Risk, summary measures, probability + distribution

Question 1

How do you calculate the absolute risk difference (ARD)?

Answer 1

|p(test) - p(control)|

Risk difference, ignoring the sign

Question 2

How do you calculate the p(test)

Answer 2

a/a+c

Proportion of subjects with a positive outcome under the test treatment

Question 3

How do you calculate the p(control)

Answer 3

b/b+d

Proportion of subjects with a positive outcome under the control treatment

Question 4

How do you calculate the relative risk?

Answer 4

p(test)/p(control)

a/(a+c))/(b/b+d)

Question 5

How do you calculate the number needed to treat? (NNT)

Answer 5

1/ARD

1/|p(test)-p(control)|

Question 6

How do you calculate odds?

Answer 6

p/(1-p)

Question 7

How do you calculate odds ratio?

Answer 7

odds(test)/odds(control)
(a/c)/(b/d)
ad/bc

Question 8

6 properties of normal distribution curve

Answer 8

1. Bell-shaped
2. Symmetrical
3. Peak of the curve lies above the mean
4. Mean + median coincide
5. Any position along the horizontal axis can be expressed as number of SDs away from the mean
6. Total area under the curve is 1 or 100%

Question 9

What proportion of observations lie outside the mean +/- 1SD of a normal distribution?

Answer 9

32% (31.7%)

Question 10

What proportion of observations lie outside the mean +/- 2SD of a normal distribution?

Answer 10

5% (actually 1.96SD)

Question 11

How do you calculate the standardised normal deviate (Z)/number of standard deviations away from the mean?

Answer 11

Z =(X−μ)/σ
X = random variable with normal distribution
μ = mean
σ = standard deviation

Question 12

When reading the probability from a table of Z values, what is it important to remember?

Answer 12

Cell in table contains probability X+Z

Question 13

What proportion of observations lie outside the mean +/- 3SD of a normal distribution?

Answer 13

0.27%