Week 6

Question 1

The three steps to assess the validity of combing studies?

Answer 1

1. Assessing the clinical similarities
2. Assessing similarities in Methods
3. Assessing the similarities stats-This is assessed by the I Stat

Question 2

I-Stat?

Answer 2

Tests if the variation found is more than what would be expected by chance.

Question 3

Three ways to access bias (ranging from good to worse)?

Answer 3

Assess the individual components (the Cochrane risk tool is a good one); Checklists and then scales.

Question 4

When assessing the individual components look for:

Answer 4

1. Random allocation
2. Allocation concealment
3. Participant blinding
4. P’s lost at follow-up
5. Early stopping for benefit

Question 5

Effect Size?

Answer 5

It quantifies the relationship between the IV’s and the DV or the relationship between a predictor variable and an outcome variable.

Question 6

Explain average effect sizes in MA’s

Answer 6

A MA gets all the effect sizes and works out the average effect size.In order to get the average effect sizes you can use either a FIXED or RANDOM effects model. You can’t just work out the mean as each person will then have the same weight. You need to treat some studies as more important than others.

Question 7

Random Effects?

Answer 7

Used when there is thought to be sig heterogeneity between studies.

Question 8

Assumption of Random Effects?

Answer 8

That there is a distribution of effect sizes, there is no one true effect size. It gets a effect size by assigning weights to the different studies based on the random error and how different the studies are from each other (systematic error)

Question 9

Assumption of Fixed Effects

Answer 9

There is less heterogeneity.
It relies on the assumption that there is one single effect size. It treats the one with the largest sample size as most important. It assumes that because there are so many samples, this study would be most likely to be a good estimate of the true effect. It says that the reason we see different true effects is because of sampling error or bias.

Question 10

Explain the I-stat:

Answer 10

When the difference between 2 studies are not due to sampling error use the I stat. It tells you the amount of variability (the diff between the effect sizes) that is due to systematic error. This is expresses as a %. If the I stat is low then we can say that the studies have a high sampling error.