#### Transcript Statistics - ROHAN Academic Computing

Chapter 12 Multiple Regression Learn…. To use Multiple Regression Analysis to predict a response variable using more than one explanatory variable. Agresti/Franklin Statistics, 1 of 141 Section 12.1 How Can We Use Several Variables to Predict a Response? Agresti/Franklin Statistics, 2 of 141 Regression Models The model that contains only two variables, x and y, is called a bivariate model Agresti/Franklin Statistics, 3 of 141 Regression Models The regression equation for the bivariate model is: x y Agresti/Franklin Statistics, 4 of 141 Regression Models Suppose there are two predictors, denoted by x1 and x2 This is called a multiple regression model Agresti/Franklin Statistics, 5 of 141 Regression Models The regression equation for this multiple regression model with two predictors is: x x y 1 1 Agresti/Franklin Statistics, 6 of 141 2 2 Multiple Regression Model The multiple regression model relates the mean µy of a quantitative response variable y to a set of explanatory variables x1, x2,…. Agresti/Franklin Statistics, 7 of 141 Multiple Regression Model Example: For three explanatory variables, the multiple regression equation is: x x x y 1 1 2 2 Agresti/Franklin Statistics, 8 of 141 3 3 Multiple Regression Model Example: The sample prediction equation with three explanatory variables is: yˆ a b x b x b x 1 1 2 2 Agresti/Franklin Statistics, 9 of 141 3 3 Example: Predicting Selling Price Using House and Lot Size The data set “house selling prices” contains observations on 100 home sales in Florida in November 2003 A multiple regression analysis was done with selling price as the response variable and with house size and lot size as the explanatory variables Agresti/Franklin Statistics, 10 of 141 Example: Predicting Selling Price Using House and Lot Size Output from the analysis: Agresti/Franklin Statistics, 11 of 141 Example: Predicting Selling Price Using House and Lot Size Prediction Equation: yˆ 10,536 53.8 x 2.84 x 1 2 where y = selling price, x1=house size and x2 = lot size Agresti/Franklin Statistics, 12 of 141 Example: Predicting Selling Price Using House and Lot Size One house listed in the data set had house size = 1240 square feet, lot size = 18,000 square feet and selling price = $145,000 Find its predicted selling price: yˆ 10,536 53.8(1240) 2.84(18,000) 107,276 Agresti/Franklin Statistics, 13 of 141 Example: Predicting Selling Price Using House and Lot Size Find its residual: y yˆ 145,000 107,276 37,724 The residual tells us that the actual selling price was $37,724 higher than predicted Agresti/Franklin Statistics, 14 of 141 The Number of Explanatory Variables You should not use many explanatory variables in a multiple regression model unless you have lots of data A rough guideline is that the sample size n should be at least 10 times the number of explanatory variables Agresti/Franklin Statistics, 15 of 141 Plotting Relationships Always look at the data before doing a multiple regression Most software has the option of constructing scatterplots on a single graph for each pair of variables • This is called a scatterplot matrix Agresti/Franklin Statistics, 16 of 141 Plotting Relationships Agresti/Franklin Statistics, 17 of 141 Interpretation of Multiple Regression Coefficients The simplest way to interpret a multiple regression equation looks at it in two dimensions as a function of a single explanatory variable We can look at it this way by fixing values for the other explanatory variable(s) Agresti/Franklin Statistics, 18 of 141 Interpretation of Multiple Regression Coefficients Example using the housing data: Suppose we fix x1 = house size at 2000 square feet The prediction equation becomes: yˆ 10,536 53.8(2000) 2.84x 2 97,022 2.84x 2 Agresti/Franklin Statistics, 19 of 141 Interpretation of Multiple Regression Coefficients Since the slope coefficient of x2 is 2.84, the predicted selling price for 2000 square foot houses increases by $2.84 for every square foot increase in lot size For a 1000 square-foot increase in lot size, the predicted selling price of 2000 sq. ft. houses increases by 1000(2.84) = $2840 Agresti/Franklin Statistics, 20 of 141 Interpretation of Multiple Regression Coefficients Example using the housing data: Suppose we fix x2 = lot size at 30,000 square