In: Statistics and Probability
What are influential observations and outliers in regression and how to determine them? Use Excel Analysis ToolPak to solve the following problem.
The following data show the annual revenue ($ millions) and the
estimated team value
($ millions) for the 32 teams in the National Football League
(Forbes website, February
2009).
Team   Revenue   Value
Arizona Cardinals   203   914
Atlanta Falcons   203   872
Baltimore Ravens   226   1062
Buffalo Bills   206   885
Carolina Panthers   221   1040
Chicago Bears   226   1064
Cincinnati Bengals   205   941
Cleveland Browns   220   1035
Dallas Cowboys   269   1612
Denver Broncos   226   1061
Detroit Lions   204   917
Green Bay Packers   218   1023
Houston Texans   239   1125
Indianapolis Colts   203   1076
Jacksonville Jaguars   204   876
Kansas City Chiefs   214   1016
Miami Dolphins   232   1044
Minnesota Vikings   195   839
New England Patriots   282   1324
New Orleans Saints   213   937
New York Giants   214   1178
New York Jets   213   1170
Oakland Raiders   205   861
Philadelphia Eagles   237   1116
Pittsburgh Steelers   216   1015
San Diego Chargers   207   888
San Francisco 49ers   201   865
Seattle Seahawks   215   1010
St Louis Rams   206   929
Tampa Bay Buccaneers   224   1053
Tennessee Titans   216   994
Washington Redskins   327   1538
a.   Develop a scatter diagram with Revenue on the
horizontal axis and Value on the vertical axis.
b.   Develop the estimated regression equation that can
be used to predict team value given the value of annual
revenue.
c. Construct a residual plot of the standardized residuals against
the independent variable.
d. Do the assumptions about the error term and model form seem
reasonable in light of the residual plot?
Soln
a)
Steps for Scatterplot in Excel

b)
Steps for Regression in Excel
Regression Output

Regression Equation
Value = 5.83 Revenue – 252.08
c)

d)
From the plot in part c we can conclude that the assumptions of about error terms are not satisfied and they suffer from heteroskedasticity. (Absence of constant variance leads to heteroskedestacity)