s20x (version 3.1-40)

eovcheck: Testing for equality of variance plot

Description

Plots the residuals versus the fitted (or predicted) values from a linear model. A horizontal line is drawn at y = 0, reflecting the fact that we expect the residuals to have a mean of zero. An optional lowess line is drawn if smoother is set to TRUE. This can be useful in determining whether a trend still exists in the residuals. An optional pair of lines is drawn at +/- 2 times the standard deviation of the residuals - which is estimated from the Residual Mean Sqare (Within group mean square = WGMS). This can be useful in highlighting potential outliers. If the model has one or two factors and no continous variables, i.e. if it is a oneway or twoway ANOVA model, and levene = TRUE then the P-value from Levene's test for equality variance is displayed in the top left hand corner,as long as the number of observations per group exceeds two.

Usage

eovcheck(x, ...)
# S3 method for formula
eovcheck(
  x,
  data = NULL,
  xlab = "Fitted values",
  ylab = "Residuals",
  col = NULL,
  smoother = FALSE,
  twosd = FALSE,
  levene = FALSE,
  ...
)
# S3 method for lm
eovcheck(x, smoother = FALSE, twosd = FALSE, levene = FALSE, ...)

Arguments

x

A linear model formula. Alternatively, a fitted lm object from a linear model.

...

Optional arguments

data

A data frame in which to evaluate the formula.

xlab

a title for the x axis: see title.

ylab

a title for the y axis: see title.

col

a color for the lowess smoother line.

smoother

if TRUE then a smoothed lowess line will be added to the plot

twosd

if TRUE then horizontal dotted lines will be drawn at +/-2sd

levene

if TRUE then the P-value from Levene's test for equality of variance is displayed

Methods (by class)

• eovcheck(formula): Testing for equality of variance plot

• eovcheck(lm): Testing for equality of variance plot

See Also

levene.test

Examples

# one way ANOVA - oysters
data(oysters.df)
oyster.fit = lm(Oysters ~ Site, data = oysters.df)
eovcheck(oyster.fit)

# Same model as the previous example, but using eovcheck.formula
data(oysters.df)
eovcheck(Oysters ~ Site, data = oysters.df)


# A two-way model without interaction
data(soyabean.df)
soya.fit=lm(yield ~ planttime + cultivar, data = soyabean.df)
eovcheck(soya.fit)

# A two-way model with interaction
data(arousal.df)
arousal.fit = lm(arousal ~ gender * picture, data = arousal.df)
eovcheck(arousal.fit)

# A regression model
data(peru.df)
peru.fit = lm(BP ~ height + weight + age + years, data = peru.df)
eovcheck(peru.fit)


# A time series model
data(airpass.df)
t = 1:144
month = factor(rep(1:12, 12))
airpass.df = data.frame(passengers = airpass.df$passengers, t = t, month = month)
airpass.fit = lm(log(passengers)[-1] ~ t[-1] + month[-1] 
                 + log(passengers)[-144], data  = airpass.df)
eovcheck(airpass.fit)