Andy Field Discovering Statistics - Chapter 8 Key Terms

Front 1

Adjusted Predicted Value

Back 1

A measure of the influence of a given case of data on the model.

Front 2

Adjusted R-2-

Back 2

Tells us how much variance in the outcome would be accounted for if the model had been derived from the population from which the sample was taken.

Front 3

Autocorrelation

Back 3

When the residuals of two observations in a regression model are correlated. That is: when the assumption of independent errors is violated.

Front 4

b(i)

Back 4

The unstandardised regression coefficient, which indicates the strength of a given predictor and the outcome. The change in the outcome associated with a unit change in the predictor.

Front 5

B(i)

Back 5

The standardised regression coefficient, which does the same as unstandardised but in standard deviation changes.

Front 6

Cook's Distance

Back 6

A measure of the overall influence of a case on a model. Values greater than 1 may be cause for concern (i.e. indicate an outlier).

Front 7

Covariance Ratio

Back 7

Measure of whether a case influence the variation of the parameters in a regression model.

Front 8

Cross-validation

Back 8

Assessing the accuracy of a model across different samples.

Front 9

Deleted residual

Back 9

A measure of the influence of a particular case of data.

Front 10

Dummy Variables

Back 10

A way of recoding a categorical variable with more than two levels.

Front 11

Durbin-Watson Test

Back 11

Assesses the assumption of independence for regression models.

Front 12

F-ratio

Back 12

Tests the overall fit of the model in regression & overall differences between group means.

Front 13

Generalisation

Back 13

Ability for a model to be applicable to situations beyond the data that created it.

Front 14

Goodness of fit

Back 14

Index of how well the model fits the data.

Front 15

Hat values

Back 15

Multivariate analogue of the F-Ratio (so good for MANOVA)

Front 16

Heteroscedasticity

Back 16

At each point of the predictor variable there is unequal variance.

Front 17

Hierarchical Regression

Back 17

The order in which predictors are entered into the model is based on prior literature.

Front 18

Independent errors

Back 18

The residuals should be uncorrelated for any given two observations

Front 19

Leverage statistics

Back 19

The influence of the value of the outcome variable on the predictor(s)

Front 20

Mahalanobis distances

Back 20

Measure the influence of a case by examining the distance of a case from the mean of the predictor(s)

Front 21

Mean squares

Back 21

Measure of average variability

Front 22

Model sum of squares

Back 22

total amount of variability for which the model can account

Front 23

Multicollinearity

Back 23

Two or more variables are very closely related

Front 24

multiple r

Back 24

Multiple correlation coefficient.

Front 25

Multiple regression

Back 25

A regression with multiple predictor variables.

Front 26

Ordinary least squares

Back 26

A method of regression where the parameters are calculated according to the method of least squares.

Front 27

Outcome variable

Back 27

that which is being predicted

Front 28

Perfect collinearity

Back 28

When one predictor is perfectly correlated with another (or multiple)

Front 29

Predicted value

Back 29

The value of an outcome based on values of the predictors

Front 30

Predictor variable

Back 30

That which is doing the predicting

Front 31

Residual

Back 31

Error. Difference between the value the model predicts and the value observed in the data.

Front 32

Residual sum of squares

Back 32

same as residual, but broader and regarding deviance. I don't get it either.

Front 33

Shrinkage

Back 33

Loss of predictive power due to sampling, rather than taking data from full population

Front 34

simple regression

Back 34

one predictor, one outcome

Front 35

standardised residuals

Back 35

residuals of a model expressed in standard deviations

Front 36

stepwise regression

Back 36

variables are entered into the model based on a statistical criterion

Front 37

studentised deleted residuals

Back 37

measure of the influence of a particular case of data

Front 38

suppressor effects

Back 38

where a predictor has a significant effect, but only when another variable is held constant (otherwise it suppresses)

Front 39

t-statistic

Back 39

in regression, to test whether a regression coefficient is significantly different from 0

Front 40

tolerance

Back 40

measures multicollinearity

Front 41

total sum of squares

Back 41

measure of the total variability within a set of observations

Front 42

unstandardised residuals

Back 42

residuals of a model expressed in the original units

Front 43

variance inflation factor (VIF)

Back 43

a measure of multicollinearity. when it hits 10 freak out