Final Project - Group 1
Final Project - Group 1
Final Project - Group 1
Prepared By:
Ana Chikovani
Tekla Tavkhelidze
Luka khmaladze
Mariam Buhagiar
Barbare Mekhrishvili
Description: As a researcher you are interested in Brand Locality. You have identified following 5 major factors of the dependent
variable: Company Culture, Creative Marketing, Customer Service quality, Company Ethics and Product Quality. You have
randomly selected 31 respondents and asked them to rate all 6 variables (dependent and independent) from 1 to 5. With 1 being the
lowest and 5 being highest. Project Requirements:
Project Requirements:
Data is obtained and SPSS multiple regression is run correctly. Student obtained Descriptive
Statistics, Correlations, ANOVA tables, Coefficients table, Coefficient Correlations, Collinearity
Diagnostics, Residual Statistics, Histogram of dependent variable, normal plot of regression
standardized residuals and scatter plots (5 points)
Using SPSS outcome, student discussed the model selection process. How and why were the
independent variables chosen (1 point).
Student used SPSS outcome and checked variation and models significance level (1 point).
Using SPSS outcome, student checked correlation and collinearity of the model (1 point)
Student checked Residual Statistics (1 point).
Student stated prediction model and concluded on its “generalization” (1 point).
What happens if you have following data, customers assess XYZ company’s Company Culture=2,
Creative Marketing=4, Customer Service quality=2, Company Ethics=5 and Product Quality=5.
How would you rate the company’s Brand Loyalty? Why? Which variable(s) play major part? Why?
What would you recommend to the company? (bonus 2 points)
Content
Assumptions of normality
Linear regression
Regression equation
Predicted funding
Assumptions of normality
Descriptives Brand Loyalty - Mean (2.41) and Median (2.00) are close to each
Statistic Std. Error other. Median is a little smaller than the Mean which indicates that
Brand Mean 2.41 .141 the data is slightly skewed right. Skewness (-0.080) is Close to 0 this
Loyalty 95% Confidence Interval Lower Bound 2.12
means that the curve is almost Symmetrical. Histogram seems bell
for Mean Upper Bound 2.69
shape. We see that Box Plot has some problems Median (Q2) is very
5% Trimmed Mean 2.40
close to Q1. We do not have outliers. The distance between the
Median 2.00
minimum and Q1 is about the same as the distance between Q3 and
Variance .636
Minimum 1 around the straight line, meaning it is easy to predict. Q-Q detrended
Maximum 4 plot doesn’t seem quite normal, we see that data point are scattered
Range 3 asymmetrically, normally they should be dispersed more
Interquartile Range 1 symmetrically and closer to the horizontal line.
Skewness -.080 .414
We clearly see that there are some problems, to conclude that we
Kurtosis -.361 .809
have normal probability distribution. But for our research purposes
Correlations
Company Creative Customer
Brand Loyalty Culture Marketing Service Company Ethics Product Quality
Pearson Correlation Brand Loyalty 1.000 .862 .739 .282 .435 .738
Company Culture 32 32 32 32 32 32
Creative Marketing 32 32 32 32 32 32
Customer Service 32 32 32 32 32 32
Company Ethics 32 32 32 32 32 32
Product Quality 32 32 32 32 32 32
Let’s analyze given result. Brand Loyalty has highest correlation with Company culture (0.862), then follows Creative marketing (0.739) and
product quality (0.738). Company ethics and customer services have lowest correlations towards Brand Loyalty 0.435 and 0.282 accordingly. But
to choose right X-s for our Y we need to look at correlations between independent variables. Company culture have high correlations with
creative marketing and product quality, 0.617 and 0.651 accordingly, meaning they explain lot of same information for Y, it’s logical to choose
one of them for Y, and that’s would be Company Culture with highest correlation with Brand loyalty which is 0.862. for Second variable it would
be logical to choose the variable which has high correlation with Brand loyalty and low correlation with Company culture and that in our case is
Company ethics, with correlations Y and X 1 0.435 and 0.2 accordingly. to sum up the most informative independent variables for Brand Loyalty
are Company culture and company ethics with correlations 0.862 and 0.435 accordingly. X 3 according SPSS program is Creative Market,
personally I wouldn’t leave it and would try to build model without that variable, but for our analysis let’s trust SPSS :D,
Variables Entered/Removed a
Model Variables Entered Variables Removed Method
1
Stepwise (Criteria: Probability-of-F-to-enter <= .050,
Company Culture .
