market research
DESCRIPTION
Market ResearchTRANSCRIPT
Market ResearchContentsMissing Value analysis2Reliability testing2Multi Collinearity3Validity Check4Multiple Regression4Discriminant Analysis5Factor Analysis8Cluster Analysis10Multidimensional scaling14
Missing Value analysis
Done when few responses are not recorded due to some problem
Reliability testing
Based in Cronbach alpha >.6Item total statistics check cronbach alpha if item deleted(if in any variable it is greater than the existing value then delete the question to increase it)
Data reduction technique Cronbach alpha should be greater than .6Analysis > Scale > Reliability analysis > Statistics > Scale is item deleted (shows if the item is deleted by how much will the cronbach alpha will increase)To identify the excluded data first use frequency table and then missing data analysis
Reliability Statistics
Cronbach's AlphaN of Items
.8105
Item-Total Statistics
Scale Mean if Item DeletedScale Variance if Item DeletedCorrected Item-Total CorrelationCronbach's Alpha if Item Deleted
Rating of Quality of Mechandise at Sears16.9712.698.616.769
Rating of Variety and Assortment of Mechandise at Sears16.6012.157.622.766
Rating of Returns and Adjustment Policy of Sears16.3312.275.619.767
Rating of Service of Personnel of Sears17.3211.869.549.793
Rating of Perception of Fair Prices at Sears16.7012.938.598.774
Multi Collinearity
Analyze> regression > linearStatistics > Collinearity diagnosistics(Compare each variable with all other variables, dependent and independent combinations)
VIF (variable inflation factor) not ok (both questions are the same remove one of them)
Coefficientsa
ModelCollinearity Statistics
ToleranceVIF
1Rating of Variety and Assortment of Mechandise at Sears.6241.603
Rating of Returns and Adjustment Policy of Sears.5801.725
Rating of Service of Personnel of Sears.6801.471
Rating of Perception of Fair Prices at Sears.6791.472
Rating of Convenience of Location of Sears.8791.138
a. Dependent Variable: Rating of Quality of Mechandise at Sears
Validity Check1. Convergent- correlation of items of the same factor should be high2. Discriminant- correlation of items of different factors should be low
Multiple Regression
Model Summaryb
ModelRR SquareAdjusted R SquareStd. Error of the Estimate
1.927a.859.830.81681
a. Predictors: (Constant), attract, strong, shiny, fresh, decay
b. Dependent Variable: cavity
Model fit via R^2 >.6(how much of the dependent variable is explained by the independent variable) ANOVAa
ModelSum of SquaresdfMean SquareFSig.
1Regression97.854519.57129.334.000b
Residual16.01224.667
Total113.86729
a. Dependent Variable: cavity
b. Predictors: (Constant), attract, strong, shiny, fresh, decay
Accept or reject the null hypothesis( no relationship) Considering the significance valueCoefficientsa
ModelUnstandardized CoefficientsStandardized CoefficientstSig.
BStd. ErrorBeta
1(Constant)3.1361.2512.506.019
shiny.134.151.093.889.383
strong.550.129.5714.272.000
fresh.162.160.1131.016.320
decay-.456.134-.439-3.399.002
attract-.252.162-.177-1.558.132
a. Dependent Variable: cavity
Again consider the significance values to reject or accept the variablesIf greater than .05 reject
Equations based of B value
Discriminant Analysis
Case 1:
Y= a + b1x1 + b2x2 + b3x3 + b4x4 + b5x5Y loyalty (highly loyal, loyal, neutral , not loyal , highly disloyal)X1 ageX2 genderX3 freq.X4 incomeX5 heightAll independent variable should be metric for multiple discriminant analysisBy metric we mean which can be measured unlike gender
If non metric variable exist we use logistic regression
NM = M (multiple discriminant analysis)NM= M/NM (logistic regression)
Analyze > Classify > Discriminant (grouping variable / Define Range) Enter Independent data Statistics (Means , univariate ANNOVAs, Winthin-group correlation) Classify( summary table , leave one out classification, all groups equal, within-groups) Save ( predictive group membership)Wilks' Lambda
Test of Function(s)Wilks' LambdaChi-squaredfSig.
1.53316.6903.001
To test Model fit (Sig. scale> multidimensional scale (Alscale)Model> Euclidean distance, level of measurement (ordinal), conditionality (matrix), Dimension (min 1 max 3 depending upon need)Options> individual subject plots
Analyze STRESS FACTOR AND R^2
Stress factor.2 Poor fit.1 - Fair.05 - Good.025 - Excellent.000 Perfect
RSQ >.6
Dimension 1 and 2 to be given by the researcherNew entrant can position the brand in the top left corner