The sample of Ugandan firms was designed as a stratified random sample on the basis of census data from the Uganda Bureau of Statistics.
The sample frame was obtained from the Uganda Bureau of Statistics, which compiled the list during a nationwide census in 2001-2002. The sample frame was stratified by location, manufacturing subsector, and size. To adequately reflect the geographic distribution of firms, three regions were defined on the basis of existing
districts: the central region, which includes the Kampala district; the northeast region; and the southwest region. Similarly, to obtain the correct distribution of activity in manufacturing, nine broad subsectors were defined using the ISIC: agro-industry; chemicals and paints; construction materials; furniture; metals; paper, printing, and publishing; plastics; textiles and leather products; and wood. Finally, to allow comparisons with datasets compiled for other African countries by the Regional Program on Enterprise Development (RPED), three standard size classes were used: small (10-49 employees), medium-size (50-99), and large (100 or more). Clusters were then defined on the basis of the location, size, and sector of firms.
Following the stratification of the sample frame, firms were selected randomly in each cluster. A total sample of 410 firms was drawn, with an overall sampling rate of 28.5 percent. The surveyed sample differs slightly from the drawn sample because of some replacement sampling, but it retains the key characteristics: the importance of firms with 100 or more employees and the prominence of the central region. The surveyed sample includes 392 firms (a sampling rate of 27.2 percent). Firms with 100 or more workers account for about 86 percent of the employment in the sample, and the central region accounts for 66 percent of the enterprises and about 66.5 percent of the employment.
The difference between the theoretical and the surveyed sample is explained by several factors: some firms refused to be interviewed, several firms changed size class (the survey was undertaken about a year after the completion of the census), and some firms either did not exist or had changed their activity. Some of these "missing" firms were replaced with firms with similar or identical characteristics. These replacements replacements from the sample frame were required to be similar in size and to operate in the same sector and the same region.
The manufacturing subsample is the largest component of the overall sample. It includes 300 firms, or about 40 percent of the existing manufacturing firms in the sample frame, and accounts for about 75 percent of the manufacturing employment. This subsample reflects the dominance of the central region and large firms. But it differs slightly from the sample frame in its structure. In the subsample 68 percent of firms are in the central region (compared with 72 percent in the frame) and 19.7 percent of firms have 100 or more employees (compared with 10.6 percent in the frame).
