3.  A random sample of 250 PPPs was chosen, as representative of the Master Database

A.  It was decided to select a sample of 275 projects. This allowed for some leeway (i.e. 10% surplus) when collecting the data in case it proved difficult to gather information on some projects.

B.  The percentage breakdown in the Master Database for each region, sector and financial close period was recorded.

C.  A script in Excel was created which carried out the following:

i.  Randomly selected 275 projects from the Master Database, creating a 'Target Database'.

ii.  The percentage breakdown for that Target Database was calculated for each region, sector and financial close period.

iii.  The differences between the Master Database percentage breakdowns and the Target Database percentage breakdowns were calculated and the differences added together. For example, the Master Database had 17.2% of eligible projects in Europe and 3.4% of eligible projects in North America. A sample with 15% and 3% of projects in those regions respectively would have a difference of 2.2 + 0.4 + … = 2.6 + … for these characteristics.

iv.  The process was repeated 10,000 times, and the Target Database with the smallest difference to the Master Database was selected.

Once this process was completed (including removing the additional 25 projects, as described below), the result comprised the 'Sample Database'. The composition of the Sample Database is shown below.

Table 1: Composition of the Sample Database by Region

Table 2: Composition of Sample Database by Sector and Financial Close period