represents the relation between one response variable and q< 1 predictor variables
confounding variable
a predictor variable that was not included in the model and is correlated with y and one or more of the q predictor variables
confounding variable bias
when a confounding variable exists, it leads to a misleading description of a possible causal relation between x1 and y
multiple correlation
a Pearson correlation between yi and a linear function of predictor variables
coefficient of multiple determination
describes the proportion of the response variable variance that can be predicted from the q predictor variables
adjusted squared multiple correlation
squared multiple correlation adjusted to remove some of the positive bias
centering
subtractring from the x1 scores and subtracting from the x2 scores.
simple slope
obtained by factoring x1i out of the and terms
effect coded variable
variable that is assigned values of 1 and -1 (and - if there are more than two categories
semi-partial correlation
correlation between xj and y that statistically removes the linear effects of one or more quantitative variables from xj
partial correlation
correlation between xj and y that statistically removes the linear effects of one or more quantitative variables from xj and y
standardized slope
a slope coefficient that has been computed using standardized response variables and predictor variables
studentized deleted residual
Deleted residuals divided by their standard errors that follow a t-distribution
DFBETAS
the influence of participant i on assessed by comparing the values of with the ith participant included and ith participant omitted from the analysis and then dividing the difference between these two estimates by the standard error of
Cooks' D
a measure of influence that describes the effect of participant i on all n residuals and the least-squares estimates of all parameters
residual plot
scatterplot of the residuals with the xji scores for each predictor variable
multivariate normal distribution
all variables are normally distributed and linearly related, all prediction errors are normally distributed