Logistic regression

Question 1

X variable can be

Y variable can be

Answer 1

continuous or categorical

BINARY ONLY

Question 2

equation?

Answer 2

y = e^(B0 + B1X1…)/1+e^(B0 + B1X1…)

Question 3

Y can range from

Answer 3

0 to 1

Question 4

odds ratio formula?

Answer 4

e^B

Question 5

OR > 1,, then B = __ and it is a ___ factor

Answer 5

> 0

risk

Question 6

OR = 1,, then B = __ and it is a ___ factor

Answer 6

= 0

neither

Question 7

OR < 1,, then B = __ and it is a ___ factor

Answer 7

< 0

protective factor

Question 8

difference between univariate and multivariate regression?

Answer 8

univariate looks at each X variable INDIVIDUALLY

multivariate looks at each X variabe IN COMBINATION

Question 9

two ways of evaluating the logistic regression model?

Answer 9

Hosmer-Lemeshow goodness of fit statistic

Receiver operator characteristic and c index

Question 10

Hosmer Lemeshow goodness of fit looks at what two statistics

Answer 10

chi 2

p value

Question 11

X^2 is low, then

Answer 11

P value is high, model fits the data well

Question 12

X^2 is high, then

Answer 12

P value is low, model doesn’t fit data well

Question 13

Receiver operator characteristic describes

Answer 13

how well the equation/model discriminates between patients who do and dont develop the outcome of interest

Question 14

Y axis is

Answer 14

sensitivity

Question 15

X axis is

Answer 15

1 - specificity

Question 16

C index measures

Answer 16

area under ROC curve

Question 17

C index of
0.5
0.75
1

Answer 17

no better than chance
good fit
perfect fit