feet The prediction equation becomes: yˆ 10,536 53.8 x 2.84(30,000) 1 74,676 53.8x 1 Agresti/Franklin Statistics, 21 of 141 Interpretation of Multiple Regression Coefficients Since the slope coefficient of x1 is 53.8, the predicted selling price for houses with a lot size of 30,000 sq. ft. increases by $53.80 for every square foot increase in house size Agresti/Franklin Statistics, 22 of 141 Interpretation of Multiple Regression Coefficients In summary, an increase of a square foot in house size has a larger impact on the selling price ($53.80) than an increase of a square foot in lot size ($2.84) We can compare slopes for these explanatory variables because their units of measurement are the same (square feet) Slopes cannot be compared when the units differ Agresti/Franklin Statistics, 23 of 141 Summarizing the Effect While Controlling for a Variable The multiple regression model assumes that the slope for a particular explanatory variable is identical for all fixed values of the other explanatory variables Agresti/Franklin Statistics, 24 of 141 Summarizing the Effect While Controlling for a Variable For example, the coefficient of x1 in the prediction equation: yˆ 10,536 53.8 x 2.84 x 1 2 is 53.8 regardless of whether we plug in x2 = 10,000 or x2 = 30,000 or x2 = 50,000 Agresti/Franklin Statistics, 25 of 141 Summarizing the Effect While Controlling for a Variable Agresti/Franklin Statistics, 26 of 141 Slopes in Multiple Regression and in Bivariate Regression In multiple regression, a slope describes the effect of an explanatory variable while controlling effects of the other explanatory variables in the model Agresti/Franklin Statistics, 27 of 141 Slopes in Multiple Regression and in Bivariate Regression Bivariate regression has only a single explanatory variable A slope in bivariate regression describes the effect of that variable while ignoring all other possible explanatory variables Agresti/Franklin Statistics, 28 of 141 Importance of Multiple Regression One of the main uses of multiple regression is to identify potential lurking variables and control for them by including them as explanatory variables in the model Agresti/Franklin Statistics, 29 of 141 For all students at Walden Univ., the prediction equation for y = college GPA and x1= H.S. GPA and x2= study time is: Find the predicted college GPA of a student who has a H.S. GPA of 3.5 and who studies 3 hrs. per day. a. 3.67 b. 3.005 c. 3.175 d. 3.4 Agresti/Franklin Statistics, 30 of 141 For all students at Walden Univ., the prediction equation for y = college GPA and x1= H.S. GPA and x2= study time is: For students with fixed study time, what is the change in predicted college GPA when H.S. GPA increases from 3.0 to 4.0? a. 1.13 b. 0.0078 c. 0.643 d. 1.00 Agresti/Franklin Statistics, 31 of 141 Section 12.2 Extending the Correlation and R-Squared for Multiple Regression Agresti/Franklin Statistics, 32 of 141 Multiple Correlation To summarize how well a multiple regression model predicts y, we analyze how well the observed y values correlate with the predicted y values The multiple correlation is the correlation between the observed y values and the predicted y values • It is denoted by R Agresti/Franklin Statistics, 33 of 141 Multiple Correlation For each subject, the regression equation provides a predicted value Each subject has an observed y-value and a predicted y-value Agresti/Franklin Statistics, 34 of 141 Multiple Correlation The correlation computed between all pairs of observed y-values and predicted y-values is the multiple correlation, R The larger the multiple correlation, the better are the predictions of y by the set of explanatory variables Agresti/Franklin Statistics, 35 of 141 Multiple Correlation The R-value always falls between 0 and 1 In this way, the multiple correlation ‘R’ differs from the bivariate correlation ‘r’ between y and a single variable x, which falls between -1 and +1 Agresti/Franklin Statistics, 36 of 141 R-squared For predicting y, the square of R describes the relative improvement from using the prediction equation instead of using the sample mean, y Agresti/Franklin Statistics, 37 of 141 R-squared The error in using the prediction equation to predict y is summarized by the residual sum of squares: ( y yˆ ) 2 Agresti/Franklin Statistics, 38 of 141 