that the numbered data could be quantitatively used in a statistical program for analysis. The questionnaire utilized text fields (to gather written comments), text areas, radio buttons, list options or drop-down menus, and check boxes so that respondents could select the appropriate answers as easily and efficiently as possible.
After subjects completed the survey and clicked the “Submit” button, their answers and corresponding field names were immediately stored in an Access database file in the Center for Interactive Advertising, linked via a ColdFusion file. Due to this method of online data collection, error was minimized as the respondents' exact answers were transferred directly into the database and no manual coding was needed. The data received and retrieved was exported into the analytical software system, SPSS, in order to perform various statistical analyses.
1. Paired t-tests
Brand Index Scores
Paired t-Tests
Brand
Mean Brand Index Score
Standard Deviation
Paired Samples
t-ratio
Quaker
40.7
6.8
Quaker -- Nature Valley
.605
Nature Valley
40.3
6.6
Nature Valley -- Kudos
.881
Kudos
39.7
7.3
Quaker -- Kudos
1.39*
Sample size = 86 * (p) ≤ .15
Paired t-Tests: Brand Index Score:
From the 86 respondents, it appears that in terms of brand index score, or consumer preference, all three brands are close in favorability. For the comparison of Quaker Valley to Kudos, in 85 or more samples of every 100 samples drawn from the same population as this sample, we would expect to find a more favorable brand perception of Quaker Chewy Granola Bars than Kudos and Nature Valley and Nature Valley with Kudos.
Each brand index score was correlated with another to determine if results were significant and could therefore be projected onto the population represented by the sample. In the first correlated t-test, Quaker was paired with Nature Valley. Quaker was preferred slightly to Nature based on a t score of .605. The second correlated t-test compared Nature Valley with Kudos. Nature Valley was preferred slightly over Kudos based on the t- ratio of .881. The third correlated t-test compared Quaker to Kudos, where Quaker was preferred to Kudos. Quaker was preferred slightly over Kudos with a t-ratio of 1.39. The results of the third t-test were found to be significant. Thus we can conclude that when Quaker was paired with Kudos, in 85 or more samples out of every 100 samples drawn from the same population as this sample, we would expect to find mean score for Quaker to be about the same as shown in this sample.
2. Between-Groups-t-Tests: Quaker
Brand Index Score: Between-Groups-t-Test
Pre-Post Change ScoreNumber of People
Mean Brand Index Score
Standard Deviation
t-ratio
Moved up
5
36.4
15.0
1.441*
Moved down
32
28.5
3.8
* (p) ≤ .15
Ad Index Score: Between-Groups t-Test
Pre-Post Change ScoreNumber of People
Mean Bran Index Score
Standard Deviation
t-ratio
Moved up
5
20.2
3.5
.818
Moved down
32
21.4
3.1
* (p) ≤ .15
Brand and Ad Index Score by Pre-Post ad Exposure Change Score:
Using a sample of 86 respondents, two between groups t-test were conducted in order to determine how many respondents moved up (people with a more favorable perception of Quaker after seeing the Quaker ad than before viewing it) and those who moved down (people with a less favorable perception of Quaker after seeing the ad than before viewing it) pre-to-post exposure to the advertisements. The tests showed that the brand index score, or consumer preference, is significantly correlated with pre-post testing movement. Therefore, in 85 or more samples out of every 100 samples drawn from the same population as this sample, we would expect to find mean scores for upward movement and downward movement in pre-post testing for Quaker about the same as shown in this sample.
