# Handout17 - Lecture 17 1. Fitted probabilities from...

This preview shows pages 1–4. Sign up to view the full content.

Lecture 17 1. Fitted probabilities from logistic regression 2. Hosmer-Lemeshow lack-of-±t test 3. Sensitivity, speci±city, false positive, false negative 4. Percent correctly predicted from logistic regression 5. ROC curve 1 Logistic regression example: hypertension in NHANES data National Center for Health Statistics posted a tutorial dataset, hypertension.xls , of 1019 NHANES observations for people over age 20. High blood pressure (hypertension) was either by blood pressure or by prescribed medication to treat hypertension. 37% of sample had hypertension. Proc Logistic descending data= NCHS ; class bmi_class; model hypertension = age male age*male bmi_class; bmi_class has 3 values: normal (18 BMI 25), overweight (25 < BMI 30), and obese (30 < BMI). 2

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
The LOGISTIC Procedure Model Information Data Set WORK.A Response Variable hypertension hypertension Number of Response Levels 2 Model binary logit Optimization Technique Fisher’s scoring Number of Observations Read 1019 Number of Observations Used 952 Response Profile Ordered Total Value hypertension Frequency 11 3 1 9 2 0 633 Probability modeled is hypertension=1. NOTE: 67 observations were deleted due to missing values for the response or ex 3 Type 3 Analysis of Effects Wald Effect DF Chi-Square Pr > ChiSq age 1 120.9322 <.0001 male 1 13.3272 0.0003 bmi_class 2 28.9211 <.0001 age*male 1 13.4863 0.0002 Analysis of Maximum Likelihood Estimates Standard Wald Parameter DF Estimate Error Chi-Square Pr > ChiSq Intercept 1 -5.6453 0.4917 131.8290 <.0001 age 1 0.0914 0.00831 120.9322 <.0001 male 1 2.2425 0.6143 13.3272 0.0003 bmi_class 1 Obese 1 0.6313 0.1175 28.8729 <.0001 bmi_class 2 Overwt 1 -0.2912 0.1140 6.5283 0.0106 age*male 1 -0.0386 0.0105 13.4863 0.0002 Will we get any odds ratios? 4
Odds Ratio Estimates Point 95% Wald Effect Estimate Confidence Limits bmi_class 1 Obese vs 3 Normal 2.642 1.745 3.998 bmi_class 2 Overwt vs 3 Normal 1.050 0.702 1.571 Why only these odds ratios? Interpretation: 5 Fitted probabilities from logistic regression Proc Logistic descending data= NCHS ; class bmi_class; model hypertension = age male age*male bmi_class/ output out =B predicted =phat lower =lowerCI upper =upperCI; Logistic regression is on the log scale (log-odds), so ftted probabilities ˆ p i are back-transFormed From log scale.

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

## This note was uploaded on 11/21/2011 for the course PUBH 6470 taught by Professor Williamthomas during the Fall '11 term at University of Florida.

### Page1 / 14

Handout17 - Lecture 17 1. Fitted probabilities from...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online