A Facebook comment today by Alexander Russo reminded me that this fall is the 20th anniversary of Origins of the “Dropout Problem” ($ or JSTOR’s Register and Read program), published by the History of Education Quarterly. The article came out of the part of my dissertation that asked when, how, and why “dropping out” became the dominant term for those who leave high school without a diploma. My argument then (and now) is that while the term “dropout” appeared decades before it became dominant, the primary use of the term “dropout” reflected a new expectation in the middle of the 20th century, the norm that teenagers complete high school. I wince a bit at the phrasing in spots — how could I have used so many parenthetical phrases?? — but for the most part, I am still happy with it and very grateful to Bill Reese for midwifing my first article. I forget whether I revised the manuscript twice or three times, but he saw something valuable in the first manuscript and wrote the kindest, most encouraging rejection note. I would like to think he was right, and not just in my biased eyes: the article is the most cited piece in that volume of HEQ.
One part of that research troubled me at the time, the documentation I accumulated that “dropout” was in fact becoming the dominant term in the 1960s. Here is the original figure from the 1993 article:
I mentioned the limitations in the usual new-scholars’ caveat: by checking the emergence of a term by periodical indices, I was relying on the raw counts of articles rather than the proportion of pieces published in a year that included the term “dropout” or something similar. There was also the fact that at the time, I went through several decades of paper indices, each year’s volume hundreds of pages long. Was I making some tremendous error in my work?
Today, we can pull down the same data with a database, or look at a different body of work, in this case books. Google Books Ngrams is problematic in several regards, but not in ways that are that different from the limitations of Education Index or Reader’s Guide to Periodical Literature. And Ngrams displays the data in terms of the proportions of all books published in a year (in this case, in American English). I used a few different terms, but the results are similar, with one important difference:
You see the same hump, except shifted by a little more than half a decade, and the decline also shifts, by more than a decade. Some part of this is explainable by the fact that this data comes from books, which have a different lead time. I am amused by the percentage numbers on the left: the first peak is when about 3 out of every 10 million words in 1971 books used the term “dropout rate.” Whooeee! But since there are many books published, each with between 60,000 and 200,000 words, and very few discuss dropping out, we should expect any term to be rare, even if it was at the peak of use in a decade.
The basic story remains unchanged in terms of the timing of the word dropout: the usage climbed in the 1960s, with articles first and books later. I would be very curious to see the data from different databases, if any graduate student is interested in trying to replicate this: it shouldn’t take more than a few hours with any database.