3. Chi-Squared Significance Test: Quaker
Brand Index Score Relative to Median, Pre-Post Ad Exposure Change Score
Pre-Post Change ScoreNumber of People
Brand Index Score above median (41.5)
Brand Index Score below median (41.5)
Moved Up
Count
35
31
% Row
53.0%
47.0%
% Column
89.7%
86.1%
% Total
46.7%
41.3%
Same
Count
2
2
% Row
50.0%
50.0%
% Column
5.1%
5.6%
% Total
2.7%
2.7%
Moved Down
Count
2
3
% Row
52.0%
48.0%
% Column
100.0%
100.0%
% Total
52.0%
48.0%
Chi-Squared Value
.323
Sample size = 86 Median = 41.5
* (p) ≤ .15
Brand Index Score Relative to Median, Pre-Post Ad Exposure Change Score:
This Chi-Square analysis shows the relationship between the brand index score, or consumer preference, of Quaker and how many respondents moved up or down in regards to brand preference from pre to post exposure of the Quaker ad. The majority of respondents’ preferences changed positively, with 35 respondents moving up. 2 respondents changed their brand preference more negatively towards Quaker post-exposure to the advertising and 2 remained the same after seeing the ads. There results were not significant, however, and could not be applied to the population as a whole, but merely the respondents of the sample.
4. Frequency Test for Changed Scores
(Score change of pre-to-post ad exposure on the constant-sum scale)
Brand# of People who Moved Up
# of People that stayed the Same
# of People who Moved Down
Quaker
70
4
12
Nature Valley
76
4
6
Kudos
74
8
4
This table shows how many respondents changed preferences from pre to post viewing of the ads. Here, a frequency test was conducted for all three brands of snack/granola bars to analyze pre-to-post exposure changes in respondent’s attitudes of favorability for each brand’s advertisement.
All three brands showed significant numbers in regards to respondents having a positive opinion change of the brand. Nature Valley received the most positive attitude changes, though not by much, with 76 respondents’ impressions improving after viewing the Nature Valley ad. 74 respondents improved their impressions of Kudos and 70 respondents had a better image of Quaker Valley after seeing the respective ads. Respondents had the least favorable pre-to-post exposure change scores in regards to the Quaker ad with 12 moving down. 6 respondents had a negative opinion change regarding Nature Valley and 4 reported a negative attitude change in regards to Kudos.
5. Frequency for Brand Index Scores (Consumer Preference)
BrandFrequency
Quaker
45
Nature Valley
41
A brand index score was developed to indicate respondents’ attitudes toward each brand. In an attempt to establish preference of one snack/granola bar over another, frequency tests were run with the brand index scores of Quaker and Nature Valley. Respondents preferred Quaker to Nature Valley, though but not by much with Quaker having a preference of 45 and Nature Valley at 41.
6. Simple Correlation Coefficient
BrandsCorrelation Coefficient
Quaker and Nature Valley
.423
To determine if any relationship exists among the 86 participants, a simple correlation coefficient test was conducted between Brand Index Scores, or consumer preferences, of Quaker and Nature Valley. The simple correlation coefficient between Quaker and Nature Valley is .423, which indicates a moderate correlation between the two brands. The more favorably a respondent’s opinion was of Quaker the more favorable was their response toward Nature Valley or vice versa.
7. Subset Correlation Coefficient (males only)
BrandCorrelation Coefficient
Quaker and Nature Valley
.284
To determine if any relationship exists among the subset of the 32 male respondents, a correlation coefficient test was conducted between Brand Index Scores, or consumer preferences, of Quaker and Nature Valley. The results indicate there was a low correlation (.284) between Quaker and Nature Valley’s brand preference when the consumer was male.
The Discriminant Analysis is used to examine the relationship between multiple continuous independent variables and one discrete dependent variable. Kudos brand snack/granola bars was analyzed using this method, where the independent variables used were the brand’s 10 Likert scale items from the questionnaire, and the dependent variable the respondents’ constant-sum scores (a measure of buying intention) increase or decrease measurements in order to determine how effective the 10 Likert items were at predicting whether respondents’ constant-sum scores increased or decreased.
