Hypothesis Testing Study Notes

Hypothesis testing • Procedure for deciding whether the outcome of a study supports your research questions • In hypothesis testing you consider the probability that the experimental procedure has NO EFFECT and that the observed result could have occurred by chance alone o IF that probability is sufficiently low: o Reject the notion that the experimental procedure had no effect o Affirm the hypothesis that the procedure did have an effect In hypothesis testing: • You have what’s called a null hypothesis (H0) and an alternate hypothesis (H1) o You use your data to prove or disprove the null Hypothesis Null hypothesis • The null hypothesis is characterised by “=“ (equal to) • It assumes no effect or no difference • If you have the following research question: ‘is mean BSL higher in Australian males or females?’ 🡪 null hypothesis would be = mean fasting BSL is the same in Australian males and females o Mean fasting BSL in Australian males = mean fasting BSL in Australian females Alternate hypothesis (H1) • The alternate hypothesis is characterised by ≠🡪not equal to o Opposite of the Null hypothesis 25 • Your alternative hypothesis can specify a direction i.e. mean fasting BSL is higher in males than in females, but the alternative hypothesis doesn’t have to specify a direction i.e. Mean fasting BSL is not the same in Australian males and females Hypothesis testing process 1. Re-state the study question in terms of H0 and H1 2. Determine the test statistic (t-test, ANOVA, Chi 3. square etc will discuss in lecture 4) 4. Choose the significance level α and determine the cut-off value 5. Compute the test statistic and p-value 6. Draw and state the conclusion: o Reject the null o Do not reject the null Significant level α • α is set beforehand o = level of significance • It is the probability of rejecting the null hypothesis (H0) when the null hypothesis is actually true • 0 You’d want α to be as small as possible 🡪 in stats it is usually set at: o 0 0.05 (i.e. 5% level of significance – most commonly used) or o 0 0.01 (1% level of significance) P value • P-value is the probability of obtaining our results (or more extreme) if the null hypothesis is true • P-value is not the probability that the null hypothesis is true • Smaller p-values tend to discredit the null hypothesis Statistical significance • Statistical significance level will help you decide whether to reject or accept the null hypothesis • P-value: o If p-value < α then the result is statistically significant and you can reject the null hypothesis ▪ i.e. p <0.05 (most commonly) or ▪ p <0.01 • Confidence Interval o If the CI does not include the value of the null hypothesis you can reject the null hypothesis o E.g. H0: mean weight of all baby guinea pigs is 50g o If the sample 95% CI is (25, 45) you reject the null hypothesis as it does not include 50 Relationship between alpha and p-value • Alpha determines p-value but is not the same as p- value. • The significance level alpha is the threshold value that we measure the p-value against. • The significance level alpha is typically set at 5% or 1%. 0 If alpha is set at 5%, then if the p-value <0.05, you reject the null hypothesis. • If alpha level is set at 1%, then if p<0.01 you reject the null hypothesis. Examples 26 One-tailed vs. two-tailed tests • One-tailed: you expect the experimental procedure to have an effect in a particular direction o Test 1 > test 2 o Test 1 < test 2 o E.g. pain threshold is lower in guinea pigs than rats • Two-tailed: you expect the experimental procedure to have an effect but you do not preduct a direction (could be > or <) e.g. pain threshold is different in guinea pigs and rats Types of error