Comparison of discriminant vs. logistic regression analyses for predicting consumer acceptance and purchase decision Darryl Holliday, Ashley Bond, Witoon.

Slides:



Advertisements
Similar presentations
Analyzing Significant Differences between Means
Advertisements

Assumptions underlying regression analysis
Lecture 28 Categorical variables: –Review of slides from lecture 27 (reprint of lecture 27 categorical variables slides with typos corrected) –Practice.
Lecture 3: Chi-Sqaure, correlation and your dissertation proposal Non-parametric data: the Chi-Square test Statistical correlation and regression: parametric.
Statistical Methods Chichang Jou Tamkang University.
PY 427 Statistics 1Fall 2006 Kin Ching Kong, Ph.D Lecture 12 Chicago School of Professional Psychology.
PPA 415 – Research Methods in Public Administration Lecture 8 – Chi-square.
© 2005 The McGraw-Hill Companies, Inc., All Rights Reserved. Chapter 13 Using Inferential Statistics.
Business Statistics - QBM117 Statistical inference for regression.
Chapter 7 Correlational Research Gay, Mills, and Airasian
Review for Exam 2 Some important themes from Chapters 6-9 Chap. 6. Significance Tests Chap. 7: Comparing Two Groups Chap. 8: Contingency Tables (Categorical.
Chapter 9: Introduction to the t statistic
SW388R7 Data Analysis & Computers II Slide 1 Multiple Regression – Basic Relationships Purpose of multiple regression Different types of multiple regression.
Relationships Among Variables
Chapter 12 Inferential Statistics Gay, Mills, and Airasian
1 of 27 PSYC 4310/6310 Advanced Experimental Methods and Statistics © 2013, Michael Kalsher Michael J. Kalsher Department of Cognitive Science Adv. Experimental.
Sensory Evaluation Technique
Accuracy Assessment. 2 Because it is not practical to test every pixel in the classification image, a representative sample of reference points in the.
Statistics for the Behavioral Sciences (5 th ed.) Gravetter & Wallnau Chapter 17 The Chi-Square Statistic: Tests for Goodness of Fit and Independence University.
Statistical Hypothesis Testing. Suppose you have a random variable X ( number of vehicle accidents in a year, stock market returns, time between el nino.
LEARNING PROGRAMME Hypothesis testing Intermediate Training in Quantitative Analysis Bangkok November 2007.
1 STATISTICAL HYPOTHESES AND THEIR VERIFICATION Kazimieras Pukėnas.
Hypothesis Testing Charity I. Mulig. Variable A variable is any property or quantity that can take on different values. Variables may take on discrete.
CENTRE FOR INNOVATION, RESEARCH AND COMPETENCE IN THE LEARNING ECONOMY Session 2: Basic techniques for innovation data analysis. Part I: Statistical inferences.
Multiple Discriminant Analysis and Logistic Regression.
Key Features and Results
Multivariate Statistical Data Analysis with Its Applications
Multivariate Data Analysis Chapter 8 - Canonical Correlation Analysis.
Statistical Decision Making. Almost all problems in statistics can be formulated as a problem of making a decision. That is given some data observed from.
Learning Objectives Copyright © 2002 South-Western/Thomson Learning Multivariate Data Analysis CHAPTER seventeen.
Educational Research: Competencies for Analysis and Application, 9 th edition. Gay, Mills, & Airasian © 2009 Pearson Education, Inc. All rights reserved.
Statistical analysis Prepared and gathered by Alireza Yousefy(Ph.D)
Correlational Research Chapter Fifteen Bring Schraw et al.
Chapter 1 Measurement, Statistics, and Research. What is Measurement? Measurement is the process of comparing a value to a standard Measurement is the.
Hypothesis Testing A procedure for determining which of two (or more) mutually exclusive statements is more likely true We classify hypothesis tests in.
Inference and Inferential Statistics Methods of Educational Research EDU 660.
Logistic Regression Database Marketing Instructor: N. Kumar.
Logistic (regression) single and multiple. Overview  Defined: A model for predicting one variable from other variable(s).  Variables:IV(s) is continuous/categorical,
Chapter 4 Linear Regression 1. Introduction Managerial decisions are often based on the relationship between two or more variables. For example, after.
Discriminant Analysis Discriminant analysis is a technique for analyzing data when the criterion or dependent variable is categorical and the predictor.
Kano Model & Multivariate Statistics Dr. Surej P John.
Research Seminars in IT in Education (MIT6003) Quantitative Educational Research Design 2 Dr Jacky Pow.
Multiple Discriminant Analysis
Educational Research Chapter 13 Inferential Statistics Gay, Mills, and Airasian 10 th Edition.
