MEASURES REGISTRY USER GUIDE
Researchers and practitioners must consider several important issues as they select food environment measures for a project. This section discusses matching measurement choice to the overarching project purpose, the population to be targeted, food environment domains or venues of interest, data collection and analysis resources, and the health behavior(s) or outcome(s) of interest.
The place to start in deciding what measures to use is to carefully define the project purpose or specific research question. An objective that is too broad (e.g., “To understand how the food environment affects students’ consumption of sugar-sweetened beverages”) will likely result in time and resources that are poorly used. In comparison, a more specific objective (e.g., “To understand how availability of sugar-sweetened beverages in school vending machines is related to student intake of sugar-sweetened beverages”) will provide a clear direction for what measures are needed to answer the question. Careful conversations among the team are needed to clarify the focus and intent of the project before choices are made about the measures to be used. The most successful projects start with clearly identified aims that are targeted toward specific change, but broad enough to have population health relevance. There should be a plan for how to use each piece of data collected. The purpose of the project and the question to be answered should drive the decision about which measure to use; choosing a measure and then trying to match it to some purpose is rarely a fruitful endeavor.
Population and Venues of Interest
As the team considers food environment measures that fit the project purpose, a consideration of the population of interest is very important. If the team is interested in young children (preschool or early elementary age), the physical environment of the home and school is important to assess. As children get older, stores and restaurants surrounding schools may also be important to assess.44 Relevant social environments are also affected by the age of the population of interest. Younger children are heavily influenced by their parents and the important adults in their lives, while older children are heavily influenced by their peers and the larger culture. Age is important to consider when elements of the person-centered environment will be assessed. Reading and cognitive abilities differ greatly across the age-spectrum of youth. Very young children may not be capable of describing their perceptions of their food environment, as the ability to think abstractly does not develop in some children until about age 12 years.45
Another important step in food environment measure selection is the access the team has to the population of interest. Access to homes, early child and education centers, preschools, schools, and community venues to assess the availability and accessibility of foods to youth is limited by the willingness of parents, school administrators, and community leaders to let a team of evaluators into their space. Access to stores, restaurants, and other public spaces may be easier, but in many cases permission must still be granted or data collected covertly. Assessing elements of the social or person-centered environments also typically requires cooperation and consent from the population of interest. Related to this consideration is how the population of interest is sampled. Will the purpose of the project be adequately served using a convenience sample or is a more representative sampling required? Careful attention to how such access to spaces and individuals is obtained is a critical consideration for the team.
Data Collection and Analysis Resources
For any type of data collection, identifying the sample on which to collect data and identifying how to connect and obtain permission or consent to collect data is always an important issue. In addition, being clear about how many data collection periods are necessary and, if more than one data collection period will be used, the time period between data collection periods is important to know. A sense of when data collection will begin and when it needs to be completed to meet project timelines is also essential. The number, skill level, and training needs of data collectors is an important consideration. The anticipated cost of data collection, including personnel costs, travel, phone and postage costs, and the cost of any incentives that might be needed to encourage participation must also be factored into resource needs. Similar issues must be considered with regard to data analysis. In particular, the cost of any specialized software and the level of sophistication needed for data analysis need to be considered.
Health Behavior or Outcome of Interest
As measures are selected for either a research purpose or a more practice-based purpose, it is important to be clear about the relationship of a measure to an important social, behavioral, or biological health outcome. Being able to measure something that is not related to an outcome of importance is both irrelevant and an inefficient use of resources. As an example, one could spend a great deal of effort to accurately assess the amount of shelf space dedicated to a particular food item, but if shelf space is not related to intake of that food, the degree of accuracy afforded by that measure is irrelevant. It is also important to consider whether the outcome can be expected to change within the timeframe of the study. If not, a more proximal change should be considered. Expecting change at the biologic level (for example weight or BMI) from an environmental intervention may be unrealistic, especially in youth. Change in behaviors, perceptions, or beliefs may more likely to be seen within a shorter timeframe.
Suggested Process for Using the Measures Registry
Searching and Filtering Results
The NCCOR Measures Registry allows users to search for measures within four domains: Individual Diet, Food Environment, Individual Physical Activity, and Physical Activity Environment. For each domain, measures are organized according to “Measure Type” (i.e., measurement method), “Ages” covered, and "Context" (i.e., urban or rural). The check boxes within each category can be used to query the database and narrow the results fields.
Once the user selects the desired measure type and context, a list of measures fitting those criteria is displayed. When a measure has been evaluated (i.e., tested for reliability or validity), the Registry will most likely include the measurement development/evaluation publication. If this does not exist, the Registry will include the first paper published using the measure. Many of the publications in the Registry do not include measurement/evaluation studies or report on psychometric properties of the measures. The “Search” feature of the Registry can be used to narrow the query beyond the standard criteria, for example, a particular venue (i.e., home, school, preschool).
Within the food environment domain, the Measures Registry includes six measures types when the food environment is selected (GIS, 24-hour dietary recall or food frequency questionnaires, environmental observation, questionnaire, record or log, and other). Although the types of measurement tools that would be found within the GIS and questionnaire measure type categories are fairly well defined by those terms, the distinction between the type of measures that would be found under “environmental observation” and “records and logs” is less clear. As an example, NEMS-S is a widely used measurement tool used to document the extent of and types of foods found in a store. It is typically called an “audit tool” but could also be viewed as an environmental observation, record, or log. Articles using NEMS-S are found when the search term is “stores” and either “environmental observation” or “record or log” is checked as measurement type. In such cases, the user might need to examine several types of measurement tools to identify an appropriate measure that could fit under several categories.
The Measures Registry also includes four age categories of interest when the food environment is selected (2–5 years, 6–11 years, 12–18 years, adults) and two contextual categories for food environment measures (metro/urban, small town/rural). Given that measures are often used across these categories, the user might need to examine all categories or use the "Search" function to thoroughly review available measure choices.
Navigating the Information Tabs Within Each Publication
Clicking on a publication’s title will open a link with more detailed information about the measurement. It is recommended that each tab be viewed in detail while keeping in mind the selection considerations outlined above. The “At a Glance” tab includes helpful information when available, such as the length, constructs covered, and how to obtain the measure. The “Study Design” tab reports the characteristics of the sample used to develop and evaluate the measure, so users can consider whether the tool is appropriate for the population they intend to study. A "How to Use" tab includes information on how the measure is administered and whether data collection and/or analysis protocols exist. In circumstances where the Measures Registry does not include a link to the measure or protocols, the user should contact the authors of the study. Finally, the “Validity” and “Reliability” tabs include specific results from the publication on the tool’s measurement properties. If a tool has multiple publications in the Measures Registry, the user should view the tables for each publication.
The Compare Function
Another useful feature of the registry is the "Compare” function. This function allows users to identify multiple measures and make comparisons among them. As an example, assume that users want a measure to assess elements of a preschool environment. Including "preschool" in the search function may yield six measures (one “environmental observation,” two “questionnaires,” and three “other”). A user reviewing the listing of these six tools under the "Results" heading may try to decide between three of them ("Food outlet accessibility for low-income preschool children"46; "Home food inventory for preschoolers"47; or the "Home-inventory: describing eating and activity development for preschoolers."48) The user can click the “Compare” button and a “Comparing Measures” document is immediately available that shows how the three studies compare with regard to a variety of factors, including the availability of psychometric properties and the sample on which each was evaluated.