Analysis and data generation took place in parallel whereby further interviews were sought to test emerging patterns which were modified using constant comparison, 'cycling' between sets of data, the developing analysis and further sampling and interviews.
The inspectors were able to reach the site and collect samples and interview witnesses.
In 1993, an additional cohort (the Asset and Health Dynamics among the Oldest Old, or AHEAD, cohort), born between 1890 1923, was sampled and interviewed.
These items include participation rate, mean time between diagnosis and interview, reporting of sample size calculation, breast cancer assessment method, interviewer blinding, reporting of missing data, and quality of exposure definition which was addressed based on previous research regarding breastfeeding difficulties.
To address the discrepancy in years of experience between the surveyed and interviewed samples, an analysis of the 42 surveyed medics with at least two full years of field experience was done, yielding similar results to the full sample with a GHQ-12 mean of 2.2 (SD: 2.2) and PCL-C mean of 36.5 (SD: 10.1).
Approximately 6800 households were selected using a two-stage sampling process [10] and interviewed between January and December 2011.
random route sampling and mixed frame telephone interviews, are sufficient.
Sample the product in the store, and interview other potential consumers or users of the product.
Age, sex and type of cancer did not differ between the interview sample and the sample who declined to participate.
Sample weights were developed to correct for sampling bias and differences between eligible patients screened and interviewed.