Kudos Brand Attributes: Up-movers vs. Down-movers
UP-MOVERS
n=74
DOWN-MOVERS
n=12Standardized Discriminant Coefficient
Unstandardized Discriminant Coefficient
Brand Attribute (Likert Item)
Mean
Standard Deviation
Mean
Standard Deviation
Product is recommendable
4.0
1.1
4.6
1.1
0.7
0.6
Product makes you physically fit
3.9
1.0
4.1
1.1
0.3
0.3
Product gives you more energy
3.9
0.9
3.7
1.0
-0.5
-0.5
Product fits lifestyle
3.9
1.0
4.1
1.1
0.1
0.1
Bar is more expensive than others
4.0
0.8
3.9
0.5
-0.1
-0.2
Packaging is appealing
4.1
0.9
4.1
0.8
0.6
0.7
Consuming product makes my day better
3.8
0.9
3.9
0.9
-0.7
-0.8
Bar tastes good
4.6
1.0
4.6
0.7
-0.4
-0.4
Bar is a bad snack/granola bar
4.2
1.0
4.3
1.0
-0.3
-0.4
Bar is a better bar than others
3.7
1.0
4.3
0.6
0.6
0.6
Kudos Brand Attributes: Up-movers vs. Down-movers
Comparing the Likert items, or mean brand attribute scores fails to reveal any dramatic differences between up-movers and down-movers. The largest differences between mean scores for up-mover and down-mover groups were on the following items: the product is recommendable and the bar is better than others. These items revealed the largest differences between groups mean scores. Six of the ten brand attribute items can be considered important, meaning they best account for the differences in pre-post exposure change scores between the two groups, up-movers and down-movers based on the highest absolute value of 4.6 divided by 2 (2.3). They are: the product is recommendable, product gives you more energy, packaging is appealing, consuming the bar makes my day better, the bar tastes good and the bar is better than others.
The unstandardized discriminant coefficients suggest the relationship between each brand attribute and how much respondents liked the ad. They can be interpreted as follows:
- The more respondents think Kudos is recommendable, the more they like the ad.
- The more respondents think Kudos will help them be more physically fit, the more they like the ad.
- The more respondents think Kudos will give them more energy, the less they like the ad.
- The more respondents think Kudos fits their lifestyle, the more they like the ad.
- The more respondents think Kudos is more costly, the less they like the ad.
- The more respondents think Kudos has appealing packaging, the more they like the ad.
- The more respondents think Kudos makes their day better, the less they like the ad.
- The more respondents think Kudos snack/granola bars taste good, the less they like the ad.
- The more respondents think Kudos bars are bad snack/granola bars, the less they like the ad.
- The more respondents think Kudos is a better quality brand, the more they like the ad.
Wilks’ Lambda(Λ)
Wilks’ Lambda(Λ)
Chi – Squared (χ²)
0.80
17.6*
Functions at Group Centroids
Up - Movers
Down - Movers
-0.20
1.23
* indicates p ≤ .15
The group centroids, -0.20 and 1.13 are significant at their current levels due to the Wilks’ Lambda value of 0.80 being statistically significant based on the Chi-Squared value. This means that in 85 or more samples out of 100 samples drawn from the same population as this sample, we would expect to find group centroid values of the same magnitude and signifies that a relationship does exist between the up and down-movers and the ten brand characteristics.
Kudos Classification Matrix
Actual GroupPredicted Group
Totals
Up - Movers
Down - Movers
Up – Movers
74
0
14
Down - Movers
7
5
8
91.9% of cases correctly classified
t observed = ______ (0.919 - .5)_______
√[(0.919 x .081)/86] + [(.5 x .5)/86]
t = 7.62*
*indicates t 0 <= 1.04 p < = 0.15
The discriminant function predicts the classification of up-movers and down-movers with 91.9% accuracy. This is an excellent degree of accuracy when compared with the random probability 50%. The observed t-ratio of 7.62 is significant when compared with t value of 1.04 at the p ≤ 0.15 level. Therefore, in 85 or more samples out of 100 drawn from the same population as this sample, we would have about the same accuracy in predicting the classification of up-movers and down-movers.
