code? Flashcards
emmeans
provides the estimated means for each group
lm()
calculates a linear model
DV ~ IV, data =
multiple predictors are separated by +
slice()
shows a specified section of the data set
e.g. rows 1 - 6 when specify (1:6)
summary()
provides a summary output of the model
provides values for coefficients, r^2, f tests , degrees of freedom
round()
rounds values to a specified number of decimal places
confint()
provides the confidence intervals of the model
z_score()
standardises coefficients
tab_model
formatted full results table
head
shows the top rows of the dataset
t.test()
does a t-test
contrasts()
used when dummy coding the data when you have a factor variable in the model
selects the first group as the baseline
contr.treatment()
specifies dummy coding
in dummy coding in r, what does base =
the level number of the baseline you want (e.g. 2)
how do you specify interactions in R
asterisk or :
which of these - (*), (:), (+), (+ and *) does not specify full model results?
:
scale = F is used to do what?
mean centre
scale(variable name, scale = F)
probe_interactions
simple slopes
only works for catetgorical * continuous interactions and continuous * continuous predictors
cat_plot()
visualises categorical interactions
geom_smooth(method = loeess)
adds a loess line to plot
crPlots()
component-residual plots
for multiple predictors
also known as partial residual plots
hist()
histograms
plots frequency distribution of residuals
residualPlot()
plots residuals vs predicted values
rstudent()
studentised residuals excluding case outlier in diagnostics
rstandard()
standardised residuals including case outlier in diagnostics
hatvalues()
hat values - assesses leverage
cooks.distance()
average distance the predicted y vaues will move if a given case is removed
influence.measures()
DFFit, DFbeta, DFbetas
covratio()
gives covratio values
influence on standard errors
vif()
quantifies the extent to which standard errors are increased by predictor correlations
gives VIF value for each predictor
anova()
applies the F-test for model comparison
- evaluates statistical significance of improvement in variance explained in an outcome with the addition of further predictors (incremental f test)
AIC() / BIC()
compares specific models by comparing values and choose the model with the smaller value
contr.sum()
changes constract scheme from default
contr.sum = sum to zero coding
plot()
plots the model
levels()
provides the levels of the dataset in the order they arise in
contrast()
tests the effects you have specified
pairs()
pairwise comparisons compares all levels of a given predictor with all levels of the other
adjust =
adjusts the p-value and compares the adjusted p value to the original alpha
Boot
takes the fitted model,
f = which bootstrap statistics to compute on each bootstrap sample, (default is f = coef returning the regression coefficients) ,
r = how many bootstrap samples to compute,
ncores = to perform calculations in parallel (default ncores = 1)
glm()
runs a generalised linear model
family =
in glm()
what family of probability distribution you want for DV (what type of variable is it)
family = binomial
for a binary variable
exp()
exponentiates the coefficients (converts log-odds to odds ratio)
test = chisq
in anova, this performs a likelihood ratio test
pwr.t.test
t test power calcuation
can be directional (alternative = less or greater)
or two sided (alternative = two.sided)
pwr.r.test
correlations
pwr.f2.test
for linear models
code for coefficients of a model
coefficients(model)
coef(model)
model$coefficients
model$coef
How to treat data as categorical data?
factor()
data as continuous data?
numeric()
plotMod$simplesslopes
provides simples slopes data and johnson-neyman plot
plotMod$interactplot
provides a simples slopes plot for cat * cont or cont * cont interactions
how to set control group as reference level
data$group <- relevel(data$group, ‘control’)
geom_line()
connects points of each bar to the other in the same group (colour) on a cat_plot
predict()
can be used to get predicted values of y from a model object
how to get residual from a model
model$residuals
resid$model
residuals(model)
predict(model) - model$y
group_by()
groups the data into the different groups you want to measure by
