To explain the process of preparing a technical paper, it is best to first explain the basics of the questions, worksheets, and formulas upon which it is based. The research question being examined focused on broad, frequently studied topics such as total population and labor force participation at the county level. Also, reproducibility is always important when processing data. For all these reasons, I structured the spreadsheet to use formulas whenever possible instead of copy-pasted strings. This meant that any update to the primary data would instantly propagate out across the document and automatically update the final results. Although my focus was on Los Angeles County, data for any other county could be swapped in without much of a challenge. In this way, the worksheets served as a block of programmatic statements that ran off of starting data imported from the Census website, akin to arguments in a function; the spreadsheet document was, in essence, a calculator.
In my opinion, the technical paper reflected the underlying data approaches, and various aspects of the research question became a sort of logic flow:
If a certain result is observed, then a corresponding statement may be made. However, if that result is not observed, then another statement may be made. With these basic steps, much of the technical paper was laid out neatly, concisely, and reproducibly. Its primary value came from conducting the underlying work, as well as providing a digestible and not-entirely-technical answer to the question.
Another aspect of the technical paper was going beyond the
logic flow approach in order to understand and seek out deeper explanations, when necessary. This relied on an understanding of the local context and the eras being considered. Just as importantly, it was necessary to reflect carefully on whether there had been any disruptive events from 2000 to 2010 that may have further explained (or negated) any underlying assumptions, such as the Great Recession. In addition to the primary value of the report, I felt that this was essential to creating the sort of deliverable that I would feel confident providing in my career.
Feedback from earlier assignments was key: a tendency to leave off citations had been raised, as was an occasional weakness when it came to fully applying key demographic principles. By doing more thorough citations, a deeper literature review and research process was necessary, which ultimately strengthened and facilitated the overall development of the assignment. In addition, returning to
the basics and applying fundamental principles about population age structure and other core concepts helped to harden the research-based nature of the report. Leaning into major concepts that have been deeply investigated for decades was a straightforward and meaningful way to enhance the authoritativeness of the report without turning to unlikely solutions like demanding additional data. In the field of Applied Demography, where resource limitations often restrict the researcher, being able to deftly apply well-accepted principles is particularly essential. In other words, the Applied Demographer oftentimes has to
percolate up and fit limited, incomplete data into an overall framework which does much of the heavy lifting, rather than
percolate down and exercise a framework through years or decades of expensive data collection that is authoritative in and f itself.
The main challenge along the way was the changing Census website. While a subdomain like data.census.gov would suggest a central portal for accessing data, much of the Census’ information remains scattered across various posts and other URLs. This meant that research had to rely to some degree on a search engine, doing site:census.gov searches for some information. I considered this a weak point in the research process, and a challenge because it was inelegant and required some improvisation. It was baffling why data from the 2000 decennial census was not also available from the 2010 decennial census, all in the same place; for 2010, information was gathered from a survey published on a standalone page of the Census website. Another issue was that different rows of data did not perfectly match up, such as for age groups, which meant that revised age groups were not only useful but necessary in order to have comparable data columns from different years.