MEASURES REGISTRY USER GUIDE
Measures with Evidence of Reliability and Validity
This section highlights a sample of measures for each environmental level setting that have some evidence for reliability and/or validity. It is important to note that this information is not based on a comprehensive review, and that measures not listed here may also have strong measurement properties.
Physical Food Environment
As mentioned in Section 3, two types of measures are used to answer questions about the physical environment: (1) geo-spatial analyses such as GIS and (2) observational scans or assessments (also called logs, records, or audits) of food product availability, pricing, placement/merchandising, advertising, and information in stores, restaurants, homes, schools, or community venues.
Geographic Information Systems
The majority of food environment research with a focus on the physical environment has used GIS data as the measure. In their review of the food environment research from 1990 to 2007, McKinnon et al.12 found that of the 137 articles identified as assessing the food environment, nearly half used GIS. Another review, which picked up when the McKinnon review ended, examined the food environment literature from 2008 to 201513 and found that GIS continues to dominate the field as a measure used to assess the food environment. This updated review showed that of the 432 articles published between 2008 and 2015, 65 percent used GIS. Many useful and important findings have resulted from an examination of the food environment using GIS. GIS data have helped reveal that neighborhood socioeconomic, racial, and ethnic composition is related to food access and availability in many neighborhoods, revealing the existence of food deserts in many low-income neighborhoods.25-29
However, GIS’s limitation related to construct validity is being revealed as more evidence builds in the field. Gamba et al.14 studied 51 peer-reviewed articles that examined the relationship between obesity risk in communities and the community nutrition environment as assessed using GIS (defined by the presence of stores; the proximity to stores; and the density, count, and ratio of types of stores within given neighborhoods). Of the total number of associations between the environment and obesity evaluated across these 51 studies, only 32 percent of the associations were in the expected direction (i.e., healthier environments were associated with lower obesity risk), 10 percent were in the unexpected direction, and 58 percent showed no association between the environment as assessed by GIS and obesity-related measures. Therefore, while GIS is commonly used, its utility as a predictor of obesity in a population is poor.
Observational Scans or Assessments in Stores and Restaurants
One of the most commonly used measures to assess the physical environment within retail food stores is Neighborhood Environment Measurement Study-Stores (NEMS-S).30 The NEMS-S is an environmental observation form designed to measure the availability and prices of milk, meat, frozen dinners, baked goods, beverages, breads, chips, and cereal. It also measures the availability, prices, and quality of fresh fruits and vegetables. An online training is available at the NEMS website. Both inter-rater and test-retest reliability were assessed as part of the development of the tool and found to be robust (rates of agreement for both inter-rater reliability and test-retest is 76 percent or higher). In a recent review of 128 studies examining measures of the food environment,31 27 reported on using NEMS-S as the observational scan and many of those articles reported on the reliability and validity of the instrument. However, the vast majority of both the reliability and validity information reported for NEMS-S in this review was based on the developmental work with NEMS-S. Very little independent reliability and validity testing was done to confirm the psychometric properties of the measure. Future efforts should begin to establish construct validity and to establish the utility of the measure to assess change.
A similar tool was developed to assess the physical environment related to restaurants (NEMS-R).32 This tool is an environmental observation that uses trained data collectors to collect data on factors believed to contribute to food choices in restaurants, including availability of healthy foods, facilitators, and barriers to healthy eating, pricing, and signage/promotion of healthy and unhealthy foods. Inter-rater reliability and test-retest reliability were assessed and found to be satisfactory. Construct validity was found to be in question as the tool showed that fast food restaurants had greater healthy entrée availability and main-dish salad availability as compared to sit-down restaurants.24
Environmental Observations for Homes, Schools and Preschools, Including Records, Logs, and Questionnaires
Homes: Very few measures attempt to assess a broad range of foods available in the home. Although some measurement tools focus on fruits and vegetables33 and some focus on prepared foods,34 the home food inventory by Fulkerson, Nelson, et al.35 is one of the few instruments available that attempts to assess a full range of foods in the home. They have also used data collected from the home food inventory to compute an obesogenic score for ranking home environments. Testing of the measurement tool showed good inter-rater reliability for a variety of food categories assessed (level of agreement between raters ranging from 0.61 to 0.83). In addition, criterion validity was assessed by having the instrument completed at the same time in the home by trained observers and parents. The level of concordance was compared with the trained observers considered to be the gold standard. For six food categories assessed, criterion validity ranged from 0.71 to 0.97. Construct validity was assessed by comparing four categories of food present in the home with dietary intakes of parents (using the National Cancer Institute’s Dietary History Questionnaire) and youth (using 24-hour recalls). As expected, availability of foods in the home was positively related to dietary intake, and a higher obesogenic score for the home food environment was associated with higher caloric intake of both parents and youth.
