Aspects of the design and data collection methodology of the National Survey of Lead and Allergens in Housing carry significant implications for the statistical analysis of data emerging from the survey. Examples include the following: The design necessitates the use of established, specialized approaches to achieve unbiased estimation and variance calculations in complex sample surveys, while selected analyses move beyond the capabilities of commercially available software. Insufficient dust amounts to support assays of all allergens from all samples necessitates consideration of randomization algorithms for assigning allergen assays, as well as and other measures to minimize the potential for introduction of bias. The prevalence of allergen concentrations beyond both upper and lower detection limits leads to consideration of various imputation strategies and other approaches for handling censored data. The design and implementation of inter-laboratory comparisons of assay results is complicated by the analysis of multiple allergens in the dust from each sample. Multivariate statistical models carry additional issues including the handling of complex interaction terms. Other specialized analyses involve an examination of relationships between different groups of measurements (lead, allergens, endotoxin). An overview is given on considerations such as these that may arise in environmental sampling surveys, along with a highlighted discussion of approaches used to address specific issues of interest.
Learning Objectives: N/A
Keywords: Environmental Health, Methodology
Presenting author's disclosure statement:
Organization/institution whose products or services will be discussed: None
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.