PSY 301 Lecture Notes - Lecture 17: Models 1, Randomness, Horoscope
Document Summary
Evaluation how you combine data and model to reach an conclusion. Absence of evidence is not the same as evidence of absence: a(cid:271)se(cid:374)(cid:272)e of e(cid:448)ide(cid:374)(cid:272)e (cid:373)ea(cid:374)s (cid:449)e do(cid:374)(cid:859)t ha(cid:448)e data, e(cid:448)ide(cid:374)(cid:272)e of a(cid:271)se(cid:374)(cid:272)e (cid:373)ea(cid:374)s (cid:449)e ha(cid:448)e so(cid:373)e data a(cid:374)d do(cid:374)(cid:859)t see so(cid:373)ethi(cid:374)g (cid:449)e are looki(cid:374)g for. Absence of data (lack of data) > we are not sure > usually means not be able to reject. Evidence of absence > we know something > able to reject some models (because we have data) Yes- imply we can rule out that there was no breach in protocol. We cannot rule out a breach in protocol. Yes: our (cid:862)fair(cid:863) (cid:373)odel allo(cid:449)s for a(cid:374)(cid:455) out(cid:272)o(cid:373)e, but, (cid:1005)(cid:1004) 6(cid:859)s i(cid:374) a ro(cid:449) is (cid:448)er(cid:455) i(cid:373)pro(cid:271)a(cid:271)le. 0. 05 (1 in 20) is the common threshold (not magic just convention) In long term, science relies on a rule of repeatability. if the observation eventually fails to support a model, we reject it.
