Regression

Question 1

purpose of ordinary least squares regression

Answer 1

technique for finding the best fitting straight line for a set of data

Question 2

when would you use the sum of squares residular

Answer 2

some residuals are positive, some negative

Question 3

simplest model (null model)

Answer 3

uses the mean as model

Question 4

coefficent of determination

Answer 4

amount of variance in the outcome that is explained by the regression line compares to that explained by the mean

Question 5

MSm

Answer 5

How much the model has improved the prediction

Question 6

MSr

Answer 6

level of inaccuracy of the model

Question 7

spearman’s correlation coefficent

Answer 7

non parametric statistica based on ranked data

Question 8

can the coefficient of determination be used to determine causality?

Answer 8

nope

Question 9

square of pearson’s gives you what

Answer 9

portion of squared variance

Question 10

square of spearman’s gives you what

Answer 10

proportion of variance in ranks that 2 variables show

Question 11

can you square kendall’s tau

Answer 11

nope

Question 12

outcome variable

Answer 12

dependent variable

Question 13

predictive variable

Answer 13

independent variable

Question 14

simple regression

Answer 14

1 predictor

Question 15

multiple regression

Answer 15

multiple predictors

Question 16

residulars

Answer 16

difference between what the model predictes and observed data

Question 17

how to assess error in a regression model

Answer 17

sum of squared residulas

Question 18

F-Tests are vased on what

Answer 18

ratio of improvement due to the model

Question 19

degrees of freedom for the sum of squares of model

Answer 19

number of variablesin model

Question 20

degrees of freedom for sum of squares of residula

Answer 20

number of observations - number of parameters being estimated

Question 21

standardized residulars

Answer 21

residulars are converted to z scores

Question 22

studentized residual

Answer 22

unstandardized residula dvidied by an estimate of it’s standard deviation that varies point by point

Question 23

deleted residual

Answer 23

adjusted predicted value - original observed variable

Question 24

cook;s distance

Answer 24

considers the effect of a single case on the model as a whole

Question 25

mahaladi's distance

Answer 25

measure the distance of cases from the mean(s) of the predictor variable(s)

Question 26

what type of distibution does mahaldi's distance have

Answer 26

chi-squared

Question 27

how do you determine degrees of freedom for mahalandi's distance

Answer 27

number of predictors

Question 28

DFBeta

Answer 28

parameter estimated using all cases - estimated when 1 case is excluded

Question 29

DFFit

Answer 29

predicted value for a case when the model is calculated including that case - excluding that case

Question 30

if a case is not influential what would DFFit be

Answer 30

0

Question 31

Covariance ratio

Answer 31

where a case influences the variance of regression parameteres

Question 32

assumptions of linear model

Answer 32

``` additivity ahnd linearity independent errors homoscedasticity normally distributed errors predictors are uncorrelated with external variables variable types ```

Question 33

additivity

Answer 33

combined effect of predictors is best descrived by adding their effects together

Question 34

Durbin-watson test

Answer 34

tests for serial correlation between errors (assumption of independent errors)

Question 35

what does the Durbin Watson test statistic vary from

Answer 35

0 - 4

Question 36

Durbin Watson test statistic of 2

Answer 36

residuals are uncorrelated

Question 37

Durbin watson test statistic > 2

Answer 37

negative correlation

Question 38

Durbin watson test statistic < 2

Answer 38

positive correlation

Question 39

What happens to a linear model if the independent errors assumption is broken

Answer 39

CI and significance tests are invalid | method of least squares still ok

Question 40

types of predictor variables allowed in a linear regression

Answer 40

quantiative or categorical (dichotomous)

Question 41

types of outcome variables allowed in a linear regression

Answer 41

quantitative, continuous, and unbounded

Question 42

unbounded variable

Answer 42

no constrants on the variability of outcome

Question 43

no perfect multicollinearity

Answer 43

no perfect linear relationship between 2+ predictors

Question 44

cross-validation fxn

Answer 44

assess the accuracy of a model across different samples

Question 45

adjusted R^2

Answer 45

tells us how much of the variance in Y is accounted for if the model had to be derived from the population from which the sample was taken

Question 46

Data splitting

Answer 46

split your data then run a regression equation on both halves of your data then compare

Question 47

sample size needed for an expected large effect

Answer 47

77

Question 48

sample size needed for an expected medium effect

Answer 48

160

Question 49

b-value

Answer 49

tells us the gradient of the regression line and the strength of relationship between a predictor and the outcome

Question 50

F

Answer 50

tells us how much variability the model can explain vs. what it doesn't explain

Question 51

hierarchiacal regression

Answer 51

predictors are selected based on past work and research decides in which order to enter predictors into the model (known predictors get entered first)

Question 52

forced entry

Answer 52

all predictors are forced into model simultaneously

Question 53

stepwise regression

Answer 53

decision about order predictors are entered are purely mathmatical

Question 54

suppressor effects

Answer 54

predictor has a singificant effect but only when another variable is held constant

Question 55

forward method has a higher risk of what type of errors

Answer 55

type II

Question 56

akaike information criterion (AIC)

Answer 56

measure of fit which penalizes the model for having more variables

Question 57

perfect coliinearity

Answer 57

at least one predictor is a perfect linear combination of the others (correlation coefficent of 1)

Question 58

As collinearity increases what else increases

Answer 58

standard errors

Question 59

test that looks at collinearity

Answer 59

variance inflation factor (VIF)

Question 60

tolerance

Answer 60

1/VIF

Question 61

When should we be concerned about VIF

Answer 61

when the largest is >10 or the average is >1

Question 62

When should be be concerned about tolerance

Answer 62

<0.2 is potential problem | <0.1 is seirous problem