Example from pp 149-154 of Belsley (1991), Conditioning Diagnostics
Format
A data frame with 28 observations on the following 5 variables.
- year
1947 to 1974
- cons
total consumption, 1958 dollars
- rate
the interest rate (Moody's Aaa)
- dpi
disposable income, 1958 dollars
- d_dpi
annual change in disposable income
References
Belsley, D.A. (1991). Conditioning diagnostics, collinearity and weak data in regression. New York: John Wiley & Sons.
Examples
data(consumption)
ct1 <- with(consumption, c(NA,cons[-length(cons)]))
# compare (5.3)
m1 <- lm(cons ~ ct1 + dpi + rate + d_dpi, data = consumption)
anova(m1)
#> Analysis of Variance Table
#>
#> Response: cons
#> Df Sum Sq Mean Sq F value Pr(>F)
#> ct1 1 300165 300165 23721.4022 < 2.2e-16 ***
#> dpi 1 1653 1653 130.6481 1.006e-10 ***
#> rate 1 24 24 1.9073 0.1811
#> d_dpi 1 10 10 0.7685 0.3902
#> Residuals 22 278 13
#> ---
#> Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# compare exhibit 5.11
with(consumption, cor(cbind(ct1, dpi, rate, d_dpi), use="complete.obs"))
#> ct1 dpi rate d_dpi
#> ct1 1.0000000 0.9973816 0.9746852 0.3138422
#> dpi 0.9973816 1.0000000 0.9669347 0.3765246
#> rate 0.9746852 0.9669347 1.0000000 0.2290986
#> d_dpi 0.3138422 0.3765246 0.2290986 1.0000000
# compare exhibit 5.12
cd<-colldiag(m1)
cd
#> Condition
#> Index Variance Decomposition Proportions
#> ct1 dpi rate d_dpi
#> 1 1.000 0.000 0.000 0.000 0.003
#> 2 3.767 0.000 0.000 0.003 0.178
#> 3 26.437 0.005 0.003 0.827 0.053
#> 4 256.573 0.995 0.997 0.169 0.765
print(cd,fuzz=.3)
#> Condition
#> Index Variance Decomposition Proportions
#> ct1 dpi rate d_dpi
#> 1 1.000 . . . .
#> 2 3.767 . . . .
#> 3 26.437 . . 0.827 .
#> 4 256.573 0.995 0.997 . 0.765