ITEC6310 Research Methods in Information Technology Instructor: Prof. Z. Yang Course Website: c6310.htm Office:
Copyright © 2011 by The McGraw-Hill Companies, Inc. All rights reserved. McGraw-Hill/Irwin Model Building and Model Diagnostics Chapter 15.
Chapter 10 Copyright © Allyn & Bacon 2008 This multimedia product and its contents are protected under copyright law. The following are prohibited by law:
IMPORTANCE OF STATISTICS MR.CHITHRAVEL.V ASST.PROFESSOR ACN.
Linear Discriminant Analysis and Logistic Regression.
Logistic Regression Analysis Gerrit Rooks
Brian Lukoff Stanford University October 13, 2006.
1 Chi-square Test Dr. T. T. Kachwala. Using the Chi-Square Test 2 The following are the two Applications: 1. Chi square as a test of Independence 2.Chi.
Outline of Today’s Discussion 1.The Chi-Square Test of Independence 2.The Chi-Square Test of Goodness of Fit.
26134 Business Statistics Week 4 Tutorial Simple Linear Regression Key concepts in this tutorial are listed below 1. Detecting.
Review: Stages in Research Process Formulate Problem Determine Research Design Determine Data Collection Method Design Data Collection Forms Design Sample.
Chapter 13 Understanding research results: statistical inference.
Chapter Seventeen Copyright © 2004 John Wiley & Sons, Inc. Multivariate Data Analysis.
Data Analysis: Statistics for Item Interactions. Purpose To provide a broad overview of statistical analyses appropriate for exploring interactions and.
Chapter 7: Hypothesis Testing. Learning Objectives Describe the process of hypothesis testing Correctly state hypotheses Distinguish between one-tailed.
Educational Research Inferential Statistics Chapter th Chapter 12- 8th Gay and Airasian.
DISCRIMINANT ANALYSIS. Discriminant Analysis  Discriminant analysis builds a predictive model for group membership. The model is composed of a discriminant.
STA248 week 121 Bootstrap Test for Pairs of Means of a Non-Normal Population – small samples Suppose X 1, …, X n are iid from some distribution independent.
LOGISTIC REGRESSION. Purpose  Logistical regression is regularly used when there are only two categories of the dependent variable and there is a mixture.
26134 Business Statistics Week 4 Tutorial Simple Linear Regression Key concepts in this tutorial are listed below 1. Detecting.
Logistic Regression: Regression with a Binary Dependent Variable.
Chapter 8 Introducing Inferential Statistics.
BINARY LOGISTIC REGRESSION
Analysis of Variance and Covariance
Using sensory analysis in Food Technology
Chapter 6 Logistic Regression: Regression with a Binary Dependent Variable Copyright © 2010 Pearson Education, Inc., publishing as Prentice-Hall.
Presentation transcript:

Comparison of discriminant vs. logistic regression analyses for predicting consumer acceptance and purchase decision Darryl Holliday, Ashley Bond, Witoon Prinyawiwatkul Department of Food Science, Louisiana State University, Baton Rouge, LA INTRODUCTION Consumer sensory evaluation is defined as the use of consumers to establish food preferences. Consumer research is quintessential for food companies in determining consumer acceptance and purchase decision prior to launching a new product or when mimicking a competitor’s product. Categorical scales are often used for assessing consumer acceptance and purchase decision. Sensory analysts prefer the 9-point hedonic scale (1=extremely dislike; 5=neither like nor dislike; 9=extremely like) while business and marketing analysts prefer the binomial scale (yes or no). The 9-point hedonic scale evaluates acceptability on consumer sensory attributes while the binomial scale measures overall product acceptance and purchase intent. The correlation of these two scales gives insight into what the most discriminating attributes for a product are and is used to determine product acceptance and purchase decision and is useful for further product refinement allowing for a more accurate identification of consumers’ hook therefore increasing the chance for product success. Regression analysis is often used to predict hedonic acceptability. However, it requires the assumption of normality and variance homogeneity which is not always met due to skewness of sensory data, type of scale used, and heterogeneity of consumer responses The two analysis methods used were logistic regression analysis (LRA) and predictive discriminant analysis (PDA). Logistic regression attempts to fit the categorical variables which in most cases the dependant variable is dichotomous (yes/no) and the independent variables are quantitative or categorical. Meanwhile, PDA is used for classification prediction and to predict consumer acceptance and purchase decision. Whether or not both tests are to be used is up to the researcher since PDA and LRA are likely to provide similar results. However, PDA is more robust if multivariate normality of consumer data is met. MATERIALS AND METHODS Materials: The data tested came from a three-component constrained simplex lattice mixture design butter cake taste test using 300 consumers following the balanced incomplete block design and was only used as a model. Methods: Two sets of attributes for each cake sample were measured. The first set of attributes were evaluated using the 9-point hedonic scale