The following report discusses the statistical research methods used to analyze three brands of shampoo and for copy testing purposes. This research was conducted by a Master's Degree Candidate at the University of Texas at Austin in the Advertising Department. This was done as an assignment for the course ADV 380j.

Executive Summary

This report consists of several statistical methods which were used to perform copy-testing analysis.  Three brands of shampoo and their advertisements were used to conduct the research: Garnier Fructis, Pantene, and Aussie.  A questionnaire was set up online and several respondents were asked to fill it out and submit it for the uses of this research.  In total, there were 76 respondents.

Several research methods were used including: regression analysis, discriminant analysis, ANOVA/MANOVA, and factor analysis, along with other basic statistical methods.

This research was used to draw several conclusions.  An understanding of how consumers’ attitude (Likert scale items) towards the brand might have affected how they changed their opinion about the brand between before and after seeing the brand (change score) was developed using regression analysis.  Discriminant analysis was used to find relationships between up movers and down movers and their attitude towards the brand (Likert scale items).  The ANOVA test was used to determine if there was a relationship between one Likert scale item and the pre-post ad exposure change score as well as whether the respondent has kids or not. The MANOVA test was also used to help determine the relationship between the same two independent variables as in the ANOVA test and all 10 Likert items as the dependent variables. The factor analysis showed which factors affected the variance of at least one of the Likert scale items.  It also indicated whether the difference in two mean attitude scores for two different brands can be projected onto the population.

More specifically, it was found that for Aussie shampoo, there was not much of a difference in attitude according to the Likert scale items between up-movers and down-movers.  It was also found that 86.4% of original group cases were correctly classified for Aussie brand shampoo.  From the MANOVA test there were several conclusions drawn: (1) those respondents who were up movers and have kids agreed more with the statement that Pantene shampoo is just like every other brand of shampoo (mean score = 4.7); (2) those respondents who were down movers and do not have kids agreed more with the statement that they would not use Pantene shampoo (mean score = 4.3); (3) those respondents who were up movers and have kids disagreed more with the statement that Pantene is a good brand of shampoo (mean score = 4.7); (4) those respondents who were down movers and do not have kids also disagreed more with the statement that Pantene is a good brand of shampoo (mean score = 4.1). Overall, this research helped to draw the conclusion that Pantene is the more preferred brand of shampoo in this sample.

For a more in depth analysis, please refer to the main menu to read the full research report. 

Introduction

This research was a requirement for a graduate level advertising class at the University of Texas at Austin.  A questionnaire was set up online and then several potential respondents were asked to fill it out. For the sake of time and lack of funds, this was not a random sample.  In total, 76 questionnaires were used for this study.  Though there were more than that initially, it was found that several were not filled out correctly and were subsequently removed from the database.

This study was conducted for the uses of advertising research on three ads for three brands of shampoo.  These three brands included: Garnier Fructis, Pantene, and Aussie, which were chosen because they were brands of shampoo that were similar in price and quality.  The ads used for this study were taken from beauty magazines and were chosen because they advertised a shampooing product and because they were similar characteristically, such as: all were bleed page ads, all were in color, all featured an attractive woman.  However, they were different enough so that the respondents could easily make a distinction between them.  Each ad had a different amount of copy, the women were featured in different ways, and the products were positioned differently.

Through this study, it was possible to conclude what brand preferences were before the ads were shown and after the ads were shown, the attitude of consumers toward the brands, characteristics commonly associated with the brands, how the ads affected the respondents, and which brand was preferred overall.  Relationships between these elements were also studied, such as the relationship between up-movers and their attitude towards the brand.  The issue of significance will also be addressed because if a study is not found to be significant, then it cannot be projected onto the rest of the population. 

The rest of this report will describe the methodology used, the different types of analysis conducted, conclusions drawn from those analysis, a summary, and appendices which will include the questionnaire used and the mean, standard deviation, frequency, and percent for each answer given on the questionnaire.

