Presentation on theme: "1 Indirect Effects and the Test of Mediation. 2 Example Objective: Compare Path Models to Multiple Regression and Illustrate Concepts of Indirect Effects."— Presentation transcript:
2 Example Objective: Compare Path Models to Multiple Regression and Illustrate Concepts of Indirect Effects and Mediation
3 Example Data: Community Response to Wildfire in California Shrublands
4 View of Data in KeeleyDataCov.xls rich = plant species richness in 1000 m 2 plots tcov = total cover of all species coastran = distance from the coast (transformed) s_age = age of the stand that burned, in years fidx = fire severity index hetero = index of within-plot heterogeneity optabio = index of optimum abiotic conditions (low N, high sand, high rock)
6 Indirect Effects as Causal Tests: Step 1 How do we interpret the observation that plant cover the year after the fires is a function of the age of the stand that burned? Perhaps older stands burn hotter and produce more severe fires.
7 Evaluating the possibility that effect of stand age on plant cover following fire is because older stands burn hotter. Indirect Effects as Causal Tests: Step 2
8 Setting Amos "Analysis Properties" for this example.
9 Testing the hypothesis of mediation - results. Indirect Effects as Causal Tests: Step 3 Chi-square less than 3.84 indicates we pass the test for concluding that fire severity mediates the effect of stand age on vegetation recovery. standardized path coefficients R-square
10 Calculating the Magnitude of the Indirect Effect. Standardized indirect effect of s_age on tcov = 0.45 x -0.44 = -0.198.