Conclusion:
The results can be projected to the population, as Wilks’ Lambda and Chi-Square were significant. In other words, the discriminate analysis results based on this sample of 86 respondents can be projected to the entire population. If we know the attitudes someone in this population has toward Kudos (as defined by the Likert items on the questionnaire), we can predict with 91.9% accuracy whether or not they will like the ad (as indicated by a positive pre-post ad exposure change score).
Regression tests were conducted for three brands of snack/granola bars (Quaker, Nature Valley, and Kudos) with the brand attributes (the ten Likert items) as the independent variables and the pre-to-post change score as the dependent variable. The sample size was 86.
1. Multiple Regression Analysis: Quaker, Nature valley and Kudos
Multiple Regression (R)
Coefficient of Multiple Determination (R²)
Standard Error of Estimate (se)
F-ratio
Quaker
.3
10.1%
.5
.85
Nature Valley
.3
11.4%
.7
.97
Kudos
.5
22.1%
.5
2.13*
* indicates p ≤ .15
‘R’ measures the relationship between the variables; a score between -0.5 and 0.5 is considered non-existent. R measures the "multiple correlation coefficient," the relationship between respondents' feelings for the brand and their impression of the brand's ad. A score between -.05 and 0.5 is indicates a non-existent relationship. A strong and positive relationship is indicated by an R of 1.0, while a strong and negative relationship indicated by an R of -1.0, and a moderate relationship by an R of + 0.5.
The F-ratio was calculated in order to determine whether or not the results were significant and could therefore be projected to a larger population represented by the sample.
R-Squared, a number between 0 & 1 that measures the "coefficient of multiple determination," or the percentage of variation in the respondents' impression of the ad explained by the fit of the estimated sample regression line. The value of R-Squared ranges from 0 to 1, with low value meaning that the sample regression line is not a good fit to the collected data, and the high value indicating that it is a good fit for the data.
Quaker Nature valley Kudos
Brand Attribute (Likert Item)b
Unstandardized CoefficientBeta
Standardized Coefficientt- score
b
Unstandardized CoefficientBeta
Standardized Coefficientt-score
b
Unstandardized CoefficientBeta
Standardized Coefficientt-score
Constant
1.2
2.31*
2.6
3.66*
1.0
2.69*
Product is recommendable
-.0
-.0
-.06
.1
.2
1.12
.1
.2
1.13
Product makes you physically fit
.1
.2
-1.49
-.1
-.2
-1.52*
.1
.2
1.53*
Product gives you more energy
-.1
-.1
-.72
-.2
-.2
-1.41
-.1
-.3
-1.66*
Product fits lifestyle
-.2
-.3
-1.71*
.1
.2
.98
-.1
-.1
-.68
Bar is more expensive than others
.1
.1
.73
-.1
-.1
-.45
-.0
-.1
-.28
Packaging is appealing
.1
.1
1.07
-.1
-.1
-1.10
.2
-.3
2.23*
Consuming product makes my day better
.1
.1
.52
-.1
-.1
-.45
-.1
-.1
-.82
Bar tastes good
.0
.1
.22
-.1
-.2
-.93
-.1
-.1
-1.01
Bar is a bad snack/granola bar
-.1
-.2
-.97
-.0
-.0
-.19
-.1
-.1
-.78
Bar is a better bar than others
.0
.0
.13
.1
.1
.79
.1
.3
1.73*
* indicates p ≤ .15
Regression Analysis for Quaker:
Multiple regression analysis for Quaker shows a low coefficient of multiple determinants (R²), 10.1%, which indicates that there is almost no relationship between how respondents rated the Quaker brand on the ten Likert items and the respondents favorability of the Quaker ad, as indicated by their pre-post ad exposure change score. The standard error of estimate (se) is .5. This shows that respondents were an average of .5 a unit away from the regression line. The F-ratio of 0.85 was not significant, which means in 85 or more samples of every 100 samples drawn from the same population as this sample, we would not expect to find a multiple regression coefficient of the same magnitude. The item “the product fits my lifestyle (4)” was found to be significant, meaning that in 85 or more samples drawn from the same population as this sample that the correlation would be about what it is in this sample for this item.