Methodology: Click on any of the following sections to read more about them:

Structure of the Questionnaire

Design

Sampling Description

Data Collection Process

Structure of the Questionnaire

The questionnaire that was used to survey the 76 respondents in the sample was created using Dreamweaver 8.  The data was recorded and analyzed using SPSS.  There are 9 sections in total.  Also, the questionnaire is not very decorated with pictures of shampoo or the brands so as not to create any bias towards any brand.  The questionnaire is made up of a combination of different answering mechanisms such as radio buttons, check boxes, drop boxes, and text boxes.

 Title Page

The title page explains to the respondent what the survey is and why it is being conducted.  It lets them know their answers will be kept confidential and will only be used for the purposes of UT’s ADV 380j class. It also lets the respondent know that they must be 18 years of age or older to take the survey and, most importantly, it thanks them for their time.

Section One

This section is composed of three questions which ask what brand of body wash, hand soap, and shampoo the respondent bought the last time that they purchased these products. The respondent is then required to type in their answer in the text box.  Of these three questions, we are only really interested in which brand is put down the most for shampoo. This may seem a little random, but it was done in order to stray the respondent from having initial bias towards any brand or specific product category.

Section Two 

This section includes three parts which ask the respondent to answer on a 1-10 scale how likely they would be to buy three certain brands each for body wash, shampoo, and hand soap.  Again, the only brands we are really interested in are the ones for shampoo.  These scores will be used to calculate pre/post ad exposure change scores which will be used many times throughout the entire statistical analysis.  This is the pre ad exposure score.   

Section Three

Section three consists of three ads for shampoo that were shown to the respondents.  Each ad was chosen based on certain key characteristics; they were similar in many ways, but diverse enough for the respondents to be able to form opinions about the differences in each one.  They were all bleed page ads, full color, advertised a shampoo, and featured a woman.  The differences were in the amount and type of copy, the way in which the woman was featured, and how the product is displayed.  The respondents were told to read and examine the ads, and, once they continued, they were told not to return to the ads.  This way, it would try to control any inconsistencies in the number of times that one person looked at each ad.    

Section Four

This question includes the exact same questions as in section two.  These numbers will also help to calculate pre/post ad exposure change scores and will be called post ad exposure score.  This section is critical because it helps determine if there are any changes in how likely the respondent is to buy the brand after they have viewed the ad.

 Section Five

Section five asks the respondent point-blank if they feel differently about any of the brands of shampoo after viewing the ads.  If the answer is NO, the respondent moves on to section six, but if the answer is YES, they will continue to answer the questions in section five.  If the answer is YES, the respondent is then asked which band they felt differently about and why. 

Sections Six

This section contains three sets of ten Likert scale questions for the three brands of shampoo: Garnier Fructis, Aussie, and Pantene.  Each question is on a five point scale containing strongly agree, agree, neutral, disagree, and strongly disagree.  These items are used to determine the respondents’ attitude towards the brand.  Analysis will also be conducted on the relationship between the attitude scores and moves scores (pre/post ad exposure change scores).  The questions were written as follows (insert the brand name into the blank). Each answer on the Likert scale was assigned a number in order to be able to record quantitative data.

• I think ___________ is a good brand of shampoo.
• I don’t like how ____________ makes my hair feel after I use it.
• I think ___________ shampoo is too expensive.
• I think ___________ shampoo is better than other shampoos.
• ___________ is a brand I trust.
• ___________ shampoo gives my hair more volume.
• I would not use __________ shampoo.
• ____________ makes my hair stronger.
• ____________ makes my hair shinier.
• ____________ shampoo is like every other shampoo.