R-squared The error in using y to predict y is summarized by the total sum of squares: ( y y) 2 Agresti/Franklin Statistics, 39 of 141 R-squared The proportional reduction in error is: ( y y ) ( y yˆ ) R ( y y) 2 2 2 Agresti/Franklin Statistics, 40 of 141 2 R-squared The better the predictions are using the regression equation, the larger R2 is For multiple regression, R2 is the square of the multiple correlation, R Agresti/Franklin Statistics, 41 of 141 Example: How Well Can We Predict House Selling Prices? For the 100 observations on y = selling price, x1 = house size, and x2 = lot size, a table, called the ANOVA (analysis of variance) table was created The table displays the sums of squares in the SS column Agresti/Franklin Statistics, 42 of 141 Example: How Well Can We Predict House Selling Prices? The R2 value can be created from the sums of squares in the table R 2 ( y y ) ( y yˆ ) 2 2 ( y y) 314,433- 90,756 0.711 90,756 2 Agresti/Franklin Statistics, 43 of 141 Example: How Well Can We Predict House Selling Prices? Using house size and lot size together to predict selling price reduces the prediction error by 71%, relative to using y alone to predict selling price Agresti/Franklin Statistics, 44 of 141 Example: How Well Can We Predict House Selling Prices? Find and interpret the multiple correlation R R 0.711 0.84 2 There is a strong association between the observed and the predicted selling prices House size and lot size very much help us to predict selling prices Agresti/Franklin Statistics, 45 of 141 Example: How Well Can We Predict House Selling Prices? If we used a bivariate regression model to predict selling price with house size as the predictor, the r2 value would be 0.58 If we used a bivariate regression model to predict selling price with lot size as the predictor, the r2 value would be 0.51 Agresti/Franklin Statistics, 46 of 141 Example: How Well Can We Predict House Selling Prices? The multiple regression model has R2 0.71, so it provides better predictions than either bivariate model Agresti/Franklin Statistics, 47 of 141 Properties of R2 The previous example showed that R2 for the multiple regression model was larger than r2 for a bivariate model using only one of the explanatory variables A key factor of R2 is that it cannot decrease when predictors are added to a model Agresti/Franklin Statistics, 48 of 141 Properties of R2 R2 falls between 0 and 1 The larger the value, the better the explanatory variables collectively predict y R2 =1 only when all residuals are 0, that is, when all regression predictions are prefect R2 = 0 when the correlation between y and each explanatory variable equals 0 Agresti/Franklin Statistics, 49 of 141 Properties of R2 R2 gets larger, or at worst stays the same, whenever an explanatory variable is added to the multiple regression model The value of R2 does not depend on the units of measurement Agresti/Franklin Statistics, 50 of 141 R2 Values for Various Multiple Regression Models Agresti/Franklin Statistics, 51 of 141 R2 Values for Various Multiple Regression Models The single predictor in the data set that is most strongly associated with y is the house’s real estate tax assessment • (r2 = 0.679) When we add house size as a second predictor, R2 goes up from 0.679 to 0.730 As other predictors are added, R2 continues to go up, but not by much Agresti/Franklin Statistics, 52 of 141 R2 Values for Various Multiple Regression Models R2 does not increase much after a few predictors are in the model When there are many explanatory variables but the correlations among them are strong, once you have included a few of them in the model, R2 usually doesn’t increase much more when you add additional ones Agresti/Franklin Statistics, 53 of 141 R2 Values for Various Multiple Regression Models This does not mean that the additional variables are uncorrelated with the response variable It merely means that they don’t add much new power for predicting y, given the values of the predictors already in the model Agresti/Franklin Statistics, 54 of 141 In a data set used to predict body weight (in pounds), three predictors were used: height, percent body fat and age. Their correlations with total body weight were: Height: 0.745 Percent Body fat: 0.390 Age: -0.187 Which explanatory variable gives by itself the