Direct observation provides detailed information about frequency, duration and intensity of bouts of activity occurring during the period of observation. This method exceeds most methods in its capacity to capture the type and context of physical activity, including components such as where and with whom it occurs. Direct observation can be tailored to fit the needs of the specific research question and in particular the behavioural and contextual variables of interest. The dimensions of physical activity assessed by direct observation are described in Table P.3.17.
Table P.3.17 The dimensions which can be assessed by direct observation.
|Dimension||Possible to assess?|
|Total physical activity energy expenditure||✔|
|Timing of bouts of activity||✔|
|Contextual information (e.g. location)||✔|
Direct observation requires recording of the dimensions of physical activity by an (ideally independent) observer in real-time. Systems of direct observation vary by the characteristics set out below.
Time frame and setting
The time frame and setting of direct observation are constrained by the necessity for the observer to be present at the time of monitoring. This requirement means that this method is generally used in more controlled settings, such as schools or care homes, although some instruments are designed for coding activity in diverse settings.
It follows that observation periods are typically relatively short, as observation for long periods is burdensome for the participants and researcher alike, and could also be intrusive. Costs of data collection and analyses for long periods of observation must also be considered. Direct observation is therefore not typically suitable for recording total daily physical activity. Although individual observation periods may be of limited duration, it is possible to conduct multiple assessments in order to account for some of the intra-individual variation in behaviours and the effect of factors such as season, weather, and time of day.
These limitations mean that it is more suitable for research questions focusing on particular time periods and settings, for example, examining the number of minutes of MVPA recorded by school children during recess.
Observation sampling frequency
Sampling can be continuous or interval based:
Categorisation of behaviour
The dimensions of physical activity recorded by direct observation vary by system and should be selected to suit the requirements of the research question. Some instruments record intensity category (e.g. sedentary, light, moderate, vigorous), while others record aspects of posture (e.g. sitting, lying, standing) or type (e.g. walking, running) of activity. Some instruments are even more detailed, and many record aspects of the context in which activity occurs.
Electronic vs. paper-pencil
Typically an observer will watch the participant(s) using a specific observational system and record a rating of physical activity into a laptop computer or by pencil into a coding form. A number of systems are available and have been comprehensively reviewed (McKenzie, 2002; Oliver et al., 2007). Electronic systems allow more rapid entry through key strokes or mouse clicks, with time stamps added automatically. Key advantages of electronic direct observation include:
Physical activity estimates are dependent upon instrument used, but can include:
Interpretation of observational data is aided by additional information, such as:
It is usual for MET values to be assigned based on the type and/or intensity data reported during observation. MET scores for each interval of observation can then be summed, either in total to provide total energy expenditure or grouped according to dimensions such as intensity category, type, or context to identify relative contributions. Since timing of activity is recorded, they can also be used to describe change in energy expenditure during the observation period. Research assistants who will be assigning MET values should be trained and assessed first to ensure inter-rater agreement and high objectivity.
This approach assumes the following:
Considerable time and effort is required to conduct observation studies but advances in technology have increased the potential of this method to assess physical activity but mostly in controlled settings. An overview of the characteristics of direct observation is outlined in Table P.3.18.
Table P.3.18 Characteristics of direct observation.
|Number of participants||Small|
|Researcher burden of data collection||High|
|Researcher burden of data analysis||High|
|Risk of reactivity bias||Yes|
|Risk of recall bias||No|
|Risk of social desirability bias||No|
|Risk of observer bias||Yes|
|Participant literacy required||No|
Considerations relating to the use of direct observation for assessing physical activity are summarised by population in Table P.3.19.
Table P.3.19 Physical activity assessment by direct observation in different populations.
|Infancy and lactation|
|Toddlers and young children||Activity can be highly intermittent – frequency of recording may need to be higher.|
Given the number of observational systems available, one of these should be used and adapted as required rather than a new one developed. Advances in technology have resulted in several bespoke software packages to collect, handle, analyse and present observational data. PDAs are particularly suited to field work and real time recording of physical activity. It is also possible to collect contextual data with these packages.
A list of specific direct observation instruments is being developed for this section. In the meantime, please refer to the overall instrument library page by clicking here to open in a new page.