Section Seven

This section contains 10 characteristics that describe the ads that were shown for all three brands of shampoo: Garnier Fructis, Aussie, and Pantene.  The respondent is asked to check off any and all characteristics that they feel apply to each brand of shampoo.  This helps to determine how much the respondent like the ad that they were shown, and it is later used to see how it negatively affected pre/post ad exposure change scores.  Each check box was assigned a number in order to be able to record quantitative data.  The list of characteristics was as follows:

• Boring
• Informative
• Hard to Believe
• Memorable
• Different/unique
• Important
• Confusing
• Creative
• Common/Ordinary
• Easy to Understand

Section 8

In this section there were 3 groups of ten statements about the ad for each brand of shampoo.  The respondent was allowed to check yes or no depending on if they felt the statement was true of the ad or not.  These also helped to determine the attitude that people had towards the advertisements.  The statement used were as follows:

• This ad was annoying.
• This ad contained useful information.
• This ad was believable.
• This ad would be easy to remember.
• This ad would catch my attention if it was in a magazine.
• This ad made me think differently about shampoo.
• This ad would appeal to women more than men.
• I would like to see this ad again.
• This would be an easy target to forget.
• This ad taught me something new about shampoo. 

Section 9

This section includes questions regarding demographic information.  The respondents will answer questions regarding age, sex, number of children, and education along with questions about how often they buy shampoo and wash their hair.  There is also a box provided if the respondent wishes to write any further comments. This is helpful because you want to know what type of person is filling out your survey because maybe demographics might play a role in their attitude towards the brands. Some analysis, like MANOVA, use demographics such as how many children the person has in order to analyze a relationship.

BACK TO TOP

Design

The survey for this research was created by a Master’s Degree Candidate in the advertising department at the University of Texas. The questionnaire was created for the purpose of finding out what brands of shampoo respondents favored and how they felt about them and also to test advertisements for those same brands of shampoo and how people felt towards the ads.  It was then put on the internet for people to fill out.  Random sampling was not used, friends were told by friends to fill it out as a favor to the student conducting the survey.  There were several mechanisms used to create the questionnaire such as: text boxes, radio buttons, check boxes, and drop boxes.  In total, 76 respondents’ surveys were used to conduct the statistical analysis. These analyses included: ANOVA/MANOVA, t-tests, chi-squared tests, f-ratio, regression analysis, discriminant analysis, paired t-tests, simple correlation coefficient, etc…

BACK TO TOP

Sampling Description

The minimum amount of surveys required was 60.  Although more people filled out the survey, 76 were used for statistical analysis because these 85 were sufficiently and correctly filled out.  A random sample was not conducted for the sake of time and lack of funds.  The survey was passed on through e-mail and word of mouth and was mainly filled out by the researcher’s friends and family.  The main form of communication was an informal e-mail which was passed on from friend to friend and contained the URL for the questionnaire. 
The URL is:   http://www.ciadvertising.org/sa/fall_06/adv380j/jenna/shampoo.html

BACK TO TOP

Data Collection Process

A questionnaire was built using Dreamweaver in order to test brand preference of three brands of shampoo: Garnier Fructis, Aussie, and Pantene.  The questionnaire used a series of techniques in order to collect the answers from respondents such as: check boxes, radio buttons, text boxes, and drop boxes.  Each answer choice or text box was tagged with a number in order to easily record the data into the MS Access file and then into SPSS.

The survey itself was loaded on the University of Texas server using WS_FTP.  The data were collected using Microsoft Access and a Coldfusion file, and once enough respondents answered the questionnaire, it was then loaded into SPSS in order to conduct the statistical analysis assignments required by the graduate level course ADV 380j. 

The following is a breakdown of the techniques used in the questionnaire in order to collect the respondents’ answers:

Section 1, 2, 4: These sections use text boxes in which the respondents were able to type in any brand name or amount between 0 and 10 that they want.  Each answer is recorded into the Access database as text or as the number entered within the questionnaire.

Section 3: The three shampoo ads for Garnier Fructis, Aussie, and Pantene were placed in this section so that the respondents could look at them once each for as long as they want and then move on to the rest of the questionnaire. 
                                                                                                                    
Section 5: This section use radio buttons to record respondents’ answers and each button is given a 1 or 0 for a YES or NO answer, respectively.  If the question is not YES/NO and is multiple choice, each button is then given a 1, then 2, then 3, and so on, depending on how many answers are given.  These numbers were recorded into the MS Access database.

