Stats 3

Question 1

Vote counting

Answer 1

When we count significant results and use the majority result as our summary outline (see example).

Question 2

What is a good way to identify an informative study?

Answer 2

Large sample size

Question 3

Meta-analysis

Answer 3

Calculates a composite effect by assigning more weight to your powerful studies.

Question 4

The file drawer problem

Answer 4

“unwanted” or “less exciting” results are less likely to be published, therefore the literature is arguably bias towards novel and exciting work.

Question 5

Outcome reporting bias

Answer 5

Focussing on positive results from the paper.

Question 6

“spin” phenomenon

Answer 6

“trends” went in the right direction

Question 7

Citiation bias

Answer 7

Papers showing positive results are more likely to be cited.

Question 8

Funnel plot

Answer 8

Small studies - more biased
Large studies - less biased (more scrutinised)

Question 9

Small study effect

Answer 9

Appears like publication bias, but it isn’t:
1. Small study with the sickest (most desperate)

2. Follow up study is bigger less sick people
3. Follow on from that would be even large and healthy people

Question 10

Positive predictive value

Answer 10

of all the people who test positive, who actually has the disease =49% (disease is so rare that statistically - if a disease is common, testing I worth while)