# Test Yourself

## True or False

### Question 1

The 't' distribution describes the variation of sample means that follow a normal distribution.

### Question 2

The paired 't' test would be appropriate to test the effect of treatment on a group of patients if measurements of a normally distributed numerical variable were made on the same individuals before and after treatment.

### Question 3

If I collected blood pressure from 20 pairs of brothers and sisters I can use a two sample t-test to compare the mean blood pressure between boys and girls.

## Multi-select

### Question

The average crown rump length of 100 fetuses examined by ultrasound was 10cm with a standard deviation of 1 cm (10mm). Which TWO of the following statements are true? (Assume that the value of 't' is 2).

a)      The standard error of the mean is 1mm.

b)      The 95% confidence intervals of the mean are 9.9-10.1cm.

c)       The 95% confidence intervals of the mean are 9.8-10.2cm.

d)      The 95% confidence intervals of the mean are 9.0-11.0cm.

e)      The standard error of the mean is 0.1mm.

### Question

A randomised controlled trial looked at the effect on outcomes of IVF/ICSI of using a particular pain suppository at the end of oocyte retrieval. Half of the women were randomised to have the suppository, whilst the remainder did not have it (parallel group study). One of the main documented outcomes was the pregnancy rate per embryo transfer expressed as a percentage. The difference between the pregnancy rates for the two groups was quoted as 6% with a 95% CI of -3% to 16%. Which of the following is not true?

There is no strong evidence to suggest that there is a difference between the two groups.

We can be 95% confident that the true difference in pregnancy rate is between -3% and 16%.

The true difference in pregnancy might be as great as 16%.

The CI is wide and a bigger study might be better able to detect a difference if it exists.

The pregnancy rates are statistically significantly different (P=0.03).

## True-False Question

### Question 1

A one tailed test can have a p-value greater than 0.5

### Question 2

If I wish to perform a two sample test, such as a t-test, and calculate a CI for the difference between groups, then the only information I need are the mean, SD and sample size in each group.

### Question 3

If I wish to perform a paired t-test and calculate a CI for the mean difference between paired measures, then the only information I need are the mean, SD and sample size in each of the two sets of paired measured.