best prediction of weight? a. Height b. Percent body fat c. Age Agresti/Franklin Statistics, 55 of 141 In a data set used to predict body weight (in pounds), three predictors were used: height, percent body fat and age. Their correlations with total body weight were: Height: 0.745 Percent Body fat: 0.390 Age: -0.187 With height as the sole predictor, what is r2? a. .745 b. .555 c. .625 d. .825 Agresti/Franklin Statistics, 56 of 141 In a data set used to predict body weight (in pounds), three predictors were used: height, percent body fat and age. Their correlations with total body weight were: Height: 0.745 Percent Body fat: 0.390 Age: -0.187 If Percent Body Fat is added to the model R2 = 0.66. If Age is then added to the model R2=0.67. Once you know height and % body fat, does age seem to help in predicting weight? a. No b. Yes Agresti/Franklin Statistics, 57 of 141 Section 12.3 How Can We Use Multiple Regression to Make Inferences? Agresti/Franklin Statistics, 58 of 141 Inferences about the Population Assumptions required when using a multiple regression model to make inferences about the population: • The regression equation truly holds for the • population means This implies that there is a straight-line relationship between the mean of y and each explanatory variable, with the same slope at each value of the other predictors Agresti/Franklin Statistics, 59 of 141 Inferences about the Population Assumptions required when using a multiple regression model to make inferences about the population: • The data were gathered using • randomization The response variable y has a normal distribution at each combination of values of the explanatory variables, with the same standard deviation Agresti/Franklin Statistics, 60 of 141 Inferences about Individual Regression Parameters Consider a particular parameter, β1 If β1= 0, the mean of y is identical for all values of x1, at fixed values of the other explanatory variables So, H0: β1= 0 states that y and x1 are statistically independent, controlling for the other variables This means that once the other explanatory variables are in the model, it doesn’t help to have x1 in the model Agresti/Franklin Statistics, 61 of 141 Significance Test about a Multiple Regression Parameter 1. Assumptions: • • • Each explanatory variable has a straightline relation with µy with the same slope for all combinations of values of other predictors in the model Data gathered with randomization Normal distribution for y with same standard deviation at each combination of values of other predictors in model Agresti/Franklin Statistics, 62 of 141 Significance Test about a Multiple Regression Parameter 2. Hypotheses: • • H0: β1= 0 Ha: β1≠ 0 • When H0 is true, y is independent of x1, controlling for the other predictors Agresti/Franklin Statistics, 63 of 141 Significance Test about a Multiple Regression Parameter 3. Test Statistic: b 0 t se 1 Agresti/Franklin Statistics, 64 of 141 Significance Test about a Multiple Regression Parameter 4. P-value: Two-tail probability from tdistribution of values larger than observed t test statistic (in absolute value) The t-distribution has: df = n – number of parameters in the regression equation Agresti/Franklin Statistics, 65 of 141 Significance Test about a Multiple Regression Parameter 5. Conclusion: Interpret P-value; compare to significance level if decision needed Agresti/Franklin Statistics, 66 of 141 Example: What Helps Predict a Female Athlete’s Weight? The “College Athletes” data set comes from a study of 64 University of Georgia female athletes The study measured several physical characteristics, including total body weight in pounds (TBW), height in inches (HGT), the percent of body fat (%BF) and age Agresti/Franklin Statistics, 67 of 141 Example: What Helps Predict a Female Athlete’s Weight? The results of fitting a multiple regression model for predicting weight using the other variables: Agresti/Franklin Statistics, 68 of 141 Example: What Helps Predict a Female Athlete’s Weight? Interpret the effect of age on weight in the multiple regression equation: Let yˆ predicted weight, x1 height, x2 % body fat, and x3 age Then yˆ 97.7 3.43x1 1.36 x2 0.96 x3 Agresti/Franklin Statistics, 69 of 141 Example: What Helps Predict a Female Athlete’s Weight? The slope coefficient of age is -0.96 For athletes having fixed values for x1 and x2, the predicted weight decreases by 0.96 pounds for a 1-year increase in age, and the ages vary only between 17 and 23 Agresti/Franklin Statistics, 70 of 141 Example: What Helps Predict a Female Athlete’s Weight? Run a hypothesis test to determine whether age helps to predict weight, if you already know height and percent body fat Agresti/Franklin Statistics, 71 of 141 Example: What Helps Predict a Female Athlete’s Weight? 