Table 1 Possible sources and examples for each error type
From: Methods to account for uncertainties in exposure assessment in studies of environmental exposures
Error Types | Possible Sources | Examples |
|---|---|---|
Variability | Differences in individual’s location, exposure or behavior, randomness, etc. | Individual-specific exposure estimates differ with distances from the pollutant source [12]; Exposure to pesticides or bacteria may vary by season [12]; Different patterns of food intake may result in different exposures across individuals [12]. |
Uncertainty | Lack of knowledge in specifying exposure pathways, simplified model assumptions, failure to account for possible correlations between variables, etc. | There is uncertainty in the level of exposure to insecticide sprays due to the unknown exposure pathway (inhalation, dermal contamination or both) [61]; There is uncertainty when calculating one’s inhalation rate because of the failure to account for the dependency of body weight and breathing volume [61]; There is uncertainty in the estimated room air concentration because of the unknown release rate of the chemical. [61]. |
Shared Error | Incomplete knowledge about the parameters that affect the exposure measurements of group. | Inaccurate estimations of the ground deposition of certain contaminants may affect the estimation of exposure for all people who live in the same area [18]; Errors from an uncertain of a biased measuring device when it is used to a group of people [54]. |
Unshared Error | Lack of knowledge about the parameters that vary randomly between subjects. | See the examples for classical error, Berkson error and unshared non-random error. |
Classical Error | Imprecise measuring device, repeated measurements that vary around the true value, etc. | Using the replicated urinary nitrogen as a measured biomarker to investigate the true long-term dietary protein intake, etc. [3, 44] |
Berkson Error | The same exposure value is assigned to a group with similar characteristics. | Air quality records collected by a monitoring station are assigned to all subjects in the study as estimates of true individual exposure to pollutants [62]; When job-exposure-matrix is used to estimate the individual exposure in occupational epidemiological studies, same exposure estimate is assigned to the groups of people with same occupation code [17, 63]. |
Unshared Non-random Error | Imprecise knowledge in individual specific parameters. | Errors in personal residence history records [7]; Errors in personal consumption rates of contaminated foods [7]. |