Consider the following computer results, which describe regression analyses involving two independent variables X1 and X2 and a dependent variable Y. Assume that your goal is to assess the relationship of X1 with Y, controlling for the possible confounding effects of X2.
a. Using an appropriate regression coefficient as your measure of association, determine whether confounding exists. Explain.
b. Suppose that confounding was defined to require a comparison of crude versus adjusted (partial) correlation coefficients. What conclusion would you draw? Explain.
c. What is the moral of this example?

Transcribed Image Text:

Regression of Y on X1 and X2 CORRELATION Variable X1 X2 X1 1.0000 0.5000 0.6527 0.50001.0000 0.9494 0.6527 0.494 1.0000 X2 ANALYSIS OF VARIANCE Sum of OF Squares Sourco Model Error Corrected Total F Value Pr> F 108.88 0001 Square 2 268 00000 134.00000 13 16.00000 15 284.00000 1.23077 [Portion of output omitted PARAMETER ESTIMATES Squared Partial Squared Partial Corr Type II Parameter Standard Variable F Estimate intercept X1 x2 t Value Pr>Corr Type l 11.80 .0001 5,00000 2.00000 0.64051 3.12 0.0081 7.00000 0.42366 0.42606 0.64051 0.93 <0001 0.90184 Regression of Y on X1 ANALYSIS OF VARIANCE Sum of DF Squares Mean Square 121.00000 F Value Pr> F Model Error Corrected Total 121,00000 163.00000 284.00000 10.39 0.0061 14 11.64286 15 Portion of output omitted] PARAMETER ESTIMATES Parameter Standard Variable DF Estimate Error tValue Pr> 6.75000 5.50000 120638 5.60 0001 1.70608 X1 3.22 0.0061