The Manga Guide to Statistics

3. Nu Hypotheses And Alternative Hypotheses

Your story

reminded me that

there was puing

in the fridge.

I’m glad it

wasn’t stolen.

I said that you

must always set up

a nu hypothesis

and alternative

hypothesis

When you do a

hypothesis test.

Yeah, but what are the

nu hypothesis and

alternative hypothesis?

You said that you

would explain

later, but so far no

explanation has bn

given.

I must confe that it is

quite diicult to explain

these two hypotheses

simply.

WHAT?

170 Chapter 7

Examples of Hypothesis Tests

Tests of difference

between population ratios

Name Example of use

Tests of

independence

Tests of

correlation ratio

Tests of

correlation

Tests of difference

between population means

Estimates whether the value of the Cramer’s coefficient for sex and

desired way of being asked out is zero for a population

Estimates whether the value of the correlation ratio for favorite fashion

brand and age is zero for a population

Estimates whether the correlation coefficient for amount spent on

makeup and amount spent on clothes is zero for a population

Estimates whether allowances are different between high school girls

in Tokyo and Osaka

*

*

Note that two populations are being considered.

Estimates whether the approval rating of cabinet X is different between

voters residing in urban areas and rural areas

*

So this is what

I’ do...

I wi explain what kind

of hypotheses should be

used as nu and alternative

hypotheses, instead of

what nu and alternative

hypotheses are.

Remember this

table I showed

you earlier?

I wi explain

using the

examples in this

table.

Yes, I do

remember.

practical.

That's

Okay!

Let’s Explore the Hypothesis Tests 171

Tests of independence

Null hypothesis The Cramer’s coefficient for sex and desired way of being asked out is 0 for a

population.

Alternative hypothesis The Cramer’s coefficient for sex and desired way of being asked out is greater

than 0 for a population.

Tests of coelation ratio

Null hypothesis The correlation ratio for favorite fashion brand and age is 0 for a population.

Alternative hypothesis The correlation ratio for favorite fashion brand and age is greater than 0 for a

population.

Tests of coelation

Null hypothesis The correlation coefficient for amount spent on makeup and amount spent on

clothes is 0 for a population.

Alternative hypothesis The correlation coefficient for amount spent on makeup and amount spent on

clothes is not 0 for a population.

or

The correlation coefficient for amount spent on makeup and amount spent on

clothes is greater than 0 for a population.

or

The correlation coefficient for amount spent on makeup and amount spent on

clothes is less than 0 for a population.

172 Chapter 7

Tests of dierence betwn population means

Null hypothesis The allowances of high school girls in Tokyo and Osaka are the same.

Alternative hypothesis The allowances of high school girls in Tokyo and Osaka are not the same.

or

The allowances of high school girls in Tokyo are larger than those of high school

girls in Osaka.

or

The allowances of high school girls in Tokyo are smaller than those of high school

girls in Osaka.

Tests of dierence betwn population ratios

Null hypothesis The approval ratings of cabinet X for voters residing in urban areas and rural

areas are the same.

Alternative hypothesis The approval ratings of cabinet X for voters residing in urban areas and rural

areas are not the same.

or

The approval rating of cabinet X for voters residing in urban areas is higher than

that of voters residing in rural areas.

or

The approval rating of cabinet X for voters residing in urban areas is lower than

that of voters residing in rural areas.

Gd

explanation!

Let’s Explore the Hypothesis Tests 173