Assignement2_Group4
-
Upload
souhardya-banerjee -
Category
Documents
-
view
212 -
download
0
description
Transcript of Assignement2_Group4
1• Managerial Problem is defined
2• Computation of aggregate factors and factor Analysis
3• Regression Analysis with impact of linear/non-linear relationships on Key Loyalty
predictors
4• Take assumptions and validate for regression
5• “Best” predictive model for aggregate data is interpretted
6• Moderating affects of variables on best aggregate model
7• customer segments to loyalty predcitors implicaions
8• Recommendations for “Driving Loyalty Up”
FLOW OF PRESENTATION
Problem Statement
• An insurance company facing stiff global competition chipping its loyal customer base
• Tough business environment • Price competitiveness need to match customer
needs for trusting relationships• To enhance the company’s competitive position in
the customer’s mind• Company Seeking for “Driving Loyalty Up” Strategy
Composites for Regression Analysis
• Beh_loyalty: (loy1+loy2+loy3+loy4)/4• Cog_loyalty : (loy5+loy6+loy7+loy8)/4• Satisfaction = (inter1 + inter2 + inter3)/3• Trust on Agent = (rep17+rep18+rep19+rep20)/4.• Trust on company
=(prac17+prac18+prac19+prac20)/4.• Value_Shortterm = (val1+val2+val3)/3.• Value_Longterm= (val4+val5+val6)/3.
• Hypothesis : – H0 : KLP’s (trust on agent, trust on company, customer satisfaction, long
term and short term value ) do not have a influence in driving up the loyalty loyalty.
– H1: KLP’s (trust on agent, trust on company, customer satisfaction, long term and short term value ) have a influence in driving up the loyalty loyalty.
• If p value > 0.05 , we accept the null hypothesis and reject otherwise
Variable Setup in SPSS Statistics
Outcome Variable : •Cog_loyalty ( for cognitive Loyalty)•Beh_Loyaltty(For behavioral Loyalty)
Predictor Variable : Satisfaction Trust_agent Trust_Comp Value_ST, Value_LT
KLP’s explain 55.2% of the variablity on Beh_Loyalty
The table shows that the independent variables statistically significantly predict the dependent variable, F(5,773) = 1095, p < .0005 (i.e., the regression model is a good fit of the data
Regression on Behavior Loyalty(Beh_Loyalty)
From the "Sig." column that all independent variable coefficients are statistically significantly different from 0 (zero)
Statistical significance of the independent variables
Results• Adjusted R square change is 0.552 means 55.2% of main factors are
influencing model.• It is significant as for p < 0.05 and hypothesis is rejected• Behavioural Loyalty is dependent on all the independent variables• The dependency is higher in satisfaction and long term value variable• VIF< 10 that satisfies that they are not multi-collinear.
Normality
Since median comes user +-2sigma mean. Also std. residual is zero at mean. Thus it is normal
Linear
Trends show that it is following a nearly a linear trial.
Independent of Errors
The linear curve shows that it is Independent of errors..
Regression on Cognitive Loyalty(Cog_Loyalty)
KLP’s explain 52.3% of the variablity on Beh_Loyalty
The table shows that the independent variables statistically significantly predict the dependent variable, F(5,762) = .912, p < .0005 (i.e., the regression model is a good fit of the data
Results• Adjusted R square change is 0.523 means 52.3% of main factors are
influencing• It is significant as for p < 0.05 and null hypothesis is rejected• Cognitive Loyalty is dependent on all the independent variables• The dependency is higher in short term value variable.• VIF< 10 that satisfies that they are not multi-collinear.
Normality
Since median comes user +-2sigma mean. Also std. residual is zero at mean. Thus it is normal
Linear
Trends shows that it is following a nearly a linear trial.
Independent of Errors
The linear curve shows that it is Independent of errors..
