Appendix E - LVSA Unmet Demand Survey methodology clarification
Perth and Kinross Council · 6 October 2025
Back to Licensing Committee — 6 October 2025 and its agenda
Text extracted from the council's PDF. Layout and reading order may differ from the original.
Page 1
LVSA Unmet demand survey methodology clarification. 1.1 Background With respect to the taxi unmet demand survey undertaken in Perth during 2024, some comments made appear to place an interpretation of the methodology used, which may not be in line with the methodology actually applied. This brief paper sets out to define some of the key elements of the Index of Significance of Unmet Demand calculation. 1.2 How the passenger delay related elements in the ISUD calculation are derived. Passenger waiting at taxi ranks is measured in two principal ways. This is: Discrete passenger waiting and Continuous passenger waiting. Discrete passenger waiting Discrete passenger waiting occurs when an individual passenger or a group of passengers arrive at a taxi rank and there are no taxis available for immediate hire. The time that the individual or group arrived at the rank is logged. The time that the passenger or group leave the rank, either in a taxi or on foot, is logged. The diƯerence in times is calculated and recorded as the wait time for the passengers concerned. Continuous passenger waiting. When lengthy passenger queues form, the queuing becomes continuous. It is often not feasible to identify each passenger or group of passengers throughout their wait in the queue. Therefore an alternative methodology is employed. The number of passengers in the queue is measured every 5 minutes. The number of passengers who departed the rank in taxis or on foot is logged every five minutes. This data is then used to calculate the average wait time for passengers waiting in the queue during each five minute period. Not all waiting passengers are counted as waiting Occasionally, when passengers are waiting at the taxi ranks, they may not all be waiting to board a taxi. For example, it is not uncommon for a group of people to wait at a taxi rank, for taxis to arrive and not all of the group board a taxi. Some of the group may leave the rank on foot, once the rest of the group have boarded a taxi or taxis. Clearly, the people who didn’t board the taxi were merely accompanying friends, until the taxi arrived. When this occurs, those waiting people who left the rank are not counted as waiting passengers in the passenger waiting statistics. Similarly, it is occasionally observed that people will wait at a taxi rank, when there are no taxis present. For example, sitting on a nearby bench. If a taxi arrived and there were
Page 2
no other people waiting and the people who were waiting didn’t board the taxi, then they were assumed not to be waiting for a taxi and were excluded from the passenger waiting statistics. How passenger waiting time is used Passenger waiting time is used in the calculation of the Index of Significance of Unmet Demand (ISUD). Two of the coeƯicients which are used in the index calculation relate to passenger waiting times. These are: The proportion of weekday daytime (off peak) hours where people are observed to have to wait for a Taxi to arrive. Also known as the Steady State Performance (SSP) statistic The proportion of passengers which are travelling in any hour when the average passenger wait in that hour is greater than one minute. Also known as General Incidence of Delay (GID) SSP Calculation The proportion of weekday daytime (oƯ peak) hours where people are observed to have to wait for a taxi to arrive is calculated separately for each taxi rank observed. The original definition of the ISUD calculation, developed in the 1980s, was based on the assumption that ranks would be sampled for periods of around two hours, during active periods and the results expanded using standardised profiles of demand variation throughout the day. However, practices have moved on and nowadays, complete observation over 24 hours each day is commonplace. As such, the original approach of counting any passenger waiting (even for a few seconds) as triggering that hour as a passenger wating hour, was excessively onerous. Alongside new, more expansive observation practices, a more conservative approach has been adopted. This more conservative approach considers the SSP trigger to occur if a passenger has been observed to wait for more than a minute at an active taxi rank. A rank is considered to be active, for the purposes of calculation of the SSP statistic, if two or more hires occur during that hour. If the level of passenger waiting at an individual rank exceeds the threshold, that rank is considered to be exhibiting SSD waiting for that hour. However, if passenger waiting at other ranks does not exceed the SSD waiting threshold for that hour, then they are not considered to be exhibiting SSD waiting. For example, assuming there are 40 hours per week of oƯ peak hours and four active ranks, this equates to 160 oƯ peak rank hours. If passenger waiting at one rank on a Friday morning exceeds the SSD threshold, this would be one hour out of the 160 rank hours, i.e. 1/160 = 0.625%
Page 3
The data collected on the Thursday surveyed is weighted in order that the Thursday results are taken to be representative of four weekdays. The Friday results are used individually, without weighting. GID Calculation Passenger waiting was observed at each rank from 7:00 am on a Thursday though to 7:00 am the following Sunday. The total number of passengers boarding taxis during each hour was recorded. Aggregate passenger waiting time was calculated for each hour. The aggregate passenger waiting time was divided by the total number of passengers boarding taxis, during the same hour, to calculate the average waiting time for all passengers during each hour. If the average waiting time for all passengers during any hour exceeded one minute, then this triggered the threshold to include the total number of passengers during that hour in the GID calculation. Data collected on Thursday was weighted to represent four weekdays. 1.3 How average passenger delay is calculated. The total aggregate passenger waiting time observed on Thursday was multiplied by four to represent four days, Monday to Thursday. The aggregate passenger waiting time on Friday and Saturday (including Sunday morning up to 7:00 am) was also added to the aggregate Monday to Thursday passenger waiting, to calculate a weighted aggregate passenger waiting time for Monday morning to Sunday morning. This total represents aggregate passenger waiting time over six days. Similarly, the aggregate number of passengers was calculated for Thursday (multiplied by four) plus Friday and Saturday. The aggregate waiting time was divided by the aggregate passengers to calculate the average wait time over the six days.