Ample Sample? An Examination of the Representativeness of Themes Between Sampling Durations Generated From Keyword Searches for 12 Months of Immigration News From LexisNexis and Newspaper Websites

Abstract

Recommendations vary on sample sizes researchers should use to represent a year’s worth of coverage from electronic databases. We use a keyword search to create two sampling frames of immigration news in 2009. We use LexisNexis to represent elite and regional news outlets and a keyword search of individual Arizona newspapers to represent state coverage. We examine the representativeness of themes, created and coded with IBM’s Modeler program. We compare the efficiency of samples created with simple random sampling and constructed week sampling and conclude most variables are representative when sampling between 10 and 14 weeks.