Table c1 Strengths and weaknesses of common methodologies used to measure the drivers of energy imbalance
Strengths | Weaknesses | |
|---|---|---|
Study sample | ||
Cross-sectional studies | Less costly, usually easier to include larger number of people, timely results. | Exposure and outcome measured at the same time, causal relationship unclear. |
Longitudinal studies | Better inference for causation by associating changes of exposure and outcome over time | More costly. Loss of follow-up (sometimes selective) |
Representative sample (for example, nationally representative) | Generalizability | Greater heterogeneity, sample size can be small for key subgroup analysis |
Selective sample (for example, obese, diabetic or Pima Indians) | Relevant to specific study questions. Convenience of sample | Generalizability |
Measurement method | ||
Recall/self or proxy report (for both diet and activity) | Easier to administer | Omission or inaccuracy, subjectivity |
Direct observation (for both diet and activity) | Objectivity. No recall bias | Labor intensive. Subjects may change behavior with the knowledge of being observed. Some inaccuracy remains. |
Objective measurements of physical activity (for example, accelerometers, DLW) | Objectivity and accuracy | Costly. Thus, usually smaller number of subjects are studied |
Macro-level measure: food balance sheet | Objective, use of publicly available data | Lack of individual level inferences. Many sources of biases in calculation |