Schools: A few measures have been developed and tested to document the physical environment of schools. The School Nutrition and Dietary Assessment Study (SNDAS) funded by USDA36 and the CATCH study (a large, multi-centered school-based study)37 both reported on methods to document the foods available from USDA reimbursable meals. Other studies have used environmental observations to document the competitive food environment of schools, including foods available in vending, à la carte, and school stores.38 Many of these instruments have been tested using longitudinal designs and shown to be robust. Several of those instruments are available in the Measures Registry.
Early Care and Education Centers and Preschools: Measures are available to assess foods available at early care and education centers and preschools. The Environment and Policy Assessment and Observation (EPAO) is an environmental observation measurement tool that allows assessment of several food categories (including fruits and vegetables, sugar-sweetened beverages, and foods of minimal nutritional value) as well as the existence of marketing or promotional messages.39 Although no assessment of validity is available, the measurement tool shows good reliability. Another useful measurement tool for early care and education centers and preschools is the Nutrition and Physical Activity Self-Assessment for Child Care assessment tool (NAP SAAC), a self-administered survey that allows services to evaluate their own nutrition and physical activity environment.40 This measurement tool provides feedback on both the physical and social environment of early care and education centers.
Social Food Environment
Measures to assess the social food environment can be broadly grouped into three types: (1) social support, role modeling, and social expectations; (2) policies, practices, or rules about eating behavior within venues; and (3) parenting practices around meal time and foods available to youth.
Social Support, Role Modeling, and Social Norms
Information on social norms and social support are often collected as part of surveys or questionnaires that youth complete. These types of psychosocial measures are widely reported in the literature and much of the published literature on these measures includes information on the questionnaires’ psychometric properties. Finding the most appropriate measurement tool should include a consideration of the specific type of eating behavior that is being assessed, the age of the respondent, and importantly, an understanding of who the most relevant referents are for the youth.
Policies and Practices
Several measures are available to assess the social environment of schools and the CDC’s School Health Policy and Practice Survey (SHPPS) is one of the most thorough. SHPPS is a national survey, administered as a questionnaire over the phone, that is periodically conducted to assess school health policies and practices at the state, district, school, and classroom levels. It was first conducted in 1994 and most recently conducted in 2014. The level at which data are collected has varied over the years. Construct validity for sections of the tool was examined and confirmed in an article by Taber et al.41 where they found that changes in state policy as assessed through SHPPS was associated with changes in student soda consumption.41 The Nutrition Environment and Services section of SHPPS might be of most interest to those exploring elements regarding the food environment and includes policies and practices related to foods offered on à la carte and vending, school stores, and in the school cafeteria.42
Another important social environment for youth is parenting practices that involve meal time and feeding practices. Several measures have been used and found to be reliable and valid. One of the most commonly used measurement tools is the Child Feeding Questionnaire (CFQ).19 This self-report measurement tool is a questionnaire assessing parenting beliefs, attitudes, and practices regarding child feeding with a focus on obesity risk. In addition to internal consistency of the questions included, construct validity has been established. Factors assessed in the CFQ designed to measure parental concerns and beliefs regarding the child’s risk for obesity were significantly and positively related to the child’s weight status. In addition, parents’ reports of their use of control in feeding their children were also related to the child’s weight status in the expected direction. This tool also has been tested with an Hispanic sample.19 Lytle et al.18 created an index to assess multiple elements of positive family meal practices. This self-report questionnaire asks parents to report on a wide range of practices around family meals, including foods typically offered, rules around talking on the phone or using phones while at a family meal, and parents’ usual presence at meal times. The index showed good construct validity; higher scores on the index, indicating healthier meal practices, were associated with lower BMI of both parent and adolescent.
Measures to assess the person-centered environment include those assessing individual’s perceptions of their physical and social environment. There are few measurement tools available and those that are available often assess only a few elements of the physical food environment. Green and Glanz43 recently developed and tested the Perceived Nutrition Environment Measures Survey (NEMS-P). This instrument was designed to assess perceptions of the availability of foods in stores, restaurants, and homes; the price of food in stores and restaurants; the promotion of healthy items in stores and restaurants; and the accessibility of healthy and less healthy foods in homes. Test-retest reliability was evaluated and found to be moderate to good. Construct validity was assessed by asking residents of four neighborhoods that differed by socioeconomic status (SES) to assess their own community nutrition environment, the store consumer environment, the restaurant consumer environment, and the availability of foods in the home environment. Although ratings of store and restaurants did not differ by community, residents of higher-SES neighborhoods reported higher availability scores in stores, a stronger belief that healthy items were available in restaurants, and higher scores for access to healthy foods in their homes as compared to the residents in lower SES-communities. This finding suggests discriminant validity, a type of construct validity. However, construct validity showing the relationship between perceptions and food choices or health outcomes was not demonstrated. Insufficient research has been conducted with this measurement tool to determine its future potential.