Imagine you wish to determine the effects of neighborhood watch programs on neighborhood crime rates. You hypothesize that the more time people spend patrolling their neighborhoods, the less crime there should be as a result. So, you gather data from a representative sample of 50 neighborhoods that have neighborhood watches in 2005. For a six-month period (January 1, 2005 through June 30, 2005) you record the number of hours citizens spend patrolling their neighborhoods. You also gather UCR data from the city police departments to compute a neighborhood crime rate for the six-months following (July 1, 2005 through December 31, 2005) patrol data collection. The number of citizens that make up each neighborhood watch is an important control variable ("Size of Watch"), since larger watches should be able to spend more time patrolling, so you gather this information as well. Here are the results of the correlation matrix: Here are the regression results when the model is run with only the independent variable: Regression Resultsa Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) .728 2.164 2.951 .025 Hours Patrolled -.152 .319 -.165 -1.567 .032 a. Dependent Variable: Property Crime Rate R-square = 0.15 Here are the regression results when the model is run with the control variable added: Regression Resultsa Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 2 (Constant) .728 2.898 3.041 .003 Hours Patrolled -.176 .495 -.163 -2.567 .057 Size of Watch .418 .635 .003 .028 .004 a. Dependent Variable: Property Crime Rate R-square = 0.31 Based on these results, what percentage of the variance was explained in the second model?