PDB_Stat_100_Lecture_23_Printable

PDB_Stat_100_Lecture_23_Printable - STA 100 Lecture 23 Paul...

Info iconThis preview shows pages 1–9. Sign up to view the full content.

View Full Document Right Arrow Icon

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

View Full Document Right Arrow Icon

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

View Full Document Right Arrow Icon

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

View Full Document Right Arrow Icon

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

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

Unformatted text preview: STA 100 Lecture 23 Paul Baines Department of Statistics University of California, Davis March 2nd, 2011 Admin for the Day I Homework 6 due today at 5pm! I Project Proposals feedback later today (very short) I Optional status report can be submitted on Monday I Practical Significance: do the results actually matter References for Friday: Rosner, Ch 10.1-10.6 (7th Ed.) References for Wednesday: Rosner, Ch 11 (7th Ed.) χ 2 Goodness of Fit Test Last time we saw how we can compare observed counts to expected expects (under certain assumptions), and use the χ 2-test to see if the assumptions (e.g., independence) are plausible. The idea of comparing what we observe to what we expect under a specific model is very general. It allows us to test the ‘Goodness of Fit’ of our model i.e., are our assumptions plausible? I H : Our model assumptions are correct, vs., I H 1 : Our model assumptions are not correct. Cholera Example Example: From lecture 9, the number of cases of Cholera in each household in an Indian village were recorded. The data look like this: > cholera house cholera_cases 1 1 2 2 3 1 4 1 ... etc ... > table(cholera$cholera_cases) 1 2 3 4 35 32 16 6 1 1 2 3 4 Indian Cholera Cases Cases per household Frequency 5 10 15 20 25 30 35 Cholera: Poisson? Let X i be the number of Cholera cases in house i . Then suppose X i ∼ Poisson ( λ ), with each house being independent. The sample mean of the data was 0.93 cases/household. I Q: Estimate λ . We estimate ˆ λ = 0 . 93. I Q: What is the probability of a house having no cases of Cholera? P ( X i = 0) = e- . 93 = 0 . 394 I Q: How many of the 90 households might we expect to have no cases of Cholera? 90 * . 394 = 35 . 5 Cholera: Poisson or not?--------------------------------------------------------------- Cholera Cases per Household: 1 2 3 4 5+--------------------------------------------------------------- Observed: 35 32 16 6 1--------------------------------------------------------------- Expected if Pois(0.93): 35.5 33.0 15.4 4.8 1.1 0.2--------------------------------------------------------------- It looks like the Poisson assumption is pretty good for this data. Later in the course we will formally answer whether it is good enough. . . (Today is the ‘later in the course’. . . ) Goodness-of-Fit Tests Just as for contingency tables, the χ 2 goodness-of-fit test requires approximations. Each ‘cell’ or ‘bin’ must not have an expected value less than 5. We can combine cells to achieve this. . . Cholera: Poisson or not? The combined version:--------------------------------------------------------------- Cholera Cases per Household: 1 2 3+--------------------------------------------------------------- Observed: 35 32 16 7--------------------------------------------------------------- Expected if Pois(0.93): 35.5 33.0 15.4 6.1--------------------------------------------------------------- X 2 = ( O 1- E 1 ) 2 E 1 + ( O 2- E 2 ) 2 E 2 + ( O 3- E 3 ) 2 E 3 + ( O 4- E 4 ) 2 E 4 = (35- 35...
View Full Document

{[ snackBarMessage ]}

Page1 / 40

PDB_Stat_100_Lecture_23_Printable - STA 100 Lecture 23 Paul...

This preview shows document pages 1 - 9. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online