1. Assumptions: • • The 64 female athletes were a convenience sample, not a random sample Caution should be taken when making inferences about all female college athletes Agresti/Franklin Statistics, 72 of 141 Example: What Helps Predict a Female Athlete’s Weight? 2. Hypotheses: • H0: β3= 0 • Ha: β3≠ 0 3. Test statistic: b 0 0.960 t 1.48 se 0.648 3 Agresti/Franklin Statistics, 73 of 141 Example: What Helps Predict a Female Athlete’s Weight? 4. P-value: This value is reported in the output as 0.14 5. Conclusion: • The P-value of 0.14 does not give much evidence against the null hypothesis that β3 = 0 • Age does not significantly predict weight if we already know height and % body fat Agresti/Franklin Statistics, 74 of 141 Confidence Interval for a Multiple Regression Parameter A 95% confidence interval for a β slope parameter in multiple regression equals: Estimated slope t (se) .025 The t-score has: df = (n - # of parameters in the model) Agresti/Franklin Statistics, 75 of 141 Example: What’s Plausible for the Effect of Age on Weight? Construct and interpret a 95% CI for β3, the effect of age while controlling for height and % body fat b t ( se) 0.96 2.00(0.648) 3 .025 0.96 1.30 (2.3,0.3) Agresti/Franklin Statistics, 76 of 141 Example: What’s Plausible for the Effect of Age on Weight? At fixed values of x1 and x2, we infer that the population mean of weight changes very little (and maybe not at all) for a 1-year increase in age The confidence interval contains 0 • Age may have no effect on weight, once we control for height and % body fat Agresti/Franklin Statistics, 77 of 141 Estimating Variability Around the Regression Equation A standard deviation parameter, σ, describes variability of the observations around the regression equation Its sample estimate is: s Residual SS df ( y yˆ ) 2 n (# of parameters in reg. eq.) Agresti/Franklin Statistics, 78 of 141 Example: Estimating Variability of Female Athletes’ Weight Anova Table for the “college athletes” data set: Agresti/Franklin Statistics, 79 of 141 Example: Estimating Variability of Female Athletes’ Weight For female athletes at particular values of height, % of body fat, and age, estimate the standard deviation of their weights Begin by finding the Mean Square Error: residual SS 6131.0 s 102.2 df 60 2 Notice that this value (102.2) appears in the MS column in the ANOVA table Agresti/Franklin Statistics, 80 of 141 Example: Estimating Variability of Female Athletes’ Weight The standard deviation is: s 102.2 10.1 This value is also displayed in the ANOVA table For athletes with certain fixed values of height, % body fat, and age, the weights vary with a standard deviation of about 10 pounds Agresti/Franklin Statistics, 81 of 141 Example: Estimating Variability of Female Athletes’ Weight If the conditional distributions of weight are approximately bell-shaped, about 95% of the weight values fall within about 2s = 20 pounds of the true regression line Agresti/Franklin Statistics, 82 of 141 Do the Explanatory Variables Collectively Have an Effect? Example: With 3 predictors in a model, we can check this by testing: H : 0 0 1 2 3 H : At least one parameter 0 a Agresti/Franklin Statistics, 83 of 141 Do the Explanatory Variables Collectively Have an Effect? The test statistic for H0 is denoted by F Mean square for regression F Mean square error Agresti/Franklin Statistics, 84 of 141 Do the Explanatory Variables Collectively Have an Effect? When H0 is true, the expected value of the F test statistic is approximately 1 When H0 is false, F tends to be larger than 1 The larger the F test statistic, the stronger the evidence against H0 Agresti/Franklin Statistics, 85 of 141 Summary of F Test That All βeta Parameters = 0 1. Assumptions: Multiple regression equation holds, data gathered randomly, normal distribution for y with same