Lecture 3: hypothesis testing

Question 1

What is a type I / type II error?

Answer 1

Type I error - finding a false positive - actual finding but this is incorrect - the probability of finding a false positive is 0.05 - alpha

Type II error - finding a false negative - there was a finding but not computed due to lack of power - Beta

Correct inference that null was false is 1-Beta
Correct inference that null was right and not rejected 1 - alpha

Question 2

What level is power set to?

Answer 2

0.80 typically

As alpha - probability of type I error - no treasure is set to 0.05

As power - probability of falsely rejecting null hypothesis type II error - set to 0.80

If we trail effect size from past literature - can compute how many participants are needed to be recruited

Can use G* power software/programme to compute sample size calculations

Question 3

What does a p value estimate?

Answer 3

The probability of computing a type 1 error

To reject null hypothesis we need to sample a value that is above 95% CI for the sampling distribution - i.e. there is a 2.5% chance that we sampled the data from this error (comes to 5% considering both low/high values) - therefore if p is set to 0.05 for it to be up to chance - values with low p value indicate a very low probability of sampling that number from our population by chance

Question 4

What are the steps to indicate whether we reject the null?

Answer 4

1 - create the null and alternate hypothesis

2 - sample from the population and compute the correct statistic to estimate the parameter

3 - create the sampling distribution for this statistic

4 - find the rejection area

5 - see if the sample value falls in the rejection area