KLP’s explain 60% of the variablity on Beh_Loyalty
The table shows that the independent variables statistically significantly predict the dependent variable, F(5,773) = .902, p < .0005 (i.e., the regression model is a good fit of the data
Regression on Total Loyalty
Results• Adjusted R square change is 0.602 means 60.2% of main factors are
influencing model.• It is significant as for p < 0.05 and hypothesis is rejected• Behavioural Loyalty is dependent on all the independent variables• The dependency is higher in satisfaction and long term value variable• VIF< 10 that satisfies that they are not multi-collinear.
Normality
Since median comes user +-2sigma mean. Also std. residual is zero at mean. Thus it is normal
Linear
Trends show that it is following a nearly a linear trial.
Independent of Errors
The linear curve shows that it is Independent of errors..
MulticollinearityWe have considered 0.9 as the cut-off value. Since no correlation value is above or equal to 0.9, it can be concluded that significant multicollinearity is not present between any of the independent variables.
Therefore, it is appropriate to proceed with multiple regression.
Multiple regression
The Adjusted R square value is 0.600.This indicates that 60% variation in Loyalty which is the dependant variable is explained by the independent variables which we have considered.The model is converging well since significance is less than 0.5
Multiple regression
All the Independent Variables have significance values <0.05. Hence all of them are going to be part of the Multiple Regression Model.Based on the unstandardized coefficients of the IVs, following model is made –
Loyalty = 0.004 + 0.127*Agent + 0.123*Trust + 0.135*Value_Shortterm+ 0.160*Value_Longterm + 0.133*Satisfaction
Analysis – Aggregate Data
• Based on the outcome of multiple regression, we can conclude that all independent variables has to be considered.
• Further, observing the coefficients tell us that a unit increase in Value_Longterm will contribute most to Dependent Variable Loyalty.
• This is followed by Value_Shortterm and Satisfaction, which are closely followed by Agent and Trust
ANOVAa
ModelSum of
Squares df Mean Square F Sig.1 Regression 500.027 5 100.005 111.642 .000b
Residual 293.813 328 .896 Total 793.840 333
a. Dependent Variable: loyaltyb. Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longterm
Multiple Regression Model significance
ANOVAa
Model Sum of Squares df Mean Square F Sig.1 Regression 423.537 5 84.707 125.127 .000b
Residual 159.088 235 .677 Total 582.625 240
a. Dependent Variable: loyaltyb. Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longterm
The US-specific data
ANOVAa
Model Sum of Squares df Mean Square F Sig.1 Regression 133.154 5 26.631 23.006 .000b
Residual 229.196 198 1.158
Total 362.350 203
a. Dependent Variable: loyalty
b. Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longterm
Germany-specific data
Holland-specific data
It is observed that the models are significant (<0.05) in case of the data specific to all three countries.
Model Summaryb
Model R R SquareAdjusted R
Square
Std. Error of the
Estimate
Change StatisticsDurbin-Watson
R Square Change F Change df1 df2
Sig. F Change
1 .853a .727 .721 .82278 .727 125.127 5 235 .000 2.170
a. Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longterm
b. Dependent Variable: loyalty
Model Summaryb
Model R R SquareAdjusted R
Square
Std. Error of the
Estimate
Change StatisticsDurbin-Watson
R Square Change F Change df1 df2
Sig. F Change
1 .794a .630 .624 .94645 .630 111.642 5 328 .000 1.836
a. Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longterm
b. Dependent Variable: loyalty
Model Summaryb
Model R R SquareAdjusted R
Square
Std. Error of the
Estimate
Change StatisticsDurbin-Watson
R Square Change F Change df1 df2
Sig. F Change
1 .606a .367 .351 1.07590 .367 23.006 5 198 .000 2.102a Predictors: (Constant), Value_Shortterm, Satisfaction, Trust, Agent, Value_Longtermb. Dependent Variable: loyalty
The US
Germany
Holland
Adjusted R Square
The highest value of Adj. R sq. is in case of the US, which is 0.721. i.e. 72.1% variation in Loyalty is explained by the IVs considered. That value is 62.4% in case of Germany.Holland, however, has the adj. R sq. value just 0.351, which means there exist other factorsapart from the IVs considered who drive the customer loyalty.