Probability-of-F-to-remove >= .100).
2
Stepwise (Criteria: Probability-of-F-to-enter <= .050,
Company Ethics .
Probability-of-F-to-remove >= .100).
3
Stepwise (Criteria: Probability-of-F-to-enter <= .050,
Creative Marketing .
Probability-of-F-to-remove >= .100).
SPSS Conducted F-Test And excluded variables that weren’t important. And left following variable as we assumed earlier:
X1 Company Culture
X2 Company Ethics
X3 Creative Market
Excluded Variablesa
Collinearity Statistics
Minimum
Model Beta In t Sig. Partial Correlation Tolerance VIF Tolerance
d. Predictors in the Model: (Constant), Company Culture, Company Ethics, Creative Marketing
SPSS used stepwise method to exclude variables from our model. According To T-test it excluded as predicted previously Customer Service and
Product quality. As we see in the first step Company Ethics was removed from excluded variables, then running second t test Creative Market
Was removed from the list. And on the third step two Excluded variables were left Customer service and Product quality.
Model Summaryd
Change Statistics
Adjusted R Std. Error of the
Model R R Square Square Estimate R Square Change F Change df1 df2 Sig. F Change
R square shows us how well regression model explains observed data. From Table we see that the first model with just X 1 Company Culture can
explain 0.743 of Brand loyalty, by adding X 2 Company Ethics variable our R square grew to 0.815 and by adding X 3 Creative Market variable to
the model our square grew to 0.856. The model with just X 1 variable was good enough by itself (R square was more than 70%) but by adding X2
and X3 increased accuracy. We can conclude that dependent variable is explained well by the independent variables.
ANOVAa ANOVA test is used to determine the influence that
Model Sum of Squares df Mean Square F Sig.
independent variables have on the dependent variable
1 Regression 14.645 1 14.645 86.602 .000b
in a regression study. We conducted F-Test: At
Residual 5.073 30 .169
significance level 0.000 which is less than α (0.05),
Total 19.719 31
we can conclude that the means are not equal,
2 Regression 16.061 2 8.031 63.670 .000c
Residual 3.658 29 .126 meaning there is significant connection between
Total 19.719 31 independent variables and dependent variables, so we
3 Regression 16.872 3 5.624 55.316 .000 d
have regression.
Residual 2.847 28 .102
Total 19.719 31
Coefficientsa
Standardized
Unstandardized Coefficients Coefficients Correlations Collinearity Statistics
Model B Std. Error Beta t Sig. Zero-order Partial Part Tolerance VIF
Company Culture .636 .068 .862 9.306 .000 .862 .862 .862 1.000 1.000
2 (Constant) .254 .207 1.230 .229
Company Culture .596 .060 .807 9.887 .000 .862 .878 .791 .960 1.042
Company Ethics .277 .083 .273 3.350 .002 .435 .528 .268 .960 1.042
3 (Constant) .021 .203 .105 .917
Company Culture .482 .067 .653 7.158 .000 .862 .804 .514 .619 1.616
Company Ethics .223 .077 .220 2.904 .007 .435 .481 .208 .900 1.112
Creative Marketing .195 .069 .266 2.824 .009 .739 .471 .203 .580 1.724
We conducted t-test and based on these data we concluded that B 0 and B1 are not equal to zero. Step 1 contained Just One independent variable,
step two contained two variables and step three contained all 3 independent variables, in all cases B0 and B1 were different than 0.
Collinearity Diagnosticsa
Variance Proportions
Model Dimension Eigenvalue Condition Index (Constant) Company Culture Company Ethics Creative Marketing
Collinearity – here we need dependent and independent variables to have significant correlations. When we don’t need correlation among
independent variables, because we want them to measure different parts of variance on Y. We see that t statistics is 7.5 and significance level is 0.
Therefore, there is significant correlation between dependent and independent variables. We can also see that there is high level .737 correlation,
with significance level of .000, therefore we reject null hypothesis, that correlation is significantly different from zero.