# Test Yourself

## True or False

#### Feedback

**False**

**False: For risk ratios and odds ratios and other measures of association and point estimates that are bound by zero, we often calculate the CI on the log scale so that we don’t end up with unrealistic CIs that contain zero. This creates assymetrical CI when back transformed to the OR or RR scale.**

#### Feedback

**False**

**False: The risk difference is a very useful measure because it an estimate of the absolute risk reduction/increase in the population. However, if we wish to compare the strength of association of a factor across different outcomes then a relative risk can be more useful because it is a relative measure that is independent of the baseline proportion or rate.**

## True or False

#### Feedback

**False**

**False: We never accept a hypothesis, just fail to reject it. The null hypothesis in this case was there was there was no difference in the proportion of achieved pregnancies between the two IVF procedures. The p-value of 0.15 just indicates that this observed difference is quite likely to occur from a sample of this size just due to random sampling if in fact there is no difference between IVF procedures. A larger study may find evidence of a difference.**

#### Feedback

**False**

**No, a p-value is a measure of strength of evidence against the null, this classification of a result as significant or not is not encouraged. We also need to consider the size of the effect or association.**