Coefficientsa
Model
Unstandardized Coefficients
Standardized
Coefficients
t Sig.
95.0% Confidence Interval for B Collinearity Statistics
B Std. Error BetaLower Bound
Upper Bound Tolerance VIF
1 (Constant) .312 .220 1.421 .157 -.121 .745
Satisfaction .081 .037 .115 2.172 .031 .007 .154 .414 2.418
Agent .128 .044 .185 2.884 .004 .041 .216 .282 3.543
Trust .158 .050 .231 3.165 .002 .060 .256 .219 4.569
Value_Shortterm
.096 .053 .131 1.838 .067 -.007 .200 .228 4.394
Value_longterm
.190 .057 .278 3.341 .001 .078 .303 .168 5.958
a. Dependent Variable: loyalty
Multiple Regression Model
All the Independent Variables have significance values <0.05. Therefore all of them need to be considered for the Multiple Regression Model.Based on the unstandardized coefficients of the IVs, following model is made –
Loyalty = 0.312 + 0.081*Satisfaction + 0.128*Agent + 0.158*Trust+ 0.096*Value_Shortterm + 0.190*Value_Longterm
The US
Coefficientsa
Model
Unstandardized Coefficients
Standardized
Coefficients
t Sig.
95.0% Confidence Interval for B Collinearity Statistics
B Std. Error BetaLower Bound
Upper Bound Tolerance VIF
1 (Constant) -.255 .237 -1.077 .282 -.722 .211 Satisfaction .226 .054 .292 4.185 .000 .120 .332 .232 4.308Agent .064 .053 .083 1.207 .228 -.040 .168 .239 4.191
Trust .120 .060 .151 1.980 .049 .001 .238 .194 5.145
Value_Shortterm
.184 .051 .218 3.626 .000 .084 .283 .313 3.195
Value_longterm
.110 .053 .137 2.079 .038 .006 .213 .261 3.825
a. Dependent Variable: loyalty
Multiple Regression Model
All the Independent Variabless have significance values <0.05 except for the Agent. Rest of them are going to be part of the Multiple Regression Model.Based on the unstandardized coefficients of the IVs, following model is made –
Loyalty = -0.255 + 0.226*Satisfaction + 0.120*Trust+ 0.184*Value_Shortterm + 0.110*Value_Longterm
Germany
Coefficientsa
Model
Unstandardized Coefficients
Standardized
Coefficients
t Sig.
95.0% Confidence Interval for B Collinearity Statistics
B Std. Error BetaLower Bound
Upper Bound Tolerance VIF
1 (Constant) .236 .459 .515 .607 -.669 1.142
Satisfaction
.096 .100 .098 .956 .340 -.101 .293 .302 3.308
Agent .172 .082 .176 2.086 .038 .009 .335 .447 2.238
Trust .091 .089 .097 1.023 .308 -.084 .266 .353 2.831
Value_Shortterm
.010 .112 .011 .088 .930 -.211 .230 .221 4.517
Value_longterm
.280 .093 .316 3.025 .003 .098 .463 .293 3.415
a. Dependent Variable: loyalty
Multiple Regression Model
Only Agent and Value_Longterm are significant variables.Therefore only these two are going to be part of the Multiple Regression Model.Based on the unstandardized coefficients of the IVs, following model is made –
Loyalty = 0.236 + 0.172*Agent + 0.280*Value_Longterm
Holland
Analysis – Country-specific Data
•Based on the coefficients obtained from Multiple Regression, it is observed that a unit increase in Value_LongTerm contributes most to the dependent variable Loyalty in case of US and Holland.•For Germany, Satisfaction has the most bearing on the Loyalty for every unit increase.•The results of multiple regression on the US-specific data indicates that customer loyalty is significantly dependent on all the Independent Variables under consideration.•For Germany, Agent is not significant. •For Holland, only Agent and Value_Longterm are significant.•Thus, factors influencing customer loyalty varies from country to country