Sampling Methods for On-Street Parking


By Linden Smith

When I began working in the parking industry 10 years ago now, we had a “how to” report from Chance Management Advisors that became my introduction to parking analysis. This report was the starting point for a journey that now includes technological achievements which were unseen at that time. For me, I saw that the basic methods of gathering sample data were not as precise as I would like. What I came up with I called “Santa Sampling,” and it opened our eyes to things we did not expect to see. I’ll go over both methods to show the strong points and weak points and what sampling will look like in the not-so-distant future.

The first data collection method is the “route” survey.  I call it this because it involves walking a route, just as the PEO’s would. Turnover surveys should always be conducted during peak hours of parking. Generally, this is the lunch hour rush between 11:00 AM and 2:00 PM (here in Lexington, KY). The occupancy survey is conducted at least once a month and needs to be documented. For surveys, I use prints of the meter map, and then mark meters that are occupied. Use a “V” to mark violations and “VC” to mark violation captured. I keep monthly data on a spreadsheet that shows the rates over months and years. This method is simple and easy to conduct and can track patterns in parking behavior.

It does have shortcomings; basically, the violation capture rate is going to be very difficult to pin down using this sampling method. There are sampling biases in this method and the violation rate, and the violation capture rate are going to be invalid.

To explain the bias in this method, I will use an example from Lexington. In our urban core we have a very short average length of stay and a very high turnover rate in our downtown. The ALOS is 30 minutes and the turnover is over 200%. In the occupancy survey, the math assumes that all parking, paid and not paid, have the same characteristics in ALOS and turnover rate. However, that is not borne out by the reality of parking. If someone knows that they are going to be more than 30 minutes, they are more likely to pay, if they feel they are just going to be a short time in the space, they are more likely to not pay. This means that the length of time that the violation and capture are “exposed” and counted in the survey are less than that of other parkers. So, if you count the numbers up, violations are occurring more often, but for a shorter period of time than the regular parkers. It may be that you have an under count in the number of violations, but the overall time of violations may be correct only by coincidence.

There is a survey method that gives an accurate and precise count of violations, but it can be time consuming so that you would not do this every month, only periodically to verify the results of the occupancy survey. I call this the Santa Sample. You are going to make a list and check it twice, and at the end you will find out who is naughty and nice.

The Santa Sample differs from the occupancy/turnover survey in one important way. While in the occupancy survey you move around and count cars in spaces, in the Santa Sample you sit in one place and let the cars come to you. Think of it this way; imagine you have an apple pie and you want to know all about the pie. Would you randomly drill holes in the pie and log each sample or would you cut a slice and extrapolate out from there? The pie slice would be the better choice to get precise information.

Find a spot where you can directly observe many parking spaces, say 10-20 spaces. You are going to track every movement in and out of each space. This survey should be done at least as long as the meter limit, or a full day if possible. Once you have made your “list” you will check it against 1) the meter transaction reports and 2) the tickets written by the enforcement officers. You should be able to measure very precisely, for just those meters, the number of violation occurrences and the length of time of violations. You can then work your way through the different areas of parking generation and find patterns between parking generators and the characteristics of violations.

The math for the Santa sample is a little different from the occupancy surveys. I won’t go into that in depth here. But from this data you will find you can generate any information relevant to parking analysis.

What I call “counting cameras” are now being used to create the list of cars going into and out of metered spaces. These are different from LPR cameras since they have lower resolution and frame rate, making them inexpensive and mobile. The data from the camera is then run through software that recognizes parking events and creates the list. It then pulls data from the meter transactions and the PEO citations to produce what I’ll call “perfect knowledge” of the sample area. It’s at a level comparable to parking garages.

With this data every statistic may be drawn out. Occupancy, paid occupancy, turnover, violations, duration of violations and even time left on the meter, and free ride time, that used by the next car in the space with time left over.

What we found was pretty shocking. Paid occupancy was just 70%. Violation capture rate was under 10%. ALOS was just 22 minutes in a 2 hour limit area. And turnover was 240%. These numbers were all well outside of what we were expecting. But, the results showed us two things; we could find revenue in the paid occupancy, getting that up to anything over 85%, and violation capture could easily double with better coverage of high turnover areas.

On the plus side, it showed us just where the pot of gold was located. Now we just need a rainbow to get us there.

Linden Smith is a Parking Analyst with the Lexington & Fayette County Parking Authority. He can be reached at