PostEdit Home Help

Key Pages

Course Home |
Projects |
----------------- |
Project Home |
----------------- |
Final Report: |
Introduction |
Data Collection Protocol |
Protocol Implementation |
Data Analysis |
Discussion |
----------------- |
Work Space: |
Discussion Board |
Shared Files |
----------------- |

Changes [Dec 12/05]

Data Analysis
Discussion Board
Protocol Implementa...
Data Collection Pro...
   More Changes...
Changes [Dec 12/05]: Home, Discussion, Data Analysis, Discussion Board, ... MORE

Find Pages

Posted at Nov 27/2005 05:28 PM:
peter y: It should, perhaps, be unsurprising that the collected data does not suggest any profound revelation about the secret of J. K. Rowling’s literary success. There is no decisive evolution in Rowling’s use of parts of speech. On an intuitive level, we can feel the books become more “advanced” in their literary styling as Rowling’s audience grows up from grade schoolers to teenagers. Her language is a little more advanced, and her plotlines are grimmer. Still, this study has not been able to capture that perceived maturation. Perhaps, as suggested in the introduction, writing is too complex to yield itself to simple statistical analysis.

Still, it would have been practical to assume, based on the background information behind this study, that two trends would have appeared as Ms. Rowling grew into her series. Firstly, that her sentences would become more complex and lengthy as she moved from writing to 9 year-olds toward writing for teenagers (and, increasingly, adults); and secondly, that her use of adverbs—those words that show so much youthful insecurity—would decline as she wrote more and could become more confident in her success.

Neither hypothesis manifested itself. Indeed, the average sentence length in each book showed no clear direction, with book 6 being far less wordy than any of its predecessors. With such a scatterplot as the one for sentence length, it is worthless to try to extrapolate the number of words per sentence in book 7, despite our ability to make the calculation from the regression line we plotted. The fifteen words-per-sentence estimate could be completely off, since there is no reason to believe the data have a true linear orientation.

The second assumption, that Ms. Rowling’s use of adverbs would decline, is also challenged by the data. In fact, instead of declining, the proportion of adverbs in her books rose steadily from book 2 onward. This is interesting, and it will be fruitful to see if our extrapolated estimates for adverb use in book 7 turn out to be accurate. If the new book’s proportion of adverbs is nearer to 10.22