Section 6&8: These sections use radio buttons on a Likert scale to record respondents’ answers.  The Likert scale reads like so: Strongly Agree, Agree, Neutral, Disagree, and Strongly Disagree.  There is one radio button for each item, and each radio button was assigned a number.  If the question is positive, the numbering starts with ‘strongly agree’ as 5.  If the question is negative, the numbering starts with ‘strongly agree’ as 1.  These numbers were then recorded into the access database.

Section 8: This section uses check boxes in order to record the respondents’ answers.  Each check box was given assigned a 1 for recording purposes into the Access file.  This was done because the respondent is able to check as many boxes that apply to the corresponding brand.

BACK TO TOP

Click on any of the following sections to read more about them:

Analysis: Click on any of the following sections to read more about them:

Click on any of the following sections to read more about them:

Table 1: Mean and Standard Deviation of Likert Scale Items

Table 1 shows the mean and the standard deviation of the brand index score for each brand of shampoo. The lowest brand index score that a respondent could give to each brand was 10, and the highest score could be 50. According to the mean brand index scores given for the three brands, respondents had, on average, fairly neutral feelings towards all three brands. A lower standard deviation, such as Aussie’s 4.7, means that individual brand index scores that respondents gave did not deviate as far from the mean brand index score. From a managerial standpoint, the marketing department for each brand needs to work on creating positive associations with their brand in the mind of the consumer.  They could revamp the image of the products by using advertising forms such as magazine ads, TV ads, guerilla marketing, or sponsoring events in order to change people’s opinions about the shampoos.  

Table 2: Paired Samples Test

Table 2 gives the t-ratios for the 3 pairs of brands of shampoo which are being compared. The relationship between Garnier Fructis and Pantene and the relationship between Aussie and Pantene is significant. This means that for every 85 or more samples out of every 100 samples drawn from the same population as this sample, it would be expected that just as Pantene is preferred more than Garnier in this sample, the rest of the population would also prefer Pantene more than Garnier. Also, this means that for every 85 or more samples out of every 100 samples drawn from the same population as this sample, it would be expected that just as Pantene is preferred more than Aussie in this sample, the rest of the population would also prefer Pantene more than Aussie. From a managerial standpoint, the brand managers for Aussie and Garnier should find out why their brand is not preferred over Pantene. This is bad for these two brands because they are losing to the competition. They should conduct further surveys and analysis on how people feel about their brand and why they do not prefer it to Pantene. Then, the correct action should be taken to fix consumer perception about the brand so that they will ultimately buy the product. Pantene should keep themselves ahead of the competition by constantly doing market research and updating their product to meet consumer’s needs. 

Click on any of the following sections to read more about them:

Table 1: Group Statistics

Table 1 shows the mean ad index score and the mean brand index score and their standard deviations for two sets of groups: those who increased (up movers) in how likely they were to buy Pantene after they saw the ad, and those who decreased (down movers) in how likely they were to buy Pantene after they saw the ad. The higher the mean ad index score, the more positively the respondents felt about the ad. The higher the mean brand index score, the more positively the respondents felt towards the brand. The ad index score’s standard deviation for both the up and down movers indicates that neither group of people deviated very far from the mean ad index score. The brand index score’s standard deviation indicates that, for up movers, people tended to deviate further away from the mean brand index score than down movers.From a managerial standpoint, the brand managers should be worried about the ad that they are running for Pantene because it had a really low mean ad index score. An average of 5.8 for an ad index score suggests that many people did not like the ad shown for this study.   Pantene needs to do research on what people do not like about their ads and take action to correct it. The mean brand index score for respondents (both up and down movers) was still fairly neutral at around 35. Perhaps if Pantene can fix what people do not like about the ads that they are running, then people may start to feel more positively about the brand rather than neutral. These suggestions would only be worthwhile if the information provided was statistically significant.                          

