**Hypothesis Testing**

Step 1

Step 2

Step 3

1SHT Template

Define the Hypotheses

Calculate the Test Statistics

**State the Conclusions**

**Introduction to**

**Define the Hypotheses**

**Hypothesis testing uses data to test a model**

The first step in hypothesis testing is to translate your claims into hypothesis statements.

Set up a significance level to control the possible frequency of type 1 errors

alpha-level = probability of Type I errors

(typically 0.05)

If the sample value is too close to the reference value, DO NOT REJECT the null hypothesis.

The Null Hypothesis

Begin with the assumption that

the null hypothesis is true...

...similar to the notion of

innocent until proven guilty

Generally means that not enough evidence has been presented to indicate a change has taken place.

NULL the sample value is the same as I expected

ALT the sample value not the same as I expected

NULL the sample value is not bigger than I expected

ALT the sample value is bigger than I expected

…uses comparisons like ‘the same as’…

Implies equality

The type of test depends on what you want to prove…

One-Tailed vs. Two-Tailed

Implies some direction

…uses comparisons like ‘bigger’, ‘less than’, etc….

A hypothesis is a claim or assumption about a population parameter

Every statistical test tests the null hypothesis against an alternate hypothesis.

NULL vs. ALT

Every statistical test tests the

null

hypothesis against an

alternate

hypothesis.

Use this Excel template to store all of your calculations relating to your hypothesis test(s)…

Is the complement of the null hypothesis

...similar to the notion of

guilty beyond a reasonable doubt.

Is generally the hypothesis that the researcher is trying to prove (that a change has taken place, etc.)

Implies that sufficient evidence has been presented to indicate a change has taken place.

Covers the area of the population not covered in the NULL

NULL -

ALT -

remember that they are COMPLEMENTARY

NULL -

ALT -

ONE-TAILED

TWO-TAILED

Z-SCORE

PROBABILITY

...based on the Normal Distribution

Standard Normal Distribution

OBSERVED VALUES

CRITICAL VALUES

need to STANDARDIZE the data first...

**Calculate the Test Statistics**

**State the Conclusions**

...obtained from the data

...obtained from the distribution

How far is “far enough” to reject the NULL?

If the sample value is far enough away from the reference value, REJECT the null hypothesis.

The critical value creates benchmark for decision making.

Define the Hypotheses

Calculate the Test Statistics

State the Conclusions

One-Sample

Hypothesis Tests

for the Mean

Define the Hypotheses

Calculate the Test Statistics

State the Conclusions

One-Sample

Hypothesis Tests

for the Proportion

Summarize salary by team.

**This Prezi was arranged by**

Jennifer J. Edmonds, PhD

Wilkes University

Jennifer J. Edmonds, PhD

Wilkes University

jennifer.edmonds@wilkes.edu

null & alternative

observed & critical values

do not

reject the null?

develop a visual to

let the data guide you...what do you see?

what you see in the data

NULL the sample value is the same as I expected

ALT

NULL

ALT

this

sample value is bigger than

that

sample value

The type of test depends on what you want to prove…

One-Sample vs. Two-Sample

what you see in the data

NULL

ALTERNATIVE

NULL

ALT the sample value is bigger than I expected

a performance standard

a performance standard

a reference value

a reference value

average highway MPG=28.08

compares one sample to a performance standard

compares two samples to one another

pick ONE sample

and compare it to the performance 'standard'

NULL

this

sample value is the same as

that

sample value

ALT

pick TWO samples

and compare their highway fuel efficiency