To measure is to know. If you cannot measure it, you cannot improve it. — Lord Kelvin
Measurement is the fundamental basis for making a decision or taking action. Accordingly, most government departments have a clear method of measuring the services they provide. For example, the police can measure the number of enforcement actions, the fire department can measure the number of establishments inspected and fires fought, and the education department can measure the number of pupils taught as well as their performance on standardized tests.
Until now, park and recreation departments have had limited tools to assess the use of the services they render to their communities. Park departments provide a wide array of community assets, maintain multiple facilities and coordinate many programs, yet they seldom can precisely measure how many people are being served. Without benchmarks and measures, it is impossible to know whether a department is meeting the needs of its community members or whether it is doing better or worse, from one year to the next.
Typically, parks only have partial information about how their facilities are used. They can count registered participants of park-sponsored programs and get lists of team members for groups that obtain permits to use park facilities, but this method will likely assess only a fraction of users, since along with registrants, many family members, friends and acquaintances also visit parks to observe or take advantage of other park facilities.
Another method has been to conduct population-based surveys and ask people about their use of park facilities. However, these methods may not be representative, may capture only a fraction of the local residents and may not yield precise information about the specific facilities that are being used. Surveys also rely on the memory and ability of individuals to accurately report their park use.
Yet, methods exist that provide robust and comprehensive assessments of park use, and these could be adopted by park and rec departments to document park use, not only by registered users, but also by the population. Following are some of these methods, including potential logistics that would be feasible for park and rec departments to adopt.
Why Measure Park Use?
There are many benefits to being able to measure the uptake and utilization of recreational services and facilities. It provides accountability by showing the public and elected officials the return they are receiving on their investments through tax dollars. Although park departments tend to be underfunded, they still receive substantial funding and showing how these investments benefit the community is critical to increasing trust, as well as potentially being able to demonstrate the impact of changes in investments. For example, if a park had to be closed, it would be clear how many people would be denied access to services, or, conversely, if an additional sum of money were given to the park for a special program, it would be possible to estimate the additional number of people who could be served with these additional resources. It’s one thing to claim that everyone living within a quarter mile of a park would be influenced, but not everyone uses the available facilities.
Learning more about usage patterns and their fluctuations could be highly valuable to park planners and managers and to understanding how to best optimize the limited resources that are available. Routine measurement can also inform park managers about other issues related to park use; for example, there may be barriers that are not obvious to those who only have limited contact with parks. There may be intimidating groups of people, mentally ill or intoxicated patrons who may discourage others from using the park. Maintenance problems, like litter, graffiti, overgrown lawns and broken equipment, can also be assessed.
Goals of Measurement
The main goals of measurement should be to quantify the use of facilities and to describe the characteristics of their users. Park-use measurement can be readily implemented through systematic observation. By selecting representative times of the day and days of the week, it is possible to get a general estimate of park use. Many studies have been conducted assessing park use, identifying peak-use periods and low-use periods. If these varied park-use times are sampled, it is possible to estimate the total park use over time.
Park facilities can refer to any park element or structure, including lawns, fields, courts or play areas. User characteristics refer to attributes, such as gender, age group and race/ethnicity. This level of information would provide evidence of the equity of services: are men served more than women, children more than seniors or minority populations more than majority?
More detailed assessment of characteristics could also include a tally of activities, either summarized as a level of energy expenditure (e.g., sedentary, moderate physical activity or vigorous physical activity), which is an important measure when it comes to health and well-being, or described as a type of activity (e.g., tennis, basketball, picnicking), which provides substantive information on recreational preferences. This information can provide insights about which facility investments are favored by residents and can guide planning for staffing, programming and future renovations. All this data can be used to justify budgetary needs and expenditures.
Logistics of Measurement
The method of measurement consists of counting and categorizing individual use at specific points in time. In previous research studies on park use, the most convenient approach was to count park users up to three or four times per day at three-hour intervals for a minimum of four days, including at least two weekdays and one weekend day. The aggregated results could provide a relatively accurate assessment of park use over a one-week period. However, the limitation of these closely spaced observations is that the findings may not generalize to different seasons when the parks may be used (or unused due to inclement weather) or when the attraction of the park might vary based on seasonal sports activities.
Although jurisdictions may have hundreds of parks, it isn’t necessary to collect data on each park to understand how the parks are being used. A park system may be interested in tracking how its facilities are used. It may want to get a lot of detail about a specific park or about specific areas within one or more parks. Typically, a sampling strategy consists of randomly selecting a small percentage of the parks. A random selection should help ensure that the sample is representative and the findings may be applicable to all parks in the system, even those that were not specifically studied.
Sampling Theory and Representativeness of Distant Observations
Conceptually, a sample is representative if it can be used to accurately estimate the characteristics of the target population, which is the actual interest for measurement. Among the many technical issues in sampling theory, there are three critical issues in achieving representativeness.
First, the sampled population, from which the sample is drawn, should match with the target population as best as it can. For example, suppose that we are interested in studying all parks and facilities managed by the park department. Because of logistical constraints, we may decide to sample only the parks with on-site staff. Then, the target population (all parks) is different from the sampled population (staffed parks), and the sample would not be representative for the target population. In practice, a perfect match between the sampled and target population is usually very difficult. However, evaluators should make the best efforts to align these two populations.
Second, whenever possible, we should always use a probability sample instead of a convenience sample. With a probability sample, everyone in the population has a chance to be chosen, and the drawing process is random. Any sample that is not a probability sample is a convenience sample; for example, choosing parks because some residents who live nearby volunteer to help is a convenience sample. Note: Subjective opinions in the sampling process, if not incorporated properly, will result in convenience samples.
Third, the chosen sample should be sufficiently large to estimate the population characteristics. Statistically, it may be reasonable to sample around 10 percent, if the population is not too large (e.g., sample 10 parks if the city park system has 100 parks). If the population is small, a reasonable sample may be more than 20 percent (e.g., sample four parks if a system has 20 parks). Likewise, if the population is huge, the sampling proportion should also be smaller (e.g., sample 50 green spaces if the city has more than 1,000 public green spaces). A sample must have a minimum of two but usually should have four or more parks. If the park system is very heterogeneous, the sample size should be increased to ensure most or every type of neighborhood will be represented.
Besides sample size, the spacing between observation needs to be considered to provide more information. An estimated three to four hours is necessary between consecutive measurements in the same park, because measuring the parks during consecutive hours may be redundant (e.g., 10 a.m. and 11 a.m. will likely not be very different compared to the observations taken at 11 a.m. and 3 p.m. Geographically sampled parks should also not be close to each other, since it is likely they serve the same populations.
Conclusion
Measuring park use is a key step toward putting this sector of government service provision on par with other sectors that are more likely to be considered essential to the lives and well-being of residents. With concrete numbers to support the demand for park services and facilities, park and rec departments can justify the need for park staff, maintenance, and additional parks and facilities. If equity is a goal, the numbers can also provide guidance on how to better develop and implement services. For example, parks that are not reaching women might need some more exercise classes, or parks not visited by seniors may need to add benches or have their accessibility issues addressed. Park assessment will provide important benchmarks to guide park planning and management strategies.
Customizing Measurement Logistics to Meet Specific Assessment Needs
Following are a range of scenarios and potential strategies that can provide the park-use information desired:
Monitoring the entire park system: If the system has approximately 200 parks, a 10 percent sample will usually be sufficient to get a robust picture of the park use. However, for systems with less than 200 parks, a minimum of 10 – 20 parks is recommended.
The frequency of assessment should be spread throughout the year and should include a variety of different days and times, included in mornings, afternoons and evenings, as well as weekdays and weekends. By spreading out the observations over time, it is possible to get an estimate of park use over a longer period and, possibly, across different seasons. Roughly 24 observations (e.g., two times a day, one day per month) for each park may be sufficient to understand overall park use.
Assessing the use of small parks or plazas: This could be applied to a single park or several parks that have similar characteristics. In parks where turnover is more rapid (i.e., where people tend to stay for shorter periods), it is necessary to conduct more frequent observations to better capture the changes. For a small park or plaza where the average person may stay less than one hour, observations may be scheduled every two hours over several days for 20 – 40 hours of observations per park. The higher end of observations assumes a greater level of turnover.
Assessing changes in park use after renovation or other changes: In this situation, it is important to conduct assessments before and after the change. Depending on the scope of the changes and the length of time it takes to make the changes, the spacing between the pre- and post-assessments may be months or years. If there are strong seasonal variations in park use, it would be critical to conduct the pre- and post-assessments during the same season, and, ideally, at the same times of day and week. Here, we recommend a minimum of 12 – 16 hourly observations before the change is implemented and another 12 – 16 observations after, synchronized to the same time of a week.
Because of potential novelty factors, it may be best to perform the post evaluation several months after the change, rather than immediately after, as the highest use of the park may be driven by curiosity, but may not be sustained after the first few months. It is also advisable to have comparison parks, which have not undergone any changes, to have a more robust assessment. Without a comparison park(s), it will be difficult to determine whether any difference in park use is because of the changes in the park or are a consequence of natural trends in park use.
Assessing seasonal use: In many cases, the use of parks varies considerably across seasons. For a more valid assessment of these differences, we recommend at least 12 observations each season.
Summarizing data collected: The primary data collected using SOPARC (System of Observing Play and Recreation in Communities) is the number of users, the characteristics of users by gender and age group, physical activity levels, race/ethnicity (if desired), types of activities and time of day for each of the park target areas. The data can be aggregated by any one of these factors. Because the observations are only a sample of the park use, the total use can be imputed by multiplying the average hourly observations times the total number of hours over which the sampling was done.
If the hours sampled were representative of 12 hours per day, seven days a week, then the average hourly use would be multiplied by 84 to estimate the weekly use over this period (in the unit of person*hours). Furthermore, the estimates can be improved by separately weighting the weekdays and the weekend days, multiply the average hourly use by 12 hours and five days and the average weekend use by 12 hours times two days, and then summing the weekday and weekend uses. This can be applied to all the subcategories and characteristics of users. The results can be examined by the target area or the type of facility as well. Comparisons can be made across parks, between different time periods and different populations groups.
Deborah Cohen, MD, MPH, is a Senior Scientist at the RAND Corporation and the Principal Investigator of the National Study of Neighborhood Parks. Bing Han, Ph.D., is Senior Statistician for RAND Corporation.