Você está na página 1de 35

19-1

Chapter 19
Measures of Association

19-2

Learning Objectives
Understand . . . how correlation analysis may be applied to study relationships between two or more variables uses, requirements, and interpretation of the product moment correlation coefficient how predictions are made with regression analysis using the method of least squares to minimize errors in drawing a line of best fit

19-3

Learning Objectives
Understand . . . how to test regression models for linearity and whether the equation is effective in fitting the data nonparametric measures of association and the alternatives they offer when key assumptions and requirements for parametric techniques cannot be met

19-4

Exhibit 19-1 Measures of Association: Interval/Ratio


Pearson correlation coefficient
Correlation ratio (eta) For continuous linearly related variables For nonlinear data or relating a main effect to a continuous dependent variable One continuous and one dichotomous variable with an underlying normal distribution Three variables; relating two with the thirds effect taken out Three variables; relating one variable with two others

Biserial

Partial correlation Multiple correlation Bivariate linear regression

Predicting one variable from anothers scores

19-5

Exhibit 19-1 Measures of Association: Ordinal


Gamma Based on concordant-discordant pairs; proportional reduction in error (PRE) interpretation P-Q based; adjustment for tied ranks P-Q based; adjustment for table dimensions P-Q based; asymmetrical extension of gamma Product moment correlation for ranked data

Kendalls tau b

Kendalls tau c

Somerss d Spearmans rho

19-6

Exhibit 19-1 Measures of Association: Nominal


Phi Cramers V Contingency coefficient C Lambda Chi-square based for 2*2 tables CS based; adjustment when one table dimension >2 CS based; flexible data and distribution assumptions PRE based interpretation PRE based with table marginals emphasis

Goodman & Kruskals tau Uncertainty coefficient


Kappa

Useful for multidimensional tables


Agreement measure

19-7

Relationships
To truly understand consumers motives and actions, you must determine relationships between what they think and feel and what they actually do.
David Singleton, Vice President of Insights, Zyman Marketing Group

19-8

Pearsons Product Moment Correlation r


Is there a relationship between X and Y?

What is the magnitude of the relationship?

What is the direction of the relationship?

19-9

Exhibit 19-2 Scatterplots of Relationships

19-10

Exhibit 19-4 Scatterplots

19-11

Exhibit 19-7 Diagram of Common Variance

19-12

Interpretation of Correlations
X causes Y

Y causes X

X and Y are activated by one or more other variables X and Y influence each other reciprocally

19-13

Exhibit 19-8 Artifact Correlations

19-14

Interpretation of Coefficients

A coefficient is not remarkable simply because it is statistically significant! It must be practically meaningful.

19-15

Exhibit 19-9 Comparison of Bivariate Linear Correlation and Regression

19-16

Exhibit 19-10 Examples of Different Slopes

19-17

Concept Application
X Average Temperature (Celsius) Y Price per Case (FF)

12 16 20 24 Mean =18

2,000 3,000 4,000 5,000 Mean = 3,500

19-18

Exhibit 19-11 Plot of Wine Price by Average Temperature

19-19

Exhibit 19-12 Distribution of Y for Observation of X

19-20

Exhibit 19-14 Wine Price Study Example

19-21

Exhibit 19-15 Least Squares Line: Wine Price Study

19-22

Exhibit 19-16 Plot of Standardized Residuals

19-23

Exhibit 19-17 Prediction and Confidence Bands

19-24

Testing Goodness of Fit


Y is completely unrelated to X and no systematic pattern is evident There are constant values of Y for every value of X
The data are related but represented by a nonlinear function

19-25

Exhibit 19-18 Components of Variation

19-26

Exhibit 19-19 F Ratio in Regression

19-27

Coefficient of Determination: r2

Total proportion of variance in Y explained by X Desired r2: 80% or more

19-28

Exhibit 19-20 Chi-Square Based Measures

19-29

Exhibit 19-21 Proportional Reduction of Error Measures

19-30

Exhibit 19-22 Statistical Alternatives for Ordinal Measures

19-31

Exhibit 19-23 Calculation of Concordant (P), Discordant (Q), Tied (Tx,Ty), and Total Paired Observations: KeyDesign Example

19-32

Exhibit 19-24 KDL Data for Spearmans Rho


_______ _____ Rank By_____ _____ _____ Applicant 1 2 3 4 5 6 7 8 9 10 Panel x 3.5 10.0 6.5 2.0 1.0 9.0 3.5 6.5 8.0 5.0 Psychologist y 6.0 5.0 8.0 1.5 3.0 7.0 1.5 9.0 10.0 4.0 d -2.5 5.0 -1.5 .05 -2 2.0 2.0 -2.5 -2 1.0 d2 6.25 25.00 2.52 0.25 4.00 4.00 4.00 6.25 4.00 _1.00_ 57.00
.

19-33

Key Terms
Artifact correlations Bivariate correlation analysis Bivariate normal distribution Chi-square-based measures Contingency coefficient C Cramers V Phi Coefficient of determination (r2) Concordant Correlation matrix Discordant Error term Goodness of fit lambda

19-34

Key Terms
Linearity Method of least squares Ordinal measures Gamma Somerss d Spearmans rho tau b tau c Pearson correlation coefficient Prediction and confidence bands Proportional reduction in error (PRE) Regression analysis Regression coefficients

19-35

Key Terms
Intercept Slope Residual Scatterplot Simple prediction tau