Unit 6 – Inference for Categorical Data: Proportions #-

key/10
L
Notes 2 - Significance Tests for Population Proportions
↓

Significance Tests - The Basics -

Mark
Shooting into the garbage can - __________________________ their claim: _______/10

We want to test this claim!

-

This is where a significance test comes in. We will use the observed data to decide between two claims

P =. 8
Claim one: ____________ p7 8
Claim two: ____________
.

These claims are called hypotheses and the test to determine which one we will accept is a significance test.
-

p
= B
Claim one: _____________=
Hotnot
This claim is called our null hypothesis, __________
We This can be a claim made by someone/something or a statement of “no difference”
startth
P 8
Claim two: _____________
. Ha
This claim is called our alternative hypotheses, ___________
This is the claim we are trying to find evidence for.

B
=. 3
Let’s have them shoot. What was their sample proportion? _________

Questions
Was it exactly the claimed proportion?
No ,
Andrew claimed . 8 and only got . 3

Can it be the sample proportion and their claimed proportion still be correct?

Possibly , maybe the stress a

heckling got to him
How can we decide if the difference is chance variation or if their original claim is incorrect?

Difference >
-

chance variation or Ho is incorrect ?

One-Sided Hypothesis Test

Example 1 Example2
𝐻0 : 𝑝 = 0.7 𝐻0 : 𝑝 = 0.2
𝐻𝐴 : 𝑝 < 0.7 𝐻𝐴 : 𝑝 > 0.2

null hypothes
These are called one-sided because we are look at one side of our ___________________________
a
Notice some similarities:
Hop =
Po
equal
● The null hypothesis is always _____________________.
● The alternative is above or below that ________________.
same value Ha p) po:

● Our hypotheses are based on our population parameter (____) or

P
p(po
Two-Sided Hypotheses Test
Example 1 Example2
𝐻0 : 𝑝 = 0.7 𝐻0 : 𝑝 = 0.2
𝐻𝐴 : 𝑝 ≠ 0.7 𝐻𝐴 : 𝑝 ≠ 0.2
>
- difference
These are called two-sided because we are look for a difference in the claim; we are not sure in which direction
-

it will be different.

***Very Important***
• Hypotheses are developed before a study is conducted. We do not adjust our hypotheses after we have
already gathered data.
• Our data will be above or below the claim but if we started off not knowing which direction the data
would trend, we have to keep running the test that way; next time it could trend the other direction!

A significance tests asks the question:

How likely is it to get a result like this just by chance when the null hypothesis is true?

Zach is an 80% free throw shooter. If Zach is really an 80% free throw shooter, is it possible for him to shoot
50% in a one game?

P(Each80
50%he
shoots
We can find the probability of this happening!! __________________________________ is
•If we get a low probability, then that is good evidence that the null hypothesis is not true. is
throw shorter

+
The probability of the event occurring given an original claim is called the ________________.

P-Value
P would take a value as
Definition: The probability, computed assuming H0 is true, that the statistic (_____)
extreme or more extreme than the one actually observed, in the direction specified by Ha

P and our null hypothesis.

● We compute this probability by using our sample statistic (____)
● We will be calculating z-scores to determine the difference between what we observed and what our
claim is.
● We will then calculate the probability we would get this value or more extreme (in the direction of the
alternative) and make a decision on if that probability is low enough that we can say it is unlikely to
significance level .
occur by chance. This is set by a ________________________________.
Significance Level
2 (alpha)
This “significance level” is ________________ and researchers set it before the test. Choosing it is based on a
few factors, which we will discuss later. But it is always given in a problem.

P-value < <

If our p-value is smaller than α, (________________________) we say that the

① The results of study is

statistically significant
2 The probability of this event
occurring by chance (if the to
is frue) is so low, we
say the difference is not due to
sampling
vaurability
rejectHo
In that case, _____________________ and conclude that there is convincing evidence in favor of the
alternative

p-value) L
If our p-value is larger than α, (________________________) we say that the

① The results
study not
of is
statistically significant
2 The difference between t andpo can be attributed to
sampling
variability

failto toreject
In that case, ______________________________ and conclude that there is not enough convincing evidence in
favor of the alternative

**Does not mean we “accept H0”!! (NEVER SAY THIS). All we showed was that we do not have enough
-

evidence against the claim. We did not prove the claim.

Usual Alpha-Levels
Significance Level 0 0
𝛼 = ________
.
m best
=_______
𝛼 Or 1
𝛼 = ________

“Tail Probability” 1%
___________ 5%
___________ 10 %
___________

Combined shaded
H :
p
=
Po

·
area is
my p-value
Two-Sided Ha :
p + Po
Hypothesis
Test
 M
Ho : P Po area is
my
=

Ha P > Po
: P-value

One-Sided
Hypothesis
Test

Ho P Po
: vale

ii
=

Ha pat :

Tests about a Population Proportion

4-Step Process
● State
Sampling
○ What hypotheses do you want to test and at what significance level?
○ Define p = in context
- distribution
● Plan
○ Random (SRS), Independent (10% Condition), Normal (Large Counts) because we assume
core e

n(pCarmis
-

● Do -
>
○ If conditions are met, calculate the test statistic (z) and find the p-value.
● Conclude n(1-P) >10
○ Make a decision about the hypotheses in the context of the problem. (Interpret the p-value!)
○ Because our p-value of __________ is bigger/smaller than our significance level, we
_____________(reject or fail to reject 𝐻0 ). We do/don’t have convincing evidence that the
population proportion _____________ (context of what this means in the problem)

Test Statistic = z = -

Po

-
 Find the p-value by calculating the probability of getting a z statistic this large or larger in the direction
specified by the alternative hypothesis.

-
·
2 = 0 05
.
𝐻𝐴 : 𝑝 < 𝑝0 𝐻𝐴 : 𝑝 ≠ 𝑝0 𝐻𝐴 : 𝑝 > 𝑝0
=
Ex) Allan reads a report that says 80% of US high school students have a computer at home. He believes the
proportion is smaller than 0.8 at HHS. Allan chooses an SRS of 60 students and records whether they have a
-
computer at home. He wants to test this at a 5% significance level. In his SRS, 41 of the students have a
computer at home. Is this evidence that the proportion of students who have a computer at home at HHS is less
than the national proportion?
at home.
p proportion st who have
=
.

computer
Ho : p =. 3 ↓: 0 05
.

P = =
63

Ha : 8
< .

3
① Random : SRS was stated One proportion
② Independend
: 681 .
1(all High schost)
8) 210 Z-Test
③ Normal 60 (
.

-
.

60 1 2)210 .

z
- 2 .
27

Conclusion
- min
normaledt
BecauseOur p-value < G 0 05
UB M y
LB
- + .

1)
&

7
, ,
we
rejectHo. -

8 -2 27 0 .
,
.

,
Context : We have convincing evidence
that the true proportion for st that .
P = 0 . 0116
have a computer at home is smaller than
O 8 .
Find the p-value by calculating the probability of getting a z statistic this large or larger in the direction
specified by the alternative hypothesis.

𝐻𝐴 : 𝑝 < 𝑝0 𝐻𝐴 : 𝑝 ≠ 𝑝0 𝐻𝐴 : 𝑝 > 𝑝0

Ex) Allan reads a report that says 80% of US high school students have a computer at home. He believes the
proportion is smaller than 0.8 at HHS. Allan chooses an SRS of 60 students and records whether they have a
computer at home. He wants to test this at a 5% significance level. In his SRS, 41 of the students have a
computer at home. Is this evidence that the proportion of students who have a computer at home at HHS is less
than the national proportion?