The unstandardized coefficients (b) suggest the relationship between each Quaker brand attribute, or Likert item, and how respondents favored the Quaker print ad. The following is an explanation of the unstandardized coefficients (b):
- The more respondents think Quaker is recommendable, the less they like the ad.
- The more respondents think Quaker will help them be more physically fit, the more they like the ad.
- The more respondents think Quaker will give them more energy, the less they like the ad.
- The more respondents think Quaker fits their lifestyle, the less they like the ad.
- The more respondents think Quaker is more costly, the more they like the ad.
- The more respondents think Quaker has appealing packaging, the more they like the ad.
- The more respondents think Quaker makes their day better, the more they like the ad.
- The more respondents think Quaker snack/granola bars taste good, the more they like the ad.
- The more respondents think Quaker bars are bad snack/granola bars, the less they like the ad.
- The more respondents think Quaker is a better quality brand, the more they like the ad.
The multiple regression equation is written as follows:
Pre-post Ad Exposure Change Score for Quaker: 1.2 - 0.0 (recommendable) + 0.1 (physically) -.1 (more energy) -0.2 (lifestyle) + 0.1 (costly) + 0.1 (packaging) + 0.1 (better day) + 0.0 (tastes good) -0.1 (bad bar) + 0.0 (better brand)
One brand attribute can be considered important, meaning it best explains the variance in pre-post ad exposure change scores relative to the other brand attributes scores on the constant sum scale according to the value of beta: “fits the respondents lifestyle (4).”
Regression Analysis for Nature Valley:
Multiple regression analysis for Nature Valley shows a low coefficient of multiple determinants (R²)at 11.4%, which indicates that there is almost no relationship between how respondents rated the Nature Valley brand on the ten Likert items and the respondents favorability of the Nature Valley ad, as indicated by their pre-post ad exposure change score. The standard error of estimate (se) is .7. This shows that respondents were an average of .7 units away from the regression line.
The F-ratio of 0.97 was not significant, which means in 85 or more samples of every 100 samples drawn from the same population as this sample, we would not expect to find a multiple regression coefficient of the same magnitude. The item “the product makes you physically fit (2)” was found to be significant, meaning that in 85 or more samples drawn from the same population as this sample that the correlation would be about what it is in this sample for this item.
The unstandardized coefficients (b) suggest the relationship between each Nature Valley brand attribute, or Likert item, and how respondents favored the Nature Valley print ad. The following is an explanation of the unstandardized coefficients (b):
- The more respondents think Nature Valley is recommendable, the more they like the ad.
- The more respondents think Nature Valley will help them be more physically fit, the less they like the ad.
- The more respondents think Nature Valley will give them more energy, the less they like the ad.
- The more respondents think Nature Valley fits their lifestyle, the more they like the ad.
- The more respondents think Nature Valley is more costly, the less they like the ad.
- The more respondents think Nature Valley has appealing packaging, the less they like the ad.
- The more respondents think Nature Valley makes their day better, the less they like the ad.
- The more respondents think Nature Valley snack/granola bars taste good, the less they like the ad.
- The more respondents think Nature Valley bars are bad snack/granola bars, the less they like the ad.
- The more respondents think Nature Valley is a better quality brand, the more they like the ad.
