For the Tanga - Horohoro road we identified 16 villages along the treatment road and 12 villages along the comparison road. Sample size for this road project was 24 villages (see Table I.1 of the survey report provided under Related Materials), spread out evenly among the treatment and the comparison road. We thus took a random sample of 12 out of 16 treatment villages. The comparison group consisted of all 12 villages along the comparison road. The listing team visited the 24 selected villages and drew up exhaustive lists of vitongoji within each village. Vitongoji were divided in two groups depending on their distance to the road and one kitongoji was randomly sampled from each group. Table I.2 shows the final sample (48 vitongoji) for the Tanga - Horohoro road project. Figures 1 and 2 in the appendix show the exact location (captured by GPS) of all surveyed vitongoji along the treatment (Figure 1 of the survey report) and comparison (Figure 2) road.
For the two stretches of road on the southern Mtwara corridor we identified 18 villages that are located on or along the roads that will be upgraded and 21 villages that are located along the selected comparison roads (see subsection 2.2). Sample size for this road project was 34 villages (seeTable I.1), spread evenly among treatment and comparison roads. We thus took a random sample of 17 out of 18 treatment villages and 17 out of 21 comparison villages. The listing team visited the 34 selected villages and drew up exhaustive lists of vitongoji within each village. Vitongoji were again divided in two groups depending on their distance to the road and one kitongoji was randomly sampled from each group. Table I.3 shows the final sample (68 vitongoji) for the Mtwara corridor road project.
For the longest road (Sumbawanga - Tunduma), we identified 27 villages along the treatment road and 26 villages along the selected comparison roads (see subsection 2.1). Sample size for the impact evaluation of this road was 42 villages spread out evenly among treatment and comparison roads. We thus took a random sample of 21 out of 27 treatment villages and 21 out of 26 comparison villages. Again, the listing team visited the selected villages and drew up exhaustive lists of vitongoji within each village. Vitongoji were again divided in two groups depending on their distance to the road and one kitongoji was randomly sampled from each group. Tables I.4 and I.5 show the final sample (84 vitongoji) for the Tunduma ‐ Sumbawanga road project (Table I.4 for the Tunduma - Mkutano part in Mbeya Region and Table I.5 for the Mkutano - Sumbawanga part in Rukwa Region).
