Resumen
Estimation of speed is an important task for various traffic engineering analyses like crash analysis, establishing speed zones, evaluation of traffic signal locations and so on. Past research on the sample size estimation was focused on finding the number of GPS based probe vehicles required. However, there are limited studies on the sample size determination of speeds obtained using video-graphic survey. Large samples increases the chances of finding the significant difference but sometimes it leads to wastage of resources. Inadequate sample size may lead to inaccurate results. Hence sample size determination is often an important step in traffic engineering studies. This paper attempts to determine the sample size based on variability and traffic conditions. An evaluation of the sample size determination such as ITE and Hybrid methods were also carried out. From the data analysis, it was found that there was not much variation in speeds for given vehicle type and location. The mean and variance in speeds obtained from first ten speed measurements for a vehicle type are not statistically different from those obtained after one hour of data collection. For some locations, where the proportion of heavy vehicles and flow were low, statistically stable mean speeds of heavy vehicles could not be obtained even after hour of data collection.