Ghana - Transportation

Road users
We sampled 12 out of 40 feeder roads for traffic counts from within groupings of roads (referred to as strata) that were created based on the road's region, surface treatment, and traffic volume (high/low). A road was selected for the sample from within the strata with a probability of selection that depends on the length of the road relative to the total length of all roads within the strata. On an ad hoc basis, two important roads were selected with 100% certainty, and we document the rationale for this choice. The remaining 10 roads in the sample were allocated to strata to ensure that at least one road was selected from each strata.

We selected six roads for vehicle intercept surveys from the sample of roads that received traffic counts. The sample selection process for selecting roads for the vehicle intercept survey was purposive and met the following criteria:
1. Roads selected for the vehicle intercept survey also had been selected for traffic counts.
2. At least one road from each region was be selected. In the case of Volta region, one road was from the cluster of roads built in the Hohoe district (northern Volta region) and another was from the southern part of Volta region.
3. The sample included the two important roads mentioned above.
4. The sample included at least one gravel road and at least one road with low traffic volume.

The vehicle intercept survey lasted three days per road, including both a market and nonmarket day. On the N1, we randomly sampled approximately 1,000 vehicles, or approximately 0.5 percent of all traffic moving in both directions along the road. On high-traffic feeder roads (more than 1,000 vehicles per day), we sampled 20 percent of all vehicles traveling along the road. For the trunk road and most feeder roads with a lower traffic volume, we sampled a larger proportion of vehicles. Actual sample sizes were 1,038 vehicles on the N1 (0.6 percent of Average Daily Traffic (ADT), 976 vehicles on the trunk road (19 percent of ADT), and 4,236 vehicles across the six sampled feeder roads (or 11 percent of total ADT on these roads). Response rates are within the targeted range when accounting for surveys not being conducted in the evening or at night out of concern for the safety of enumerators. For vehicles carrying at least one paying passenger, one passenger from each vehicle was randomly selected for the passenger survey.


Sample weights for the vehicle intercept survey were computed as the inverse of the probability of selection of the vehicle based on vehicle type and the frequency of that vehicle type in the traffic counts during the same time period. The weight variable is called "weight" and is used with the questions from vehicle intercept survey administered to all respondents. On some roads, in an effort to reduce survey nonresponse, we also conducted an abbreviated version of the survey with drivers who were especially reluctant to participate or who had already completed the full version of the survey at an earlier time. The weight variable for these questions is called "weight_complete" and is used with questions from the vehicle intercept survey that were administered during the full survey but not during the abbreviated survey.

Sample weights for the passenger survey were similarly computed as the inverse of the probability of selection of the passenger based on vehicle type, the frequency of that type of vehicle in the traffic counts, the percent of that vehicle type that were eligible (those to be carrying at least one paid passenger during the vehicle intercept survey), and the number of passengers in that vehicle. The weight variable "weight" and is used with the passenger data.


The response rate for the vehicle intercept survey was 89 percent of vehicles stopped. The response rate for passengers on the passenger and pedestrian survey was 95 percent of sampled passengers. No data was gathered on response and refusal rates for the N1 axle load survey. The overall response rate on the axle load surveys on feeder roads was 82 percent of vehicles stopped.