and included visual puffiness, appearance, color, taste, texture, moistness, and overall liking. However, the second set of attributes was evaluated using the binomial scale and measured the panelists’ overall acceptance of the sample and the purchase intent based both the hedonic scores for the first set of attributes and the overall acceptance. ercent h Predictive Discriminant Analysis (PDA) and Logistic Regression Analysis (LRA) were used to determine which sensory attributes were most critical to the products’ acceptance and panelists’ purchase decision. Also, the overall percent hit rate, sensitivity (event), and specificity (non-event) were analyzed using PDA and LRA. Statistical Analysis: All data was analyzed at  = 0.05 using RESULTS AND DISCUSSIONS Table 3. Results for full model percent hit rate through LRA and PDA based on consumer acceptance and purchase decision/intent Table 5. Skewness towards sensitivity and specificity for LRA and PDA based on consumer acceptance and purchase decision/intent CONCLUSIONS 1. This study demonstrated LRA and PDA give similar but not identical results. 2. Caution should be taken when interpreting results obtained from these two techniques. 3. Subtle difference may result when PDA is used in lieu of LRA (or vise versa). 4. Both methods should probably be used to gain the most information which will limit any potential new product entrance catastrophes. ABSTRACT This study was to compare PDA and LRA for predicting consumer acceptance and purchase decisions. Consumers (n=300) evaluated butter cake formulations (used as a model) for liking of visual puffiness, appearance, color, taste, texture, moistness, and overall liking on a 9-point hedonic scale, and for overall acceptance and purchase decision on a binomial (yes/no) scale. Hit rate (overall), sensitivity (event), and specificity (nonevent) were analyzed using PDA and LRA (α=0.05). Liking data were not multivariate-normally distributed and skewed toward positive liking. Hit rate (% correct classification) for acceptance and purchase decision was, respectively, 86.9 and 84.5 from LRA, and 85.5 and 80.9 from PDA. This implied that LRA was more effective in prediction/classification; however, sensitivity and specificity results indicated that LRA was biased toward a larger sample size. Based on % hit rate derived from both full- and single-variable models, both PDA and LRA identified overall liking, texture, and moistness as attributes dictating acceptance and purchase decisions. Based on the Wald's chi-squared test statistics from the LRA full model, moistness was, however, excluded from the model (prob. > 0.1). This implied that the hit rate should not be used to indicate the model fitting in LRA. This study demonstrated that PDA and LRA give similar but not identical results. Caution should be taken when interpreting results obtained from these two techniques. Some subtle differences may result when PDA is used in lieu of LRA. Perhaps both should be used to gain more useful information. AcceptancePurchase Decision LRA PDA AcceptancePurchase Intent LRASensitivitySpecificity PDASensitivity sensitivity = event specificity = non-event cceptance having a higher percent for sensitivity from both LRA and PDA. Meanwhile, purchase intent has a higher value for sensitivity from PDA and for specificity from LRA. The results imply that LRA was more effective in prediction and classification. LRA, however, is known to have bias lean towards sensitivity or specificity depending on which has the larger number of observations. But, based on hit rates, both PDA and LRA identified the same attributes dictating acceptance and purchase decision. The attributes that verall liking, texture, moistness. Table 3 and 5, above, illustrate acceptance having a higher percent for sensitivity from both LRA and PDA. Meanwhile, purchase intent has a higher value for sensitivity from PDA and for specificity from LRA. The results imply that LRA was more effective in prediction and classification. LRA, however, is known to have bias lean towards sensitivity or specificity depending on which has the larger number of observations. But, based on hit rates, both PDA and LRA identified the same attributes dictating acceptance and purchase decision. The attributes that were most critical to the products’ acceptance and panelists’ purchase decision were overall liking, texture, moistness. Table 1. Comparison of percent hit rate for each attribute using LRA and PDA statistical analysis for consumer acceptance AttributeLRA / %hit ratePDA / %hit rate Visual Puffiness Appearance Color Taste Texture Moistness Overall Liking Table 2. Comparison of percent hit rate for each attribute using LRA and PDA statistical analysis for consumer purchase decision/intent AttributeLRA / %hit ratePDA / %hit rate Visual Puffiness Appearance Color63.0 Taste Texture80.7 Moistness Overall Liking84.3 The single-variable model was mixed when comparing LRA and PDA values for variables affecting acceptability or purchase intent. ConsumerAcceptance PurchaseIntent VariablesPr > ChiSqOdds RatioPr > ChiSqOdds Ratio Visual Puffiness Appearance Color Taste Texture < Moistness Overall Liking< < Table 4. The acceptance and purchase intent of full model for LRA