Probability and significance Flashcards
Our experiment
Investigating the impact of CBT on depression.
Hypothesis: There will be a difference in 25 participants’ depression scores on the BDI before and after a 6 week course of CBT.
Hypothesis
For every study there are two types of hypotheses.
• The original / alternate hypothesis (H1) - there will be an effect.
• There will be a difference in participants’ depression score on the BDI before and after a 6 week course of CBT.
• The null hypothesis (H0) - which states there will not be an effect
• There will be no difference in participants’ depression score on the BDI before and after a 6 week course of CBT.
A statistical test allows us to decide which of the hypotheses is true and this whether or we accept or reject H0
Significance
Lets assumes for a minute we find a difference in depression scores before and after CBT.
• Turns out everyone improved.
• BUT - it doesn’t doesn’t mean the results are of value; that they are significant.
• Could have quite easily been down to chance, fluke or coincidence.
• → We need to use a statistical test.
Probability
All studies employ a significance level in order to check for significant differences or relationships.
• The accepted level of probability is 0.05 (5%) - this is the level at which the hypothesis is accepted.
• If the hypothesis is accepted, you are saying there is a less than 5% probability the results occurred by chance.
• There is a 95% probability that the results occurred due to a manipulation of the IV.
• Some studies employ - p<0.01 or even p<0.001
-effectively-I took 100 pregnancy tests-5 or less are showing negative-95 are showing the correct result-you are 95% certain you are pregnant
Choosing the wrong significance levels
Type 1 error-If a too lenient significance level is used, e.g. 10%. Results in rejecting the null hypothesis that is in fact true.
Null Hyp = True
Alt Hyp = Wrong
Type 2 error-If a too stringent significance level is used, e.g. 1%. Results in accepting a false null hypothesis.
Null Hyp = Wrong
Alt Hyp = True
Establishing significance
Working out whether or not results are significant requires being able to read the table on the right.
You will need several pieces of information.
1. The N value (how many participants)
2. The calculated value (S, M, T, U etc. - depends on the test)
3. Significance level (always 0.05 unless told otherwise)
4. Whether the test is one tailed or two tailed.
i. Directional or non-directional hypothesis.
5. The critical value.
The calculated value must be equal to or less than the critical value in order to be significant