The multiple regression equation is written as follows:
Pre-post Ad Exposure Change Score for Nature Valley: 2.6 + 0.1 (recommendable) + 0.1 (physically) - 0.1 (more energy) - 0.2 (lifestyle) + 0.1 (costly) - 0.1 (packaging) - 0.1 (better day) - 0.1 (tastes good) -0.1 (bad bar) + 0.1 (better brand)
One brand attribute can be considered important, meaning it best explains the variance in pre-post ad exposure change scores relative to the other brand attributes scores on the constant sum scale according to the value of beta: “packaging is appealing (6).”Regression Analysis for Kudos:
Multiple regression analysis for Kudos shows a low coefficient of multiple determinants (R²), 22.1%, which indicates that there is a low relationship between how respondents rated the Kudos brand on the ten Likert items and the respondents favorability of the Kudos ad, as indicated by their pre-post ad exposure change score. The standard error of estimate (se) is .5. This shows that respondents were an average of .5 a unit away from the regression line. The F-ratio of 2.13 was significant, which means in 85 or more samples of every 100 samples drawn from the same population as this sample, we would expect to find a multiple regression coefficient of the same magnitude. The items “the product makes you physically fit (2), product gives you more energy (3), packaging is appealing (6), and the bar is better than others (10)” were found to be significant, meaning that in 85 or more samples drawn from the same population as this sample that the correlation would be about what it is in this sample for this item.
The unstandardized coefficients (b) suggest the relationship between each Kudos brand attribute, or Likert item, and how respondents favored the Kudos print ad. The following is an explanation of the unstandardized coefficients (b):
- The more respondents think Kudos is recommendable, the more they like the ad.
- The more respondents think Kudos will help them be more physically fit, the more they like the ad.
- The more respondents think Kudos will give them more energy, the less they like the ad.
- The more respondents think Kudos fits their lifestyle, the less they like the ad.
- The more respondents think Kudos is more costly, the less they like the ad.
- The more respondents think Kudos has appealing packaging, the more they like the ad.
- The more respondents think Kudos makes their day better, the less they like the ad.
- The more respondents think Kudos snack/granola bars taste good, the less they like the ad.
- The more respondents think Kudos bars are bad snack/granola bars, the less they like the ad.
- The more respondents think Kudos is a better quality brand, the more they like the ad.
The multiple regression equation is written as follows:
Pre-post Ad Exposure Change Score for Kudos: 1.0 + 0.1 (recommendable) + 0.1 (physically) - 0.1 (more energy) - 0.1 (lifestyle) - 0.0 (costly) + 0.2 (packaging) - 0.1 (better day) - 0.1 (tastes good) - 0.1 (bad bar) + 0.1 (better brand)
Five brand attributes can be considered important, meaning they best explain the variance in pre-post ad exposure change scores relative to the other brand attributes scores on the constant sum scale according to the value of beta: Kudos “is recommendable (1),” “makes you physically fit (2),” “gives you more energy (3),” “packaging is appealing (6),” and “is better than others (10).”
A factor analysis was conducted for three breakfast bars (Quaker, Nature Valley, Kudos) represented by 86 responses to ten Likert items collected from an online survey. The purpose of the analysis was to find any relationship between the attributes via an attitude score by placing related brand attributes into independent factors. A paired t-test was then executed to determine the statistical significance for the mean differences between brand attitude scores.
Table 1. Total Variance Explained
FactorQuaker
Nature Valley
Kudos
Eigenvalue
% of variance
Cumulative variance %
Eigenvalue
% of variance
cumulative variance %
Eigenvalue
% of variance
cumulative variance %
I
4.5
44.8
44.8
4.5
44.8
44.8
4.8
48.2
48.2
II
1.3
12.7
57.5
1.3
12.6
57.3
1.4
14.5
62.7
III
1.0
9.9
67.4
1.0
9.6
67.0
0.8
8.0
70.7
IV
0.9
8.6
76.0
0.8
8.5
75.4
0.7
6.7
77.3
V
0.6
6.2
82.1
0.6
5.6
81.0
0.5
5.2
82.5
VI
0.5
5.4
87.6
0.5
4.9
86.0
0.5
4.6
87.1
VII
0.4
4.2
91.8
0.5
4.7
90.1
0.5
4.5
91.7
VIII
0.3
3.4
95.2
0.4
3.8
94.4
0.3
3.4
95.1
IX
0.3
2.7
98.0
0.3
3.2
97.7
0.3
2.8
97.9
X
0.2
2.1
100.0
0.2
2.3
100.0
0.2
2.0
100.0
Three of 10 factors created for Quaker have Eigenvalues greater than 1.0, indicating that they explain the variance of at least a single Likert item and hold as much information as the original content/question variable. These three factors combined explained 67.4% of the variance in the original ten Likert items and therefore 32.6% of variance is unexplained. In the case of Nature Valley, three factors combined accounted for 67.0% of the variance, leaving 33.0% of the variance unexplained in the original ten Likert items. Two of 10 factors created for Kudos have Eigenvalues greater than 1.0 and these factors explain 63.7% of the variance with 37.3% unexplained variance.