Table 2: Independent Samples Test

Table 2 gives the t-ratio for the ad index score and the brand index score as compared between the up-movers and down-movers that were mentioned in the previous paragraph. Neither of these t-ratios were statistically significant; therefore, the information provided in table 3 could not necessarily be projected onto the rest of the population. Since the information provided by the sample population about up movers and down movers cannot be projected onto the rest of the population in this case, more data should be collected by survey in order to create a larger sample size that might be more representative of the population.

Click on any of the following sections to read more about them:

Table 1: Cross Tabulation           

Table 1 shows how many people who increased, stayed the same, and decreased  in how likely they were to buy Aussie brand shampoo as compared before they saw the ad and were also either above or below the median brand index score for Aussie which was 31. The greatest amount of people (28) was in the category of those movers who stayed the same in how likely they were to buy Aussie and were also below the median brand index score. The table shows that 82.4% of people that were below the median brand index score stayed the same in how likely they were to buy Aussie. The table also shows that 69.2% of people who decreased in how likely they are to buy Aussie were above the median brand index score. The table shows that 61.8% of people who were above the median brand index score also stayed the same in how likely they are to buy Aussie.

Table 2: Chi -Squared

Table 2 shows the chi-squared value for Aussie shampoo. It was found that the chi-squared value is significant. This means that there is a significant relationship between people who increased, decreased and stayed the same in how likely they would be to buy Aussie shampoo and being above or below the median brand index score for Aussie. It also means that for 85 out of 100 samples taken from a population similar to the one that this sample was taken from, the results would be similar to what they are for this sample. This information can be projected onto the population. The brand managers need to conduct further research to see if the large percentage of people who were below the median and stayed the same would in fact not buy Aussie. Table 5 might indicate that people did not like the ad that they were shown in this study and it actually made them less likely to buy the product. In this case, the brand managers should consider whether to change the ad, or even the entire approach they have taken for their advertisements.     

Click on any of the following sections to read more about them:

Table 1: Frequencies

Table 1 shows how many people for each brand of shampoo either moved up, down or stayed the same as far as how likely they would be to buy the brand as compared before and after they viewed the ads shown for this study. Most people stayed the same when it came to how likely they would be to buy the brand. Both Aussie and Pantene had more people decrease (move down) in how likely they would be to buy the brand than people that increased (moved up) after they saw the ads. Garnier Fructis, however, had twice as many people increase in how likely they would be to buy the brand as people that decreased after they saw the ad. From a managerial standpoint, this could indicate that Garnier is taking the right approach with their advertising. However, Garnier brand managers also need to keep in mind that the median brand index score, on average, for Garnier is still around neutral and is not viewed as positively, on average, as Pantene. So, brand managers need to find out why people like the ads, but do not have as good of an opinion of the product. Aussie and Pantene brand managers need to find out what is wrong with their advertisements, because these numbers could indicate that consumers do not like the ads. 

Click on any of the following sections to read more about them:

Table 1: Frequencies

Table 1 shows how many people preferred Aussie over Pantene or Pantene over Aussie according to the brand index score the respondent gave the brand. This shows that Pantene is more than twice as preferred as Aussie. The brand managers at Aussie need to figure out why people prefer Pantene over their brand. It could be the formula of shampoo, the advertisements, lack of knowledge about the brand, or any number of things. Aussie could need an image makeover, or maybe a different approach to advertising, or maybe a completely new formula that makes up the shampoo. 

Click on any of the following sections to read more about them:

Table 1: Correlation Coefficient

Table 1 shows the correlation coefficient between Pantene’s brand index sore and Garnier Fructis’ brand index score. The linear association between the two variables is fairly small. Also, the correlation coefficient is negative so this means that as one variable increases or decreases the other variable does the exact opposite. To management, this means that as one person has more of a positive association with Pantene, the less of a positive association the person will have for Garnier or vice versa. However, since the linear association between these two is small, this cannot always or definitely be assumed. Brand managers for both brands should be concerned about making their shampoo the one that has the most positive association in the consumer’s mind in order to stay ahead of the competition. 

Click on any of the following sections to read more about them: