Path: janda.org/c10 > Syllabus > Outline > Topics and Readings > Correlating Two Continuous Variables
Linear Regression > Dropping Washington, D.C.

 The Effect of Dropping Washington, D.C. from the Analysis: a smaller correlation SPSS offers methods for selecting cases for analysis. We will learn them later. This simple syntax command "selects" only those cases (states) that voter greater than 15% in 1980. That excludes Washington, D.C. from the analysis. select if (reagan80 gt 15). <--drops D.C. by selecting only states that voted over 15% fo Reagan) ``` Plot of REAGAN84 with REAGAN80 ++----+----+----+----+----+----+----+----+----+----+----+----+----++ 100+ + | | | | | R | | 87.5+ + | | P | | C | | T | | 75+ 1 + V | 1 | O | 11 1 1 | T | 1 1 1 | E | 1 2 1 1 | 62.5+ 2 1 1 1 + F | 1 13431 21 | O | 1 1 | R | 111231 | | 1 1 | R 50+ 11 + E | | A | | G | | A | | N 37.5+ + , | | | | 1 | | 9 | | 8 25R + 4 | | | | | | | | 12.5+ + | | | | | | | | 0+ + ++----+----+----+----+----+----+----+----+----+----+----+----+----++ 0 15.38462 30.76923 46.15385 61.53846 76.92308 92.30769 PCT VOTE FOR REAGAN, 1980 50 cases plotted. Regression statistics of REAGAN84 on REAGAN80: Correlation= .85 R-Squared= .73 Intercept = 25.8 Slope = .67``` Note that, by dropping Washington from the analysis, the correlation drops from r=.90 to r=.85, a loss of .05. However, the R-square drops from .81 to .73, a loss of .08.