I would look at the bivariate correlations and see wether the bivariate relationship between the I.V's and dependent hold in the model.

If it stands that the signs are reversed moving from bivariate to multivariate then these are reversals, and a typical explanation for this would be multicollinearity. That you can check with a VIF option in PROC REG's ,model statement.

>From: June <spairz@YAHOO.COM> >Reply-To: June <spairz@YAHOO.COM> >To: SAS-L@LISTSERV.UGA.EDU >Subject: a question of regression >Date: Wed, 1 Sep 2004 16:48:42 -0700 > >Hi, >I have a question about my SAS proc reg, the parameter estimator of >awarenessoverall is negative , but it is positive on >defintinterestoverall these two varialbes are simliar, they should >all have the positive affect on the outcome, how could I understand >the SAS output? > >Thanks for your help. > > > > The REG Procedure > > Model: MODEL1 > >Dependent Variable: Open Open > > >Stepwise Selection: Step 13 > > Variable awareOverall >Entered: R-Square = 0.9452 and C(p) = 4.7213 > > > >Analysis of Variance > > > Sum of Mean > Source DF > Squares Square F Value Pr > F > > Model 13 >1.374133E15 1.057025E14 181.78 <.0001 > Error 137 >7.966555E13 5.815004E11 > Corrected Total 150 >1.453798E15 > > > >Parameter Standard > Variable >Estimate Error Type II SS F Value Pr > F > > Intercept >1414321 460765 5.47882E12 9.42 0.0026 > TScr >5710.41327 377.03060 1.333926E14 229.39 <.0001 > Screen >-12671 1575.12459 3.762963E13 64.71 <.0001 > awareOverall >-3368067 1980743 1.681339E12 2.89 0.0913 > aDefIntOverall >-5631360 2076179 4.278062E12 7.36 0.0075 > DefIntOverall >27376044 4105415 2.585696E13 44.47 <.0001 > T2IntOverall >-7851665 1708028 1.228807E13 21.13 <.0001 > fchoiceOverall >-18462485 3973341 1.255505E13 21.59 <.0001 > t3choiceOverall >10137471 2990400 6.682681E12 11.49 0.0009 > rtownOverall >-2811374 1878402 1.302598E12 2.24 0.1368 > unaidOverall >8268400 1353691 2.169472E13 37.31 <.0001 > seencinemaOverall >12509213 5127576 3.460873E12 5.95 0.0160 > diiOverall >-81807 31610 3.894794E12 6.70 0.0107 > t1familiarOverall >4095927 2287868 1.863771E12 3.21 0.0756

_________________________________________________________________ STOP MORE SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail