— Resources for ResearcherS
Participants
Sample size and sampling strategy
The sampling design for studies looking at the association between environmental attributes and physical activity behavior is multistage. Firstly, administrative units (e.g., census tracts) that match specific physical and social characteristics are selected. Secondly, residents are randomly selected from these units. This type of sampling scheme includes two kinds of sample sizes: the sample size of the macro-units (N; units) and the sample size of micro-units (n; residents) within each macro-unit, with N x n being the total sample size for the micro-units.
We need to establish the two sample sizes that will yield an acceptable level of statistical power (0.80) to detect individual-level (e.g., what is the relationship between perceived access to shops and walking for transport?) and individual-neighborhood cross-level relationships (e.g., what is the relationship between the average neighborhood street connectivity and walking for transport?) with minimal cost.
The suggested samples sizes were empirically derived using data from the NQLS and PLACE studies. The table below reports some sample sizes needed to achieve a power of at least 0.80, assuming that the individual-level and neighborhood-level predictors each explain 1% of the variation in the outcome and that the intraclass correlation for the outcome is 0.05.
The table shows that because the expenses of a study are largely determined by the total sample size, it is more cost-effective to aim for a large sample of tracts (if possible) than a large sample of residents within tracts. Because we will be pooling data from many countries, we are recommending a sample size per country of approximately 500 per age group (i.e., younger adults, older adults). If specific neighborhoods are studied, we are recommending approximately 50 per neighborhood.
| N (number of neighborhoods*) | n (sample size of residents within neighborhood) | N*n (total sample size) |
| 50 | 70 | 3500 |
| 55 | 50 | 2750 |
| 60 | 35 | 2100 |
| 65 | 30 | 1950 |
| 70 | 25 | 1750 |
| 80 | 20 | 1600 |
| 90 | 15 | 1350 |
| 115 | 10 | 1150 |
Implementation of Sampling Strategy
The sampling frame should be constructed in several steps.
- A list of all street addresses in targeted administrative units should be constructed. This can be completed by obtaining a published city directory in conjunction with maps or some marketing companies and government agencies are able to provide contact information at the administrative unit level.
- Using the available contact information as a proxy for the available recruitment pool in each administrative unit, units should be selected so that recruitment targets can be met in each quadrant. An estimate of a response rate will be necessary. This will help you decide how many units will need to be selected in each quadrant.
- Once dwellings are enumerated within the selected units, they should be randomly selected for contact by the research team. Only one adult per residence should participate. The adult with the most recent birthday can be targeted to allow for random selection within each household. Computer listings for sampled dwellings should then be compiled to facilitate fieldwork and tracking of respondents.
- It is important that recruitment takes place in both high and low walkable areas simultaneously. Preferably, recruitment will happen simultaneously in all neighborhoods.
Recruitment
Inclusion criteria
For the adult study, participants must be between the ages of 20 and 65 and provide signed informed consent. They also need to have lived in their neighborhood for at least 3 months before participating.
Exclusion criteria
Individuals in group living establishments (e.g., nursing homes), unable to walk, or unwilling to wear the accelerometer; should be excluded.
Random selection of participants
Once your contact lists are compiled, there may be residential listings that do not include telephone numbers. Using a random number generator in Excel, target samples of telephone and non-telephone residences should be selected for each administrative unit to be contacted.
Mail and phone recruitment
Households should be mailed informational packets, including a consent form. If there is no telephone number, include a brief questionnaire soliciting phone number, age and gender of adult residents and wait for it to be returned to initiate phone contact. A few days later, telephone interviewers should contact each household to introduce the project and encourage participation. Recruiting the adult with the most recent birthday should result in a sample of adults stratified by gender. If the selected adult isn’t interested or eligible, another adult can be recruited. Up to eight callbacks can be made to each household. Methods of in-person recruitment have been used by some countries.
A secure electronic list of names, addresses, and telephone numbers of those who agree to participate should be constructed for tracking purposes. Refusals and not eligible outcomes should also be tracked, along with basic demographics such as age and gender if possible. Once informed consent is received, materials can be mailed.
Samples should be balanced by gender (i.e., approximately 50/50 female/male within each quadrant) and should approximate the ethnic distribution of the population. When the gender quota for a quadrant has been met, only participants of the opposite gender should be recruited. Recruitment should also be balanced by walkability with approximately equal numbers in each walkability/SES quadrant. When a quota for a quadrant has been met, recruitment should only proceed in remaining quadrants.
Population Issues
| Adults | Youth | Older Adults | |
| Physical activity measurement | 18-65 years | 4-18 years | 65+ years |
| Built environment measures | IPAQ well validated across countries | NEWS-Y | NEWS with some adaptions |
| Self report issues | NEWS | Likely unreliable in younger children, parent reports often sought | Consistent and simple response options required |
| Other considerations | None | Parental consent and participation required | Measures of physical functioning or mobility required |
| Compliance | None | Low | High |
| Neighborhood recruitment | Easy to find sufficient number of adults in contiguous administrative units | Harder to get information on children’s age and families less likely to live in some neighborhoods | Harder to find older adults in contiguous administrative units, retirement community living should be assessed separately |
Seasonality
To account for variations in physical activity due to weather and time of year, we suggest that physical activity measurements for each participant are taken twice, at least 6 months apart. Although this is preferred, it is not always feasible to collect data at 2 separate time points. Therefore, we also suggest a sampling strategy that results in participants being recruited across all walkability quadrants throughout the course of the data collection period. If it is possible to complete data collection in one season, summer or spring is recommended.