standard deviation at each combination of predictors Agresti/Franklin Statistics, 86 of 141 Summary of F Test That All βeta Parameters = 0 2. H : 0 0 1 2 3 H : At least one parameter 0 a 3. Test statistic: Mean square for regression F Mean square error Agresti/Franklin Statistics, 87 of 141 Summary of F-Test That All βeta Parameters = 0 4. P-value: Right-tail probability above observed F-test statistic value from Fdistribution with: • • df1 = number of explanatory variables df2 = n – (number of parameters in regression equation) Agresti/Franklin Statistics, 88 of 141 Summary of F-Test That All βeta Parameters = 0 5. Conclusion: The smaller the Pvalue, the stronger the evidence that at least one explanatory variable has an effect on y • If a decision is needed, reject H0 if Pvalue ≤ significance level, such as 0.05 Agresti/Franklin Statistics, 89 of 141 Example: The F-Test for Predictors of Athletes’ Weight For the 64 female college athletes, the regression model for predicting y = weight using x1 = height, x2 = % body fat and x3 = age is summarized in the ANOVA table on the next page Agresti/Franklin Statistics, 90 of 141 Example: The F-Test for Predictors of Athletes’ Weight Agresti/Franklin Statistics, 91 of 141 Example: The F-Test for Predictors of Athletes’ Weight Use the output in the ANOVA table to test the hypothesis: H : 0 0 1 2 3 H : At least one parameter 0 a Agresti/Franklin Statistics, 92 of 141 Example: The F-Test for Predictors of Athletes’ Weight The observed F statistic is 40.48 The corresponding P-value is 0.000 We can reject H0 at the 0.05 significance level We conclude that at least one predictor has an effect on weight Agresti/Franklin Statistics, 93 of 141 Example: The F-Test for Predictors of Athletes’ Weight The F-test tells us that at least one explanatory variable has an effect If the explanatory variables are chosen sensibly, at least one should have some predictive power The F-test result tells us whether there is sufficient evidence to make it worthwhile to consider the individual effects, using t-tests Agresti/Franklin Statistics, 94 of 141 Example: The F-Test for Predictors of Athletes’ Weight The individual t-tests identify which of the variables are significant (controlling for the other variables) Agresti/Franklin Statistics, 95 of 141 Example: The F-Test for Predictors of Athletes’ Weight If a variable turns out not to be significant, it can be removed from the model In this example, ‘age’ can be removed from the model Agresti/Franklin Statistics, 96 of 141 Section 12.4 Checking a Regression Model Using Residual Plots Agresti/Franklin Statistics, 97 of 141 Assumptions for Inference with a Multiple Regression Model • • • The regression equation approximates well the true relationship between the predictors and the mean of y The data were gathered randomly y has a normal distribution with the same standard deviation at each combination of predictors Agresti/Franklin Statistics, 98 of 141 Checking Shape and Detecting Unusual Observations To test Assumption 3 (the conditional distribution of y is normal at any fixed values of the explanatory variables): • Construction a histogram of the standardized • • residuals The histogram should be approximately bellshaped Nearly all the standardized residuals should fall between -3 and +3. Any residual outside these limits is a potential outlier Agresti/Franklin Statistics, 99 of 141 Example: Residuals for House Selling Price For the house selling price data, a MINITAB histogram of the standardized residuals for the multiple regression model predicting selling price by the house size and the lot size was created and is displayed on the following page Agresti/Franklin Statistics, 100 of 141 Example: Residuals for House Selling Price Agresti/Franklin Statistics, 101 of 141 Example: Residuals for House Selling Price The residuals are roughly bell shaped about 0 They fall between about -3 and +3 No severe nonnormality is indicated Agresti/Franklin Statistics, 102 of 141 Plotting Residuals against Each Explanatory Variable Plots of residuals against each explanatory variable help us check for potential problems with the regression model Ideally, the residuals should fluctuate randomly about 0 There should be no obvious change in trend or change in variation as the values of the