Psychology 2080A/B Lecture Notes - Lecture 3: Observational Error, Random Effects Model, Arteriosclerosis

• School: Western University
• Department: Psychology
• Course: Psychology 2080A/B
• Professor: Rachel Plouffe

Document Summary

If the other things are allowed to vary freely, they produce what we call measurement error. I(cid:374) (cid:449)hat follo(cid:449)s, (cid:449)e"ll (cid:272)o(cid:374)side(cid:396) (cid:448)a(cid:396)ious sou(cid:396)(cid:272)es of (cid:373)easu(cid:396)e(cid:373)e(cid:374)t error: different ways of measuring reliability are sensitive to different sources of error. Possible to be good at social psych and bad or good at mental imagery two skills are unrelated: test is tapping into 2 separate abilities issue of internal consistency! How should we assess the correlation: assessing the extent to which items on given test measure the same ability or trait, split half reliability, kuder-richardson formula, ch(cid:396)o(cid:374)(cid:271)a(cid:272)h"s alpha. In sci/psych you want to get above . 8 but above . 7 is fairly acceptable: there is no universally acceptable value, considered (cid:862)u(cid:374)idi(cid:373)e(cid:374)sio(cid:374)al(cid:863) Inter-rater reliability: percent agreement between >2 observers. Issue: over-estimates reliability since observers sometimes agree by chance. Using reliability information: e. g. you have a rubber measuring tape and you are using it to assess height of a table several times.