6540 Lecture Notes - Lecture 7: Test Statistic, Alternative Hypothesis, Null Hypothesis

• School: University of Canberra
• Department: University of Canberra
• Course: 6540
• Professor: Dr Sumaira Qureshi

Document Summary

To test hypotheses: set a, the sig(cid:374)ifi(cid:272)a(cid:374)(cid:272)e le(cid:448)el, to (cid:271)e a s(cid:373)all (cid:448)alue e. g. (cid:1005)%, 5% o(cid:396) 10%: it is (cid:272)o(cid:373)(cid:373)o(cid:374) to set a = 5% = (cid:1004). (cid:1004)5, whe(cid:374) (cid:449)e test hypotheses: We assume the null hypothesis is true. We gather data and calculate a test statistic. We decide whether to reject the null as being true. If we reject the null as being true, we choose the alternative hypothesis to be true. 4 steps for hypothesis tests: null and alternative hypotheses, test statistic (t. s or t, decision rule: p-value, conclusion. In the format of ho: parameter = hypothesised value: alte(cid:396)(cid:374)ati(cid:448)e hypothesis: Looks like the null, but = has been replaced with one of or does not equal sign. Decision rule: p-value: this is the p(cid:396)o(cid:271)a(cid:271)ility of seei(cid:374)g the o(cid:271)se(cid:396)(cid:448)ed data gi(cid:448)e(cid:374) that the (cid:374)ull hypothesis is t(cid:396)ue. P-value = p(observed test statisticl ho is true: the p-value never gives the probability that ho is true.