Deck 20: Logistic Regression

Question 1

The z-statistic is used to discover whether a variable is an accurate predictor of an outcome.

Answer 1

True

Question 2

The fundamental difference between linear regression and logistic regression is that in linear regression it is possible to predict possible values,whereas in logistic regression it is the probability that a value will occur which is predicted.

Answer 2

True

Question 3

If logistic regression and discriminant function analysis are different ways of achieving the same result,then why would a researcher prefer logistic regression?

A)There are much more restrictive assumptions associated with logistic regression,meaning calculations are more robust.
B)There are much fewer restrictive assumptions associated with logistic regression,and it is generally more robust.
C)There are much more restrictive assumptions associated with logistic regression,which mean it is a simpler test to administer.
D)None of the above.

Answer 3

B

Question 4

How does logistic regression measure this relationship?

A)By converting the independent variable to probability scores
B)By converting the dependent variable to probability scores
C)By converting the independent variable to a fixed value
D)By converting the dependent variable to a fixed value

Question 5

What type of distribution does the z-statistic follow?

A)Discrete uniform distribution
B)Poisson binomial distribution
C)Degenerate distribution
D)Normal distribution

Question 6

You join a research project midway through and are handed the results of a study in which 199 teachers are asked whether or not they think corporal punishment should be used in the classroom.They are given either a 'yes' or 'no' alternative.23 respondents say 'yes',and 176 say 'no'.What type of model will give you the best prediction for the most frequently occurring outcome?

A)Bottom line
B)Baseline
C)Reduction
D)Initial

Question 7

In the z-statistic equation,when the regression coefficient (b)is large,the standard error can be inflated,resulting in an underestimated z-statistic.What does an inflated standard error increase the probability of?

A)Producing a significant standard deviation
B)Rejecting the null hypothesis
C)Rejecting a predictor as being significant,when in reality it is making a significant contribution to the model
D)Nothing,it has no effect.

Question 8

Probit regression uses similar techniques to logistic regression.

Question 9

There are two types of logistic regression,differentiated by the number of categorical outcomes trying to be predicted.What are they?

A)Egocentric and multivariate
B)Binomial and multinomial
C)Binary and multivariate
D)Binary and multinomial

Question 10

The log-likelihood statistic is analogous to what in multiple regression,as an indicator of the amount of unexplained information after a model has been fitted?

A)The residual sum of squares
B)The number of degrees of freedom present
C)The unique variance score
D)The quality control

Question 11

What does logistic regression measure?

A)The relationship between a categorical dependent variable and a continuous independent variable
B)The relationship between a continuous dependent variable and a categorical independent variable
C)The relationship between multiple continuous dependent variables and one categorical independent variable
D)None of the above

Question 12

A synonym for the z-statistic is the what?

A)Poisson regression
B)Wald statistic
C)Kuder-Richardson formula 20
D)Odds ratio

Question 13

Logistic regression can be used to?

A)Support the calculations made in the chi-square test
B)Predict the outcome of a categorical dependent variable based on one or more predictor variables
C)Confirm the analysis of variance
D)Minimize human error in calculations involving predictor variables

Question 14

Calculating the odds of an event occurring can be a lucrative thing.What is the equation to calculate odds?

A)odds = probability
B)odds = probability of event not occurring / probability of event occurring
C)odds = predictor variable effect size / probability of event occurring
D)odds = probability of event occurring / probability of event not occurring

Question 15

How is $R _ { \mathrm { L } } ^ { 2 }$ calculated?

A) $R _ { \mathrm { L } } ^ { 2 } = \frac { - 2 L L ( \text { model } ) } { - 2 L L ( \text { baseline } ) }$
B) $R _ { \mathrm { L } } ^ { 2 } = \frac { 2 L L ( \text { baseline } ) } { 2 L L ( \text { model } ) }$
C) $R _ { \mathrm { L } } ^ { 2 } = \frac { - 2 L L ( \text { baseline } ) } { - 2 L L ( \text { model } ) }$
D) $R _ { \mathrm { L } } ^ { 2 } = \frac { 2 L L ( \text { model } ) } { 2 L L ( \text { baseline } ) }$

Question 16

Logistic regression can be used for predicting which type of outcome?

A)Continuous
B)Random
C)Criterion
D)Binary

Question 17

Deviance is also referred to as:

A)DLL
B)LL/2
C)2LL
D)-2LL

Question 18

In logistic regression,what is the R-statistic?

A)The partial correlation between the outcome variable and each of the predictor variables
B)The partial correlation between the predictor variables and each of the outcome variables
C)The full correlation between the outcome variables and each of the predictor variables
D)The full correlation between the predictor variables and each of the outcome variables

Question 19

What is the equation for deviance?

A)Deviance = log-likelihood
B)Deviance = log-likelihood/2
C)Deviance = 2 × log-likelihood
D)Deviance = -2 × log-likelihood

Question 20

How does logistic regression differ from discriminant analysis?

A)It doesn't; they are one and the same.
B)Discriminant analysis separates two groups using several predictors,whereas logistic regression can be used to separate multiple groups.
C)Logistic regression separates two groups using several predictors,whereas discriminant analysis can be used to separate multiple groups.
D)None of the above.

Question 21

In linear regression,one assumption is that there is an independence of errors.In logistic regression,violating this assumption produces what?

A)Overdispersion
B)Underdispersion
C)Neither
D)Both

Question 22

What is complete separation in logistic regression?

A)When the dependent variable negatively influences the independent variable
B)When the plot of the dependent and independent variables do not intersect
C)When the outcome variable can be perfectly predicted by one other variable,or a combination of other variables
D)The use of standard errors to predict the outcome variable

Question 23

In logistic regression,it should be possible to create a table of all possible values of all variables.If there are gaps in the table,then this will increase the risk of large:

A)Standard deviation
B)Mean scores
C)Standard errors
D)Distribution anomalies

Question 24

You are carrying out a study into an area of research in which no previous relevant studies exist,and causality is not your primary concern.You are only intending to find a model to fit your data.What type of method would you employ for theory testing?

A)Likelihood ratio method
B)A simple regression
C)Forward method
D)Stepwise method