The missing methods section: Origins of the dropout problem

A quarter century ago, the History of Education Quarterly accepted my first article, on when and why people in the United States began to use dropping out as the dominant term for people who left school without a high school diploma. Spoiler: we started using the term not because dropping out was a growing problem in the 1960s but the reverse. When graduation became expected for teenagers, we needed a term for those who violated the new norm.

Like many history articles and books, it had no formal Methods section, and if I had tried to write one, I suspect HEQ editor Bill Reese would have nixed it. So: in the spirit of tech publisher O’Reilly’s The Missing Manuals series, whose motto is “the book that should have been in the box,” I present to you the missing methods section for that article, with a postscript and a personal note.



To trace evolving use of terms in both professional and general circulation, I searched through 26 years of Education Index and The Reader’s Guide to Periodical Literature for items that had been listed under elimination from schoolstudent withdrawal, or dropout, the indexing terms used for relevant materials across 1945-1970, inclusive.1 The citation data for more than 650 indexed items and the original materials comprise the primary sources for this article.


The word dropout and its definition emerged from a set of alternative terms and explanations. The analysis focuses on that emergence — both the general pattern of usage as reflected in published, indexed trade and general literature, and also the explicit arguments made about dropping out over time.

Citations. From citation data, I aggregated raw counts by year of articles and other indexed materials that were about dropping out and, within that set, the subset of materials that used the word dropout or variations on it (e.g., to drop out). This is the work that underlies Figure 1.

Figure 1 from Dorn (1993)

The number of articles on high school dropouts, 1945-1970, as indexed by the Reader’s Guide for Periodical Literature and Education Index

Content analysis. Out of the total indexed items, I read the complete text of approximately half and for each, coded how authors described the act of leaving school, what they attributed it to, to what extent they thought it was an important issue, and any prescriptions they offered. This set of denotative codes — Description, Attribution, Salience, and Prescription (DASP) — allowed the identification of emerging, convergent motifs in the 1960s as well as the clear lack of consensus before the 1960s.

People, institutions, discourse. Knowing when dropout became a dominant term is not the same as understanding why. I put the changing usage into a broader context, where people and institutions frame discourse. There is both an empirical research literature on graduation patterns and a deep historiography (literature) on the changing role of American high schools between the late 19th century and when dropout became dominant.


The analysis presented here relies on typical historical readings of primary sources as well as the selective (and purposive) use of secondary literature. The use of raw citation counts is different from the corpus linguist’s use of either density of words (typically, words per million in a corpus) or percentage of volumes. Neither denominator by year is available for this analysis.2 It is possible that different sources would identify different timing for the emergence of dropping out.3

If the analysis presented here is narrower than a corpus approach, this analysis is also broader than either philology or discourse analysis, which trace the genealogy of usage or present intense, detailed scrutiny of specific sources, respectively. The denotative coding and analysis was designed to answer questions about patterns of usage over a portion of the twentieth century. A philological approach would reach further back in the roots of dropping out and its variants, and discourse analysis would be able to delve below the surface of the denotative analysis in this article.4


Today, I would use far better tools than I had available in the early 1990s: the computerized indices that are the successors to Reader’s Guide and Education Index, and the corpus-like collections of Google Books and HathiTrust. But there is one guiding secret in this brave new tech world: There is no perfect source for tracking language usage. It’s all judgment and craftwork.

What was critical then – and remains important now – is understanding that the emergence of terms important to education policy happens in a busy arena, with lots of competing agendas and the usual sloppy (and wonderful) way that people appropriate language. It is not enough to trace the usage of a single term. From both one’s substantive historical knowledge and as one reads primary sources, there needs to be a set of alternative terms against which the term of interest is compared. That is where one can truly see dropout emerge, not just in the number of articles and books discussing the act, but how the term dropout dominates the 1960s. The same is true when I think about the use of other terms in my research projects: public, community, accountability.

I started my career knowing that the growing use of dropout would tell a story about how we value and think about schooling, and how those changed in the twentieth century. The missing methods section could be replicated, as long as one understood this is an historian’s approach.

Neither philology, discourse analysis, nor corpus linguistics

What I did in the 1990s combined a somewhat-clunky way of tracing language use with the simple act of reading many sources and assuming that what the authors wrote explicitly was important to understanding the changing use of language. This was different from both philology and discourse analysis. My advisor had Raymond Williams’s Keywords somewhere in his bookshelves, but I was more interested in the larger significance of dropout‘s usage than the genealogical structure presented in Keywords. Two of my colleagues are researchers in discourse, including James Gee, and I enormously respect that work.

Today, I would still contrast the work of the historian with both philology and discourse analysis, and also with corpus linguists. If philology and discourse analysis are different types of micro-analysis, and corpus linguists try to scan the language at a 30,000-foot level, historians are more comfortable with the mid-range generalization and the use of multiple sources at less depth than discourse analysis. In my 1993 article I used 24 sources to illustrate five motifs in the emerging discourse of dropouts. I found those sources through the more extensive work with large collections of material (literature indices), but I needed to read many sources to understand the rough measures at a human scale.

People, institutions, discourse

What made my approach different is the context. I think of this as the relationships among people, institutions, and discourse. In that research on dropout I did not expect to draw all my insights from the changing use of words alone. I had to know what was happening with high school graduation at the time — it was becoming a clear majority experience for teenagers — and also how the role of high schools was changing after World War II.

More generally, changing word use is a clue to what may be happening to the role of social institutions, but it is never enough by itself, at least in my historian’s view. Changing word use is the starting point, but it always requires a broader context: what are people doing in counterpoint to the language? what is on the agenda for the related institutions? In this way, I knew at the outset that the story of the word dropout would tell us a great deal about how we value and think about high schools, and how those changed in the twentieth century.

I think it’s time to move beyond telling stories about single words or terms. What story can we tell if we broaden our scope?

Personal note

Historical work is an example of the invisible college of academic apprenticeship: so much of the craftwork is left implicit, while theory and historiography can be debated endlessly. The social practice of completing history dissertations is part of that hidden work. I never defended a formal proposal before starting work on my dissertation on the history of dropping out; as a history student at the University of Pennsylvania, I was required neither to present nor defend a proposal. My advisor Michael Katz wanted a short prospectus, and I later turned that prospectus into a fellowship proposal, but both focused on historical questions about dropping out, as well as the critical question many historians asked themselves: could I find material in archives? I never wrote about the methods I was going to use, and none of us were trained to write about methods. Could we critique Hayden White? Sure! Explain how we were going to sift through and use archival materials? Hmmmn…

When I graduated, exhausted, I would have told you that I read through several dozen volumes each of the Education Index and Reader’s Guide to Periodical Literature, taking copious notes listing relevant materials, from which I read several hundred articles . And again took copious notes on how authors wrote about dropping out–and on occasion typed in direct quotations when they seemed especially notable. From those notes, I struggled to understand what was changing and why — so I read secondary sources to add context while I wrestled with various explanations.

Those were the tasks; those were not the method.

If you enjoyed this post, please consider subscribing to the RSS feed to have future articles delivered to your feed reader.


  1. I also searched in the years before 1945, but that was not material used in the article. []
  2. In 1990-1991, I had only paper copies of the indices to work from. []
  3. Today, the general timing looks only a little different with a corpus built from different sources (books). []
  4. In the related book and a 1999 article coauthored with Erv Johanningmeier, I was later able to trace the earlier history of to drop outdropping out, and related usage. []