Given the number of projects in the Sample Database, it is possible to draw conclusions to a certain level of confidence. Confidence intervals are used to understand how well a sample represents the whole population, in this case how well the Sample Database of 250 projects represents the Master Database. A confidence level describes how likely it is that a characteristic falls within a particular confidence interval. For example, a 95% confidence level indicates that the characteristic being investigated will fall within that interval in 19 out of 20 instances. For some specific examples:
• With the entire Sample Database of 250 projects, the confidence interval will be approximately ±5% at a confidence level of 95% (calculation not shown here). For example, this means that if 30% of projects are found to have experienced renegotiation in the Sample Database, it can be said with 95% certainty that between 25% and 35% experienced renegotiation for the entire population of 3736 PPP projects.
• For smaller subsectors of the Sample Database, this confidence interval increases. There are 126 transport projects in the sample, which increases the interval to approximately ±8%. For example, if 30% of transport projects are found to have experienced renegotiation, it can be said with 95% certainty that between 22% and 38% experienced renegotiation out of the entire population of 3736 PPP projects. The size of this interval shows that it is not reasonable to draw strong conclusions from small sample sizes.