PSYC105 Lecture Notes - Lecture 7: Statistical Hypothesis Testing, Statistical Significance, Null Hypothesis
11 Oct 2018
School
Department
Course
Professor
Document Summary
Types of t-tests: different rqs = different statistical tests, t-tests are about inspecting mean scores on a numeric (quantitative) variable. Paired samples t-test: similar to independent samples but related (non-independent) groups. Same person over 2 time points or 2 conditions. Is there a decrease in depression from baseline to post-therapy: h(cid:1004): d = (cid:1004, h(cid:1005): d 0, variables: one numeric dv, one categorical iv (with 2 levels) Not 2 groups of 32 babies: same babies measured twice. Si(cid:373)ilarities (cid:449)ithi(cid:374) a(cid:374) i(cid:374)di(cid:448)idual(cid:859)s (cid:271)eha(cid:448)iour = related s(cid:272)ores. If we can take that into account, we get a more precise estimate of the effect. Normality of difference scores: assumption of normality applies to the difference between the two scores. Score at test phase minus score at baseline = difference score. Each participant (baby) has one difference score: need to test normality of that variable, not of the original (source) variables. It(cid:859)s the differe(cid:374)(cid:272)e s(cid:272)ore that(cid:859)s dire(cid:272)tly being tested in the paired samples t-test.