A squared factor loading is the percentage of variance explained by a factor in original variables such as ten Likert items. If we add up the squared factor loading over all the factors, the result becomes communality (h).
Communalities (h²) and Factor Loadings for Quaker
Brand AttributeCommunalities
Factor Matrix
Varimax Rotated Matrix
I
II
I
II
Product is recommendable
0.7
0.8
-0.2
0.8
0.3
Product makes you physically fit
0.6
0.7
0.4
0.2
0.8
Product gives you more energy
0.4
0.4
-0.4
0.6
-0.1
Product fits lifestyle
0.7
0.9
-0.1
0.7
0.5
Bar is more expensive than others
0.5
0.3
0.6
-0.2
0.7
Packaging is appealing
0.5
0.5
-0.5
0.7
-0.0
Consuming product makes my day better
0.5
0.7
-0.1
0.6
0.4
Bar tastes good
0.6
0.8
-0.1
0.7
0.4
Bar is a bad snack/granola bar
0.7
0.8
0.4
0.3
0.8
Bar is a better bar than others
0.6
0.7
0.1
0.5
0.6
This is a bad analysis, as no items could be correlated with either variable.
Communalities (h²) and Factor Loadings for Nature Valley
Brand AttributeCommunalities
Factor Matrix
Varimax Rotated Matrix
I
II
I
II
Product is recommendable
0.6
0.8
-0.0
0.8
0.1
Product makes you physically fit
0.5
0.4
0.6
0.3
0.6
Product gives you more energy
0.6
0.8
-0.0
0.8
0.1
Product fits lifestyle
0.6
0.8
0.1
0.8
0.2
Bar is more expensive than others
0.5
0.1
0.7
-0.0
0.7
Packaging is appealing
0.6
0.6
-0.5
0.7
-0.4
Consuming product makes my day better
0.5
0.7
-0.0
0.7
0.1
Bar tastes good
0.6
0.7
-0.1
0.7
0.1
Bar is a bad snack/granola bar
0.5
0.6
0.3
0.6
0.4
Bar is a better bar than others
0.7
0.8
-0.2
0.8
-0.0
This is a bad analysis, as no items could be correlated with either variable.
Communalities (h²) and Factor Loadings for Kudos
Brand AttributeCommunalities
Factor Matrix
Varimax Rotated Matrix
I
II
I
II
Product is recommendable
0.7
0.8
-0.2
0.8
0.1
Product makes you physically fit
0.8
0.6
0.8
0.1
0.9
Product gives you more energy
0.6
0.7
-0.0
0.7
0.3
Product fits lifestyle
0.6
0.8
0.1
0.7
0.4
Bar is more expensive than others
0.6
0.4
0.7
0.1
0.8
Packaging is appealing
0.6
0.7
-0.2
0.7
0.1
Consuming product makes my day better
0.7
0.8
-0.3
0.8
0.0
Bar tastes good
0.5
0.7
-0.2
0.7
0.1
Bar is a bad snack/granola bar
0.6
0.7
0.4
0.5
0.6