NURS341 Lecture Notes - Lecture 8: Homoscedasticity, Linear Regression, Scatter Plot

• School: University of Alberta
• Department: Nursing
• Course: NURS341
• Professor: Joanne Olson

Document Summary

Or is there an association? (direction of relationship) Subjects randomly selected: level of measurement for variables is correctly identified so you can correctly choose. Non-parametric: spea(cid:396)(cid:373)a(cid:374)(cid:859)s co(cid:396)(cid:396)elatio(cid:374) coefficie(cid:374)t (cid:894)(cid:396)s or rho) One variable at ordinal level: the test that you choose will produce a correlation coefficient that ranges from -1. 0 to. Effect size and correlation: strength of the relationship (effect size) r < 0. 3, the relationship is weak. A small correlation means you have a small effect size. Tough to find statistically significant weak relationships unless the sample size is large r between 0. 3 and 0. 5, the relationship is moderate. A moderate correlation means you have a moderate effect size. An r of 0. 3 or more is considered clinically significant. r > 0. 5, the relationship is strong. A strong correlation means you have a large effect size. Coefficient of determination (r2: there is a correlation between prenatal stress and gestational diabetes: r =0. 56 r2 = (. 56)2 = . 3136.