Intro to Marketing Research: Correlation and Regression Logistic Regression for Binary Dependent Variables Odds Ratios

What It Is

Logistic Regression is a statistical method used to model the probability of a binary dependent variable (0/1, yes/no, etc.) based on one or more independent variables. A famous example is the study by King and Zeng (2001) on the impact of economic development on democracy, where they used logistic regression to analyze the probability of a country transitioning to democracy. This matters for marketing decision-making as it allows marketers to identify the factors that influence customer behavior and make informed decisions about targeting and segmentation.

Key Terms & Concepts

• Binary Dependent Variable: A variable that can take on only two values (e.g., 0/1, yes/no).
  • Example: A survey question asking if a customer has purchased a product (yes/no).
• Logistic Function: A mathematical function that maps any real number to a value between 0 and 1.
  • Formula: P(Y=1) = 1 / (1 + e^(-z)), where P(Y=1) is the probability of the dependent variable being 1, and e is the base of the natural logarithm.
• Odds Ratio (OR): A measure of the strength and direction of the association between an independent variable and the dependent variable.
  • Formula: OR = e^(β), where β is the coefficient of the independent variable.
• Logit: The logarithm of the odds ratio.
  • Formula: logit(P(Y=1)) = β0 + β1X, where β0 is the intercept, β1 is the coefficient of the independent variable, and X is the independent variable.
• Maximum Likelihood Estimation (MLE): A method of estimating the parameters of a statistical model by maximizing the likelihood of observing the data.
  • Example: A researcher uses MLE to estimate the coefficients of a logistic regression model.
• Hosmer-Lemeshow Test: A goodness-of-fit test for logistic regression models.
  • Formula: H-L statistic = Σ (observed - expected)^2 / (expected * (1 - expected)).
• Area Under the Receiver Operating Characteristic Curve (AUC): A measure of the accuracy of a logistic regression model.
  • Formula: AUC = ∫[0,1] (sensitivity + specificity - 1) d(sensitivity).
• Binary Logistic Regression Assumptions: Independence, linearity, and no multicollinearity.
  • Example: A researcher checks for multicollinearity by calculating the variance inflation factor (VIF).
• Goodness-of-Fit: A measure of how well a logistic regression model fits the data.
  • Example: A researcher uses the Hosmer-Lemeshow test to check for goodness-of-fit.
• Confusion Matrix: A table used to evaluate the accuracy of a logistic regression model.
  • Formula: TP = Σ (predicted = 1 and actual = 1), TN = Σ (predicted = 0 and actual = 0), FP = Σ (predicted = 1 and actual = 0), FN = Σ (predicted = 0 and actual = 1).
• Sensitivity and Specificity: Measures of the accuracy of a logistic regression model.
  • Formula: Sensitivity = TP / (TP + FN), Specificity = TN / (TN + FP).
• Receiver Operating Characteristic (ROC) Curve: A plot of the sensitivity against 1 - specificity.
  • Example: A researcher plots the ROC curve to evaluate the accuracy of a logistic regression model.

Common Misunderstandings

• Misunderstanding: Logistic regression is only used for binary dependent variables.
  • Correction: Logistic regression can be used for multi-class dependent variables using techniques such as multinomial logistic regression.
• Misunderstanding: The odds ratio is a measure of the strength of the association between an independent variable and the dependent variable.
  • Correction: The odds ratio is a measure of the direction and strength of the association, but it is not a measure of the strength alone.
• Misunderstanding: The logit is the logarithm of the probability of the dependent variable being 1.
  • Correction: The logit is the logarithm of the odds ratio.

Quick Application / Identification

A marketing researcher wants to identify the factors that influence the probability of a customer purchasing a product. The researcher collects data on the customer's demographics, behavior, and preferences. Which statistical method would the researcher use to analyze the data?

Answer: Logistic regression.
Explanation: The researcher would use logistic regression to model the probability of the customer purchasing the product based on the independent variables.

Last-Minute Revision

• ⚠️ Assumption of linearity: The relationship between the independent variable and the logit must be linear.
• Maximum likelihood estimation: A method of estimating the parameters of a statistical model by maximizing the likelihood of observing the data.
• Hosmer-Lemeshow test: A goodness-of-fit test for logistic regression models.
• Area under the receiver operating characteristic curve (AUC): A measure of the accuracy of a logistic regression model.
• Confusion matrix: A table used to evaluate the accuracy of a logistic regression model.
• Sensitivity and specificity: Measures of the accuracy of a logistic regression model.
• Receiver operating characteristic (ROC) curve: A plot of the sensitivity against 1 - specificity.
• Odds ratio (OR): A measure of the strength and direction of the association between an independent variable and the dependent variable.
• Logit: The logarithm of the odds ratio.
• Binary logistic regression assumptions: Independence, linearity, and no multicollinearity.
• Goodness-of-fit: A measure of how well a logistic regression model fits the data.
• Maximum likelihood estimation (MLE): A method of estimating the parameters of a statistical model by maximizing the likelihood of observing the data.
• Variance inflation factor (VIF): A measure of multicollinearity between independent variables.