Download February 21

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Mathematics 243
Completely Randomized Designs
Day 13 - Fat Tuesday
1. Completely randomized design.
2. Four examples: dolphins, distracted drivers, yawning, First-year seminars
Improve
Not
Yawn
Not
Dolphins
10
5
Seed
10
24
Control
3
12
Missed
Made
None
4
12
Cell
7
17
Retained
Dropped
Passenger
2
22
FYS
212
84
None
113
49
3. General form
”Success”
”Failure”
Treatment A
a
c
Treatment B
b
d
Null Hypothesis: There is no relationship between success rate and treatment (the true success rate is independent
of treatment)
4. Exact test. The p value of the test is the probability of getting results at least as extreme (far away from the
expected) as these if the null hypothesis is true.
Equivalently, if a + b successes are distributed randomly to the two treatment groups A and B, how likly is it that
the result would be at least as extreme as a successes in group A?
5. The hypergeometric distribution revisited.
m white balls (successes)
n black balls (failures)
a simple random sample of k balls are chosen (treatment A)
x is the number of white balls in the sample (treatment A, successes)
function (& parameters)
explanation
dhyper(x,m,n,k)
returns P(X = x)
phyper(q,m,n,k)
returns P(X ≤ q)
rhyper(nn,m,n,k)
makes nn random draws of the random variable X and
returns them in a vector.
6. Dolphin example. m = 13, n = 17, k = 15
> 1-phyper(9,13,17,15)
[1] 0.01266384
> dolphins=matrix(c(10,3,5,12),nrow=2,byrow=T)
> fisher.test(dolphins,alt='greater')
Fisher's Exact Test for Count Data
data: dolphins
p-value = 0.01266
alternative hypothesis: true odds ratio is greater than 1
95 percent confidence interval:
1.531074
Inf
sample estimates:
odds ratio
7.375228
Related documents