What are the 5 steps of NHST?
1) Create hypothesis
2) Collect data
3) Evaluate inconsistency between data and null hypothesis
4) Accept or reject null hypothesis
5) Interpret findings
What does H0 refer to?
Null hypothesis
What does H1 refer to?
Research hypothesis
If we are able to reject the null hypothesis in favour for the research hypothesis, can we claim to have evidence for the research hypothesis?
Yes
If we are unable to reject the null hypothesis, can we claim to have evidence for the null hypothesis?
No
Explain Step 3 of NHST
In Step 3 of NHST we need to evaluate the inconsistency between our data and the null hypothesis.
In practice, this is equivalent to calculating the conditional probability of having obtained a sample statistic in a particular region if we assume the null hypothesis is correct.
If values of p > a (critical value alpha) (suggests consistency with null hypothesis), do you reject or fail to reject the null?
Fail to reject null
If values of p < a (critical value alpha) (suggests inconsistency with null hypothesis), do you reject or fail to reject the null?
Reject
What critical value alpha (a) do we use?
a = 0.05 (5%)
What does the p-value refer to?
The conditional probability value associated with chosen sample statistic (e.g sample mean) assuming that the null hypothesis is true
p is the probability that the null hypothesis is true (T/F)
False
Identify the steps of conducting a z-test
1) Formulate a null and research hypothesis
2) Collect data
3) Evaluate inconsistency with H0
4) Reject or fail to reject null
5) Interpret
Explain Step 3 of the z-test for a directional hypothesis (1-tailed)
Evaluate inconsistency with H0:
-Assume null is correct
-Calculate the conditional probability that the mean is higher/lower than the population by converting your sample mean into a z-score
-Work out p-value by calculating the area of interest associated with your z-score
-If p > 0.05, reject null
Explain step 3 of the z-test for a non-directional hypothesis (2-tailed)
Calculate z-scores for both tails and sum up the p-values
Define Type 1 error
When you reject the null when the null is correct
Define Type 2 error
When you fail to reject the null when the null is false
Why might Type 1 errors occur?
Even if your p-value is small there’s still a chance that your data was unusually extreme due to sampling error
Why might Type 2 errors occur?
Often due to a problem in your study, e.g:
-biased sample
-error in experimental task
-too small sample size
etc