explanatory variable increases Agresti/Franklin Statistics, 103 of 141 Plotting Residuals against Each Explanatory Variable Agresti/Franklin Statistics, 104 of 141 Section 12.5 How Can Regression Include Categorical Predictors? Agresti/Franklin Statistics, 105 of 141 Indicator Variables Regression models specify categories of a categorical explanatory variable using artificial variables, called indicator variables The indicator variable for a particular category is binary • It equals 1 if the observation falls into that category and it equals 0 otherwise Agresti/Franklin Statistics, 106 of 141 Indicator Variables In the house selling prices data set, the city region in which a house is located is a categorical variable The indicator variable x for region is • x = 1 if house is in NW (northwest region) • x = 0 if house is not in NW Agresti/Franklin Statistics, 107 of 141 Indicator Variables The coefficient β of the indicator variable x is the difference between the mean selling prices for homes in the NW and for homes not in the NW Agresti/Franklin Statistics, 108 of 141 Example: Including Region in Regression for House Selling Price Output from the regression model for selling price of home using house size and region Agresti/Franklin Statistics, 109 of 141 Example: Including Region in Regression for House Selling Price Find and plot the lines showing how predicted selling price varies as a function of house size, for homes in the NW and for homes no in the NW Agresti/Franklin Statistics, 110 of 141 Example: Including Region in Regression for House Selling Price The regression equation from the MINITAB output is: yˆ 15,258 78.0 x 30,569 x 1 Agresti/Franklin Statistics, 111 of 141 2 Example: Including Region in Regression for House Selling Price For homes not in the NW, x2 = 0 The prediction equation then simplifies to: yˆ 15,258 78.0 x 30,569(0) 1 - 15,258 78.0x 1 Agresti/Franklin Statistics, 112 of 141 Example: Including Region in Regression for House Selling Price For homes in the NW, x2 = 1 The prediction equation then simplifies to: yˆ 15,258 78.0 x 30,569(1) 1 15,311 78.0x 1 Agresti/Franklin Statistics, 113 of 141 Example: Including Region in Regression for House Selling Price Agresti/Franklin Statistics, 114 of 141 Example: Including Region in Regression for House Selling Price Both lines have the same slope, 78 For homes in the NW and for homes not in the NW, the predicted selling price increases by $78 for each square-foot increase in house size The figure portrays a separate line for each category of region (NW, not NW) Agresti/Franklin Statistics, 115 of 141 Example: Including Region in Regression for House Selling Price The coefficient of the indicator variable is 30569 For any fixed value of house size, we predict that the selling price is $30,569 higher for homes in the NW Agresti/Franklin Statistics, 116 of 141 Example: Including Region in Regression for House Selling Price The line for homes in the NW is above the line for homes not in the NW The predicted selling price is higher for homes in the NW The P-value of 0.000 for the test for the coefficient of the indicator variable suggests that this difference is statistically significant Agresti/Franklin Statistics, 117 of 141 Is There Interaction? For two explanatory variables, interaction exists between them in their effects on the response variable when the slope of the relationship between µy and one of them changes as the value of the other changes Agresti/Franklin Statistics, 118 of 141 Is There Interaction? Agresti/Franklin Statistics, 119 of 141 Section 12.6 How Can We Model a Categorical Response? Agresti/Franklin Statistics, 120 of 141 Modeling a Categorical Response Variable When y is categorical, a different regression model applies, called a logistic regression Agresti/Franklin Statistics, 121 of 141 Examples of Logistic Regression A voter’s choice in an election (Democrat or Republican), with explanatory variables: annual income, political ideology, religious affiliation, and race Whether a credit card holder pays their bill on time (yes or no), with explanatory variables: family income and the number of months in the past year that the customer paid the bill on time Agresti/Franklin Statistics, 122 of 141 The Logistic Regression Model Denote the possible outcomes for y as 0 and 1 Use the generic terms failure (for outcome = 0) and success (for outcome =1) The population mean of the scores equals the population proportion of ‘1’ outcomes (successes) • That is, µy = p The proportion, p, also represents the probability that a randomly selected subject has a success outcome Agresti/Franklin Statistics, 123 of 141 The Logistic Regression Model The straight-line model is usually inadequate A more realistic model has a curved S-shape instead of a straight-line trend Agresti/Franklin Statistics, 124 of 141 The Logistic Regression Model A regression equation for an Sshaped curve for the probability of success p is: ( x ) e p 1 e ( x ) Agresti/Franklin Statistics, 125 of 141 Example: Annual Income and Having a Travel Credit Card An Italian study with 100 randomly selected Italian adults considered factors that are associated with whether a person possesses at least one travel credit card The table on the next page shows results for the first 15 people on this response variable and on the person’s annual income (in thousands of euros) Agresti/Franklin Statistics, 126 of 141 Example: Annual Income and Having a Travel Credit Card Agresti/Franklin Statistics, 127 of 141 Example: Annual Income and Having a Travel Credit Card Let x = annual income and let y = whether the person possesses a travel credit card (1 = yes, 0 = no) Agresti/Franklin Statistics, 128 of 141 Example: Annual Income and Having a Travel Credit Card Substituting the α and β estimates into the logistic regression model formula yields: ( 3.52 0.105 x ) e pˆ 1 e ( 3.52 0.105 x ) Agresti/Franklin Statistics, 129 of 141 Example: Annual Income and Having a Travel Credit Card Find the estimated probability of possessing a travel credit card at the lowest and highest annual income levels in the sample, which were x = 12 and x = 65 Agresti/Franklin Statistics, 130 of 141 Example: Annual Income and Having a Travel Credit Card For x = 12 thousand euros, the estimated probability of possessing a travel credit card is: 3.52 0.105 ( 12 ) 2.26 e e pˆ 1 e 1 e 0.104 0.09 1.104 3.521.05 ( 12 ) Agresti/Franklin Statistics, 131 of 141 2.26 Example: Annual Income and Having a Travel Credit Card For x = 65 thousand euros, the estimated probability of possessing a travel credit card is: 3.52 0.105 ( 65 ) 3.305 e e pˆ 1 e 1 e 27.2485 0.97 28.2485 3.521.05 ( 65 ) Agresti/Franklin Statistics, 132 of 141 3.305 Example: Annual Income and Having a Travel Credit Card Annual income has a strong positive effect on having a credit card The estimated probability of having a travel credit card changes from 0.09 to 0.97 as annual income changes over its range Agresti/Franklin Statistics, 133 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana A three-variable contingency table from a survey of senior high-school students in shown on the next page The students were asked whether they had ever used: alcohol, cigarettes or marijuana Agresti/Franklin Statistics, 134 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana Agresti/Franklin Statistics, 135 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana Let y indicate marijuana use, coded: (1 = yes, 0 = no) Let x1 be an indicator variable for alcohol use (1 = yes, 0 = no) Let x2 be an indicator variable for cigarette use (1 = yes, 0 = no) Agresti/Franklin Statistics, 136 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana Agresti/Franklin Statistics, 137 of 141 Example:Estimating Proportion of Students Who’ve Used Marijuana The logistic regression prediction equation is: 5.31 2.99 x1 2.85 x2 e pˆ 1 e 5.31 2.99 x1 2.85 x2 Agresti/Franklin Statistics, 138 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana For those who have not used alcohol or cigarettes, x1= x2 = 0 and: 5.31 2.99 ( 0 ) 2.85 ( 0 ) e pˆ 1 e 5.31 2.99 ( 0 ) 2.85 ( 0 ) 0.005 Agresti/Franklin Statistics, 139 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana For those who have used alcohol and cigarettes, x1= x2 = 1 and: 5.31 2.99 ( 1 ) 2.85 ( 1 ) e pˆ 1 e 5.31 2.99 ( 1 ) 2.85 ( 1 ) 0.628 Agresti/Franklin Statistics, 140 of 141 Example: Estimating Proportion of Students Who’ve Used Marijuana The probability that students have tried marijuana seems to depend greatly on whether they’ve used alcohol and cigarettes Agresti/Franklin Statistics, 141 of 141