Find the best model you can for modeling the dependent variable Y. There is one model which is (hopefully) better than any other. You may need to transform variables and/or remove selected cases to find this model. Then again, you may not. There is no guarantee that the optimal model uses all three variables. It may use only two, or even just one.
NOTE: the goal is model specification, not prediction at any cost. But, if you've time, think about the model you would construct if prediction were the overriding goal... In what ways would it be different?
options linesize=72; * there should be 40 cases in the data set below; data unknown; input Y X1 X2 X3; casenum = _N_; * this constructs a variable for tracking the case number; * do any needed transformations, case deletions here; cards; 41.476532 -0.980000 -0.177000 0.517000 15.012185 -0.305000 -0.966000 -0.636000 34.095757 1.117000 -1.148000 0.473000 79.121658 0.011000 1.220000 -1.513000 77.338843 -1.506000 0.566000 -0.112000 63.534287 1.134000 0.175000 0.119000 44.031970 0.269000 0.207000 0.602000 45.967824 -0.381000 0.641000 -0.070000 46.853216 -0.470000 0.348000 1.625000 43.713493 -1.066000 -0.028000 -0.252000 67.552964 -1.248000 0.541000 0.010000 51.097503 1.399000 0.267000 -0.195000 134.468353 0.777000 1.838000 0.362000 25.126183 -0.749000 -1.745000 0.415000 40.356803 0.309000 0.002000 -0.344000 34.808482 -0.757000 -0.151000 -0.883000 30.278860 0.498000 -0.674000 -0.256000 55.885217 1.307000 0.076000 0.947000 47.026038 0.513000 0.676000 0.640000 168.355813 2.035000 1.642000 0.343000 45.198592 -0.480000 0.616000 -2.639000 67.589368 1.012000 0.886000 -2.495000 41.007865 1.169000 -0.853000 0.804000 88.158704 1.291000 1.208000 -1.034000 28.423509 0.670000 -0.834000 -0.120000 20.260507 0.308000 -0.525000 -1.530000 32.261649 0.649000 -0.064000 -0.402000 31.717682 0.055000 -0.078000 -1.797000 28.475147 1.094000 -0.660000 -0.872000 26.104881 0.622000 -1.387000 0.180000 153.581928 1.293000 1.787000 0.373000 29.931081 -0.196000 -0.513000 -2.054000 31.762139 -0.459000 -0.514000 -0.997000 39.063836 1.079000 -1.331000 -2.004000 82.418881 0.634000 1.118000 0.015000 22.347592 0.180000 -0.465000 -0.600000 5.898834 0.094000 -1.108000 -0.832000 125.763731 2.632000 0.337000 1.363000 33.162800 0.177000 0.175000 0.419000 53.716938 -0.366000 0.776000 -1.117000 ;