Don’t forget HW due on Tuesday. Assignment is on web.

14
• Don’t forget HW due on Tuesday. Assignment is on web.

Transcript of Don’t forget HW due on Tuesday. Assignment is on web.

Page 1: Don’t forget HW due on Tuesday. Assignment is on web.

• Don’t forget HW due on Tuesday. Assignment is on web.

Page 2: Don’t forget HW due on Tuesday. Assignment is on web.

Hypothesis testing and p-values (Chapter 9)

We used confidence intervals in two ways:1. To determine an interval of plausible values for

the quantity that we estimate. Level of plausibility is determined by 1-. 90% (=0.1) is less conservative than 95% (=0.05) is less conservative than 99% (=0.01)...

2. To see if a certain value is plausible in light of the data:

If that value was not in the interval, it is not plausible (at certain level of confidence). Zero is a common certain value to test, but not the only one.

Hypothesis tests address the second use directly

Page 3: Don’t forget HW due on Tuesday. Assignment is on web.

Example: Dietary Folate• Data from the Framingham Heart Study

0 200 400 600 800 1000 1200

02

04

06

08

01

00

Dietary Folate (micrograms / day, calorie adjusted to 2000 calorie diet)

Co

un

t

n = 333 Elderly Men

Mean = x = 336.4

Std Dev = s = 193.4

Can we conclude that the mean is greater than 300 at 5% significance? (same as 95% confidence)

Page 4: Don’t forget HW due on Tuesday. Assignment is on web.

Five Components of the Hypothesis test:

1. Null Hypothesis = “What we want to disprove”= “H0” = “H not”= Mean dietary folate in the population

represented by these data is <= 300.

= <= 3002. Alternative Hypothesis

= “What we want to prove”= “HA”= Mean dietary folate in the population

represented by these data is > 300.

= > 300

Page 5: Don’t forget HW due on Tuesday. Assignment is on web.

3. Test Statistic

To test about a mean with a large sample test, the statistic is z = (x – )/(s/sqrt(n))

(i.e. How many standard deviations (of X) away from the hypothesized mean is the observed x?)

4. Significance Level of Test, Rejection Region, and P-value

5. Conclusion

Reject H0 and conclude HA if test stat is in rejection region. Otherwise, “fail to reject” (not same as

concluding H0 – can only cite a “lack of evidence”(think “innocent until proven guilty”)

(Equivalently, reject H0 if p-value is less than .)

Next page

Page 6: Don’t forget HW due on Tuesday. Assignment is on web.

• Significance Level: =1% or 5% or 10%... (smaller is more conservative) (Significance = 1-Confidence)

• Rejection Region:– Reject if test statistic in rejection region.– Rejection region is set by:

• Assume H0 is true “at the boundary”.• Rejection region is set so that the probability of seeing the observed test

statistic or something further from the null hypothesis is less than or equal to

• P-value– Assume H0 is true “at the boundary”.– P-value is the probability of seeing the observed test statistic or

something further from the null hypothesis.– = “observed level of significance”

Note that you reject if the p-value is less than .(Small p-values mean “more observed significance”)

Page 7: Don’t forget HW due on Tuesday. Assignment is on web.

Example:• H0: <=300, HA: >300• z (x-)/(s/sqrt(n))

= (336.4 – 300)/(193.4/sqrt(333))= 3.43

• Significance level = 0.05• When H0 is true, Z~N(0,1). As a result, the cutoff

is z0.05=1.645. (Pr(Z>1.645) = 0.05.)• P-value = Pr(Z>3.43 when true mean is 300) =

0.0003 • Reject. Mean is greater than 300.• Would you reject at significance level 0.0001?

Page 8: Don’t forget HW due on Tuesday. Assignment is on web.

Picture

1.645Test Statisistic

De

nsi

ty

-4 -2 0 2 4

0.0

0.1

0.2

0.3

0.4

Rejection region

Distribution ofZ = (X – 300)/(193.4/sqrt(333))when true mean is 300.

Area to right of 1.645=0.05 = sig level

3.43Area to right of 3.43=0.0003 = p-value

Test statistic

ObservedTest Statistic

Page 9: Don’t forget HW due on Tuesday. Assignment is on web.

One Sided versus Two Sided Tests

• Previous test was “one sided” since we’d only reject if the test statistic is far enough to “one side” (ie. If z > z0.05)

• Two sided tests are more common (my opinion):

H0: =0, HA: does not equal 0

Page 10: Don’t forget HW due on Tuesday. Assignment is on web.

Two Sided Tests (cntd)

Test Statistic (large sample test of mean)

z = (x – )/(s/sqrt(n))

Rejection Region:

reject H0 at signficance level if |z|>z/2

i.e. if z>z/2 or z<-z/2

Note that this “doubles” p-values. See next example.

Page 11: Don’t forget HW due on Tuesday. Assignment is on web.

Example:• H0: =300, HA: doesn’t equal 300• z=(x-)/(s/sqrt(n))

= (336.4 – 300)/(193.4/sqrt(333))= 3.43

• Significance level = 0.05• When H0 is true, Z~N(0,1). As a result, the cutoff

is z0.025=1.96. (Pr(|Z|>1.96)=2*Pr(Z>1.96)=0.05• P-value = Pr(|Z|>3.43 when true mean is 300) =

Pr(Z>3.43) + Pr(Z<-3.43) = 2(0.0003)=0.0006• Reject. Mean is not equal to 300.• Would you reject at significance level 0.0005?

Page 12: Don’t forget HW due on Tuesday. Assignment is on web.

Picture

1.96Test Statisistic

De

nsi

ty

-4 -2 0 2 4

0.0

0.1

0.2

0.3

0.4

Rejection region

Distribution ofZ = (X – 300)/(193.4/sqrt(333))when true mean is 300.

3.43Area to right of 3.43=0.0003

Test statistic

-3.43

Area to left of -3.43=0.0003

1.96

Rejection region

Sig level = area to right of 1.96 + area to the left of -1.96 = 0.05=

Pvalue=0.0006=Pr(|Z|>3.43)

Page 13: Don’t forget HW due on Tuesday. Assignment is on web.

Power and Type 1 and Type 2 Errors

Truth

H0 True

HA True

Action

Fail to Reject H0 Reject H0

correct

correct

Type 1error

Type 2error

Significance level = =Pr( Making type 1 error )

Power = 1–Pr( Making type 2 error )

Page 14: Don’t forget HW due on Tuesday. Assignment is on web.

• Assuming H0 is true, what’s the probability of making a type I error?

• H0 is true means true mean is 0.• This means that the test statistic has a

N(0,1) distribution. • Type I error means reject which means |

test statistic| is greater than z/2.• This has probability .