Before the Numbers
Quantitative Research Design from Question to Dataset
Before the Numbers
Quantitative Research Design from Question to Dataset
A statistic is the last step of a long process. Someone chose a question, defined a term, picked a group of people, wrote the wording, and cleaned the file. Each of those choices can change the number. This book covers those choices, in the order a researcher makes them.
The book assumes no background in research. An introduction explains how research works, and eight chapters follow. Each chapter works through one public case. Seven of the eight cases use real public data. One uses a simulated file, and the chapter says so.
Vivek H. Patil, Ph.D.
Margin of Error Media LLC
The Eight Cases
1. Counting homelessness
Two federal counts of the same problem give different totals. The chapter shows how a definition becomes a measure.
2. Four estimates of one population
Four research groups estimate the unauthorized immigrant population. Their totals differ by millions. The chapter traces where survey error comes from.
3. The resume audit
Researchers mailed 4,870 resumes with randomly assigned names. The chapter covers experiments and what they allow a researcher to claim.
4. Measuring food insecurity
The federal measure rests on a short set of questions. The chapter covers scales, reliability, and validity.
5. A community health survey
A county borrows tested questions from a national health survey. The chapter covers how to write and test questions.
6. Who the census missed
The 2020 Census was accurate in total and inaccurate for several groups. The chapter covers sampling and coverage.
7. Bots in an online panel
A simulated panel file has 120 planted bad cases. The chapter covers data collection, screening, and cleaning.
8. Reporting a wage gap
One file of 534 workers supports several different gap estimates. The chapter covers the analysis plan and the report.