Meta- Analysis

Question 1

What does a Meta-analysis involve?

Answer 1

contrasting and combining results from multiple studies to provide an aggregate measure of effect size

Question 2

Why do we use a random-effects design?

Answer 2

If measures and procedures differ from study to study

Question 3

What are some ways that studies can be weighted?

Answer 3

There are several ways to determine weights, for example: validity, sample size, and inverse variance

Question 4

What are the results of a meta-analysis displayed in?

Answer 4

A forest plot

Question 5

What is on the y and x axis of a forest plot?

Answer 5

The results of individual studies as well as the aggregate effect size are displayed in a forest plot

y axis = individual studies and summary measure
x axis = effect size

Question 6

How do we interpret the aggregate effect size on a forest plot?

Answer 6

if the points of the diamond overlap the line of no effect the overall meta-analysed result cannot be said to differ from no effect at the given level of confidence.

Question 7

In M-A, Statistical power is _____ due to an increase in ____ ____

Answer 7

increased due to an increase in sample size

Question 8

M-A increases the _____ of the data, as results from many samples are pooled together

Answer 8

generalisability

Question 9

What can also be investigated alongside a meta-analysis?

Answer 9

Publication bias

Question 10

How many meta-analyses are thought to have suffered from publication bias?

Citation?

Answer 10

25%

Ferguson & Branick, 2012

Question 11

What does publication bias lead to?

Answer 11

A biased distribution of effect sizes

Question 12

How can we investigate publication bias in a graphical form?

Answer 12

Funnel Plot

Question 13

What is a funnel plot?

Answer 13

It is a scatter plot of standard error (or sample size) on Y axis, versus effect size on the X axis.

Question 14

If publication bias is NOT present…

Answer 14

The funnel plot will appear symmetrical about the vertical:

Smaller studies have more scatter in the size of the effect and thus appear scattered at the bottom of the plot, while larger studies have less scatter and thus form the tip of the funnel at the top of the plot.

Question 15

If publication bias is present…

Answer 15

The funnel plot will appear asymetrical, with data points appearing disproportionately on the side representing larger effect sizes

Question 16

As well as visually inspecting the plot for asymmetry…

Answer 16

Eggers regression (1997) can be conducted which statistically tests for asymmetry.

Question 17

A meta-analysis of several small studies doesn’t necessarily…

Answer 17

predict the result of one large randomised control trial

Question 18

A meta-analysis does not correct for…

Answer 18

poor design or bias in the original studies

Question 19

Issue with having a strict inclusion criteria?

Answer 19

it introduces a level of subjectivity which meta-analysis should avoid

Question 20

If studies do seem unreliable, the researcher could consider….

Answer 20

assigning an appropriately low weight to them.

Question 21

who found only a fair agreement between large RCT and M-A addressing the same question

Answer 21

LeLorier et al (1997)

Question 22

What is ‘best evidence synthesis’?

Answer 22

‘best evidence synthesis’ should be used in which many facets of a ‘good’ study (e.g. internal and external validity) are taken into account when defining strict a priori inclusion criteria

Question 23

In terms of a forest plot, If the confidence intervals for individual studies overlap with the line of no effect…

Answer 23

it demonstrates that at the given level of confidence their effect sizes do not differ from no effect for the individual study.

Question 24

How is each individual study represented in a forest plot?

What is the area of each square proportional to?

Answer 24

Each individual study is represented as a square on the plot; The area of each square is proportional to the study’s weight in the meta-analysis

Question 25

What is the aggregate effect size represented as in a forest plot?

Answer 25

The overall meta-analysed measure of effect is often represented on the plot as a dashed vertical line.

This meta-analysed measure of effect is commonly plotted as a diamond, the lateral points of which indicate confidence intervals for this estimate.

Question 26

What is the vertical line on a forest plot?

Answer 26

it represents the line of no effect