Download presentation
Presentation is loading. Please wait.
Published byMarsha Chase Modified over 8 years ago
1
Public transport quality elements – What really matters for users? By Dimitrios Papaioannou and Luis Miguel Martinez Presentation for the 20 th ECOMM in Athens, 1-3 June 2016
2
Introduction This presentation shows two approaches to measure importance of attributes in PT satisfaction evaluation. Non compensatory multi-criteria approach Utility based approach 2
3
Some background Behavior of attributes is not always symmetric Sometimes trade offs can be made between attributes Sometimes poor performance on one attribute cannot be compensated What are the differences between the two approaches The non compensatory multi-criteria method examines the dominance of criteria over others The utility based approach considers the linear utility function and the compensatory relation between attributes Non compensatory procedures can be used to create importance coefficients for pairwise comparison of alternatives Utility based can be used to estimate trade offs between attributes 3
4
Non-compensatory multi-criteria – Data I Survey titled “Urban mobility and PT satisfaction survey” 13 PT elements were graded depending on their importance (Likert scale 1-7) Satisfaction was asked for same elements on PT users Non users were questioned whether any of these elements prevents them of using the service 4 PT planningPT operationOther PT stop proximity Transfer timePT stop condition Service frequency Ticket / monthly pass price Seat availability Number of transfers Vehicle congestion On board travel time Riding experience Access to destinationsSchedule reliabilityReal time information
5
Non-compensatory multi-criteria – Data II 5
6
Non-compensatory multi-criteria – Methodology 6 We apply the Kemeny – Young method on the importance data Kemeny young is a voting procedure with preferential ballots and pairwise comparison. It belongs to the Condorcet “family” because it produces a Condorcet winner In this case we assume preference of attributes based on overall importance ranking This method can be used to assess the relevance of different attributes in a multi-criteria approach, including non-compensatory decision making
7
Non-compensatory multi-criteria – Estimation process We use binary dominance of attributes to estimate the preferences Importance scores are compared pairwise, The higher scoring attribute takes the value of one In the case of same evaluation the value is split between the two attributes Scores are summed and overall performance of each attribute against the others gives it its relative strength The importance of each attribute is estimated by calculating the weight pf each attribute on the multi- criteria decision 7
8
Non-compensatory multi-criteria – Results Attributes are ranked by their weighted coefficient Frequency, reliability and accessibility attributes have the highest coefficients Public transport stop condition, in vehicle crowdedness, and seat availability have the lowest coefficients Improvements in higher ranked attributes would be preferable against improvements in lower ranked attributes 8 FrequencyReliability Access to destinations Stop proximityTransfer timeTicket 0.1080.1030.0960.0910.0810.076 Real time information Riding experience On board travel time N. of transfers Vehicle crowdedness Seat availability Stop condition 0.0740.0730.0700.0680.0660.0520.041
9
Non-compensatory multi-criteria – Users, non-users, reasons for not using PT We split our data in two segments, users and non- users, and followed the same procedure Similar process was done for the reasons that non- users don’t use PT 9
10
Utility based method – Data Mini stated preference scenarios 3x binary choice for potential improvements of a bus service Each attribute is considered as a potential different alternative of service improvement 2700 responses 33 pairs between 11 variables Variables Number of transfersRiding experience Transfer timeSeat availability On board travel timePT stop proximity TicketService frequency PT stop conditionSchedule reliability Availability of information 10
11
Utility based model - Methodology Utility based approach with a linear function A binary logit model is used to estimate the utility of different attributes for customers Used to calculate the relative tradeoffs between attributes 11
12
Utility based model – Results The model has a R 2 of 0,12 The parameters access time and stop condition were not significant and were disregarded FareFrequency Real time information Reliability Riding experience Seating availability N. of transfers Transfer time Travel time Fare (€) 1-5.77-2.6114.57-8.49-2.992.1915.1631.65 Frequency (serv/hour) -0.1710.45-2.531.470.52-0.38-2.63-5.49 Real time information (yes/no) -0.382.211-5.583.251.14-0.84-5.81-12.12 Reliability (min) 0.07-0.40-0.181-0.58-0.200.151.042.17 Riding experience (yes/no) -0.120.680.31-1.7210.35-0.26-1.78-3.73 Seating availability (yes/no) -0.331.930.87-4.882.841-0.73-5.08-10.59 N. of transfers 0.46-2.64-1.196.67-3.89-1.3716.9414.48 Transfer time(min) 0.07-0.38-0.170.96-0.56-0.200.1412.09 Travel time (min) 0.03-0.18-0.080.46-0.27-0.090.070.481 12
13
Utility based model – Results An extra service per hour is valued 0,17 € People are willing to pay 0,38 € to have real time information, while reducing uncertainty in waiting by one minute has a value of 0,07 € Having one less transfer is the equivalent of 0,46 €; each minute spent transferring is valued at 0,07 € Regarding in vehicle experience, having a seat is valued at 0,33 €; a smooth ride at 0,12 €; and an extra minute of travel time at 0,03 € Comparing times, waiting time both at trip origin and during transfers is valued twice as much as in vehicle time Real time information is equivalent with saving 6 minutes of waiting time (either planned or unplanned) 13
14
Conclusions IConclusions Non-compensatory multi- criteria approach This method doesn’t produce a satisfaction score It identifies the relevance of each attribute towards satisfaction in pairwise comparisons between potential interventions to the system Generally people tend to value network elements higher than comfort and ride related attributes Extreme bad performance in an attribute can lead to non-trading behavior Utility based approach Trade offs make more sense compared with fare or time attributes Binary attributes should not be compared with each other 14
15
Conclusions II Sometimes direct trade offs exist between attributes but only within a certain acceptance range The attributes that users assign the biggest importance might be more significant for mode choice The same attributes probably behave in a non-compensatory framework For example if the service frequency is not high enough for me, I will not consider PT as an alternative Attributes that are considered less important have bigger acceptance rates These attributes might end up being the main drivers behind satisfaction For example a crowded vehicle is not as important as service reliability, therefore I am willing to accept it 15
16
Conclusions III Differences exist on how users and non-users see PT service Some attributes are valued more by users (ticket price, real time information) Others by non-users (n. of transfers, riding experience) The reasons non-users don’t use PT do not follow the same ranking as their respective importance In vehicle travel time is ranked much higher as a reason for not using compared with its relative importance 16
17
Questions? 17
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.