Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key 1bi: The null hypothesis for these relationships is Ho: slope of y on x =0. Note that this is the same as Ho: slope of y on x = 𝑦̅. 1biv-vi: Page | 1 Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key 1bvii and 1bviii: Always use adjusted r2 Independent Variable Slope P-Value r2 for model Conclusion with respect to Ho Diatom abundance 1.05 <0.0001 .968 Reject Other limpets -1.50 0.365 0 (r2 cannot be negative) Accept Tide Height 8.03 0.0127 .27 Reject Predators 4.48 0.58 0 Accept Page | 2 Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key Hence you would reject the hypotheses that predators and other limpets are associated with the Focal Limpet density and would accept the hypotheses that food and tide height are associated with the density of Focal limpets. Note that the sign of the slope will always be the same as the sign of the tvalue. 1cv: The VIF scores indicate that there is no co-linearity among variables that affects the results of the analysis. The Estimate values represent the model intercept and the slopes of the relationship between the X variables (eg diatom abundance) after accounting for all the X variables. Here there are three terms that are significant (diatom abundance (+), other limpets (-) and tide height (+). All are significant at the p<0.0001 level. In the simple regression analyses tide height was significant but much less so and other limpets was not significant. The density of predators is not significant in the multiple regression or the simple regression analyses. The r2 score for the multiple regression analysis is higher than any of the values for the simple regressions. Page | 3 Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key 1cviii 1cix: A multiple regression yields a linear equation in the slope intercept form: Focal-limpet = -23.39+1.01(diatom abundance)-1.05(other limpets)+.98(tide height)-0.07 (predators). Given that the density of predators was not significant in the model you could drop the predator effect. The answer is around 110- 111 depending on rounding. Page | 4 Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key 2b. use the loess smoother (curved line icon) and play with Lambda to the desired tension. 2c and d. Page | 5 Bio286 Worksheet 4: Multiple Regression, Non-Linear Regression - Answer Key Model: AIC corrected Mean (1 parameter): 𝑦 = 𝑦̅ 447.81 Linear Regression (2 parameter): 𝑦 = 𝑏0 + 𝑏1 + 𝜀 414.75 Gompertz 3 parameter: 𝑦 = 𝑎𝑒 −𝑒 𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑒𝑥𝑝𝑜𝑛𝑒𝑛𝑡 −𝑏(𝑥−𝑐) , where 𝑒 = Biexponential 5 parameter: 𝑦 = 𝑎 + 𝑏𝑒 −𝑐𝑥 + 𝑑𝑒 −𝑓𝑥 , where 𝑒 = 𝑛𝑎𝑡𝑢𝑟𝑎𝑙 𝑒𝑥𝑝𝑜𝑛𝑒𝑛𝑡 348.12 346.81 Page | 6