Study for SPSS Test

Question 1

Correlation

Answer 1

measure of the degree of association between two variables (x and
y) initially assumed to be numerical

Question 2

Pearson’s correlation

Answer 2

Investigates the linear relationship between x and y (scatterplot of x and y)

Question 3

Pearson’s correlation assumptions

Answer 3

Assumes both x and y to be numerical and that at least one is
normally distributed.

Question 4

Pearsons correlation H0

Answer 4

: There is no linear association in the
population between the two variables X and Y
(ρ=0).

Question 5

Spearmans rank correlation

Answer 5

Investigates any monotonic* relationship between x and y

Question 6

Spearmans rank correlation assumptions

Answer 6

At least one variable should be continuous (the other could be ordinal) –
they do not have to be normal (ranks)

Question 7

Spearmans H0

Answer 7

There is no association in the population
between the two variables X and Y (ρs=0)

Question 8

Myotonic

Answer 8

As one increases so does other but maybe not at constant rate

Question 9

Linear regression

Answer 9

e investigation of the relationship between two independent continuous
variables, X and Y (X’s for multiple linear regression). It can be used to predict Y
(dependent/response variable) from the X(s) (independent /predictive variable(s))

Question 10

Simple linear regression equation

Answer 10

Y= β0 + β1X1 + e

Question 11

Simple linear regression assumptions

Answer 11

Linear relationship between X(s) and Y – Check by Correlation, scatterplot of
X and Y, OR scatter plot of the Normalised Predicted values vs Dependent
values
* Normal distribution of RESIDUALS – Check by a Histogram of Residuals, or
by inspection of the P-P plots
* Constant variance – Check by a Scatterplot of Residuals vs Normalised
Predicted (sausage shaped, random scatter)
* Independent observations – Check by Scatterplot of Normalised Residuals
and Predicted - AGAIN (no organisation)

Question 12

Simple linear regression fit of mode

Answer 12

The coefficient of variation, R2, indicates how much of the variation in Y
is explained by the proposed model

Question 13

Simple linear regression what does ANOVA do

Answer 13

This divides the overall variation into the
variation explained by the regression model and the residual (left over or not
explained) variation and compares them.

Question 14

SLR ANOVA H0

Answer 14

The relationship between X and Y in the population is not informative. The
regression coefficients are zero i.e. Ho: β0 = β1 = 0

Question 15

SLR tests of individual coefficents H0

Answer 15

H0β0: The intercept is zero i.e. β0 = 0
H0β1: The effect of Xi is constant β1 = 0 i.e. the coefficient is flat

Question 16

SLR t value for coefficient to be sig

Answer 16

t>2

Question 17

Multiple linear regression H0

Answer 17

The relationship between all the X’s and Y in the population is not
informative. All the regression coefficients are zero

Question 18

Equation of multiple logistic regression model

Answer 18

logit(p) = β0 + β1X1 + β2X2…+ βnXn + e
where logit(p)=loge(p/(1-p))

Question 19

How to get odds ratio from logit

Answer 19

Exponential function of the coeffieient

Question 20

Assumptions for logit regression

Answer 20

None

Question 21

Bland altman plot

Answer 21

Analyse agreement between 2 methods
Both variables should be continuous

Question 22

Interpretation bland-altman plots

Answer 22

Estimate of bias shows how big average discrepancy between 2 methods are, should be mean +/- 2SD

Checks variability consistence across range of values

Question 23

When to use RR

Answer 23

Compare the risk of an event between 2 groups, used in cohort and RCT as measure of effect

Question 24

RR H0

Answer 24

Risk of x is the same in both groups

Question 25

When to use OR

Answer 25

Compare odds of exposure between those with the event to those without event

Question 26

Calculate SE

Answer 26

SE = SD/square root n

Question 27

Calculate CI

Answer 27

CI = x +/- 1.96 x SE