Hypothesis Testing The z-test for Proportions
Hypothesis Testing The z-test for Proportions
Hypothesis Testing The z-test for Proportions
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
<strong>The</strong> Data<br />
• Let x1 , x2 , x3 , … , xn denote a sample from a<br />
normal population with mean μ and standard<br />
deviation σ.<br />
• Let<br />
n<br />
∑ xi<br />
i=<br />
1 x = = the sample mean<br />
n<br />
• we want to <strong>test</strong> if the mean, μ, is equal to some<br />
given value μ 0 .<br />
• Obviously if the sample mean is close to μ 0 the<br />
Null <strong>Hypothesis</strong> should be accepted otherwise the<br />
null <strong>Hypothesis</strong> should be rejected.<br />
<strong>The</strong> Test Statistic<br />
• To decide to accept or reject the Null <strong>Hypothesis</strong><br />
(H 0 ) we will use the <strong>test</strong> statistic<br />
x − μ0 x − μ0<br />
z = = =<br />
σ σ<br />
x<br />
x − μ0<br />
n ≈<br />
σ<br />
x − μ0<br />
n<br />
s<br />
n<br />
• If H0 is true we should expect the <strong>test</strong> statistic z to<br />
be close to zero.<br />
• If H 0 is true we should expect the <strong>test</strong> statistic z to<br />
have a standard normal distribution.<br />
• If H A is true we should expect the <strong>test</strong> statistic z to<br />
be different from zero.<br />
<strong>The</strong> sampling distribution of z when H 0 is true:<br />
<strong>The</strong> Standard Normal distribution<br />
Reject H 0<br />
0 z<br />
Accept H 0<br />
Reject H 0<br />
P<br />
P<br />
Reject H 0<br />
α/2<br />
<strong>The</strong> Acceptance region:<br />
[ Accept H 0 when true]<br />
= P[<br />
− zα<br />
/ 2 ≤ z ≤ zα<br />
/ 2]<br />
= 1−<br />
α<br />
[ Reject H when true]<br />
P[<br />
z < −z<br />
or z > z ] = α<br />
0<br />
α/2<br />
0 z<br />
− zα<br />
/ 2 zα<br />
/ 2<br />
Accept H 0<br />
• Acceptance Region<br />
– Accept H0 if: − z ≤ z ≤ z<br />
Reject H 0<br />
= α / 2<br />
α / 2<br />
α / 2<br />
α / 2<br />
• Critical Region<br />
– Reject H0 if: z < −z<br />
z > z<br />
• With this Choice<br />
α / 2 or<br />
α / 2<br />
[ I Error]<br />
[ Reject when true]<br />
H P<br />
P =<br />
Type 0<br />
[ < − > ] = α z z z z<br />
= α / 2<br />
α / 2 or<br />
P<br />
Summary<br />
To <strong>test</strong> mean of a Normal population<br />
H0 : μ = μ0 (some specified value of μ)<br />
Against<br />
HA : μ ≠ μ0<br />
1. Decide on α = P[Type I Error] = the<br />
significance level of the <strong>test</strong> (usual choices<br />
0.05 or 0.01)<br />
7