Hypothesis Testing (Type 1 & Type 2 Errors)

A hypothesis is just a question we are answering, we never know the full reality just what our data relays and points to being the correct answer.

Ho = null hypothesis

Four Possible Outcomes

• Retain/fail to reject Ho when Ho is true

• Reject Ho when Ho is false

• Type 1 error

  • Rejecting Ho when Ho is true

• Type 2 error

  • Retain/fail to reject the Ho when Ho is false

Retain Ho when Ho is true

Say no effect when there is no effect

Hypothesized sampling distribution is the true distribution

Probability equals 1-a

Type 1 Error

Reject the Ho when it is true

• Say there is an effect when there is no effect

Hypothesized sampling distribution is the true distribution

Probability equals a

False positive

Reject Ho when Ho is false

Say there is an effect when there is an effect

Hypothesized sampling distribution is not the true distribution

Probability equals 1-b

Type 2 Error

Retain Ho when it is false

• Decide no effect when there is an effect

Hypothesized sampling distribution is not the true distribution

Probability equals b

False negative

Power

Ability to detect an effect when there is an effect

Ways to Increase Power

• Increase Alpha

• Increase Effect Size

• Increase Variability

• Increase Sample Size