Secondary analysis of archived data (aka: secondary data analysis) is a commonly used quantitative research design. What is "secondary" is the new analysis of the previously collected primary data taken from an existing database.

a) If one conducts such a study using a CDC database of data collected from 20,000 participants who were originally randomly selected by the CDC for their original study, and he samples 2500 cases from those 20,000 which contain the variables of interest for his study, is this sampling approach considered random or non-random convenience sampling? How so?

b) To what extent will one be able to generalize the study findings (external validity) from this sampling procedure?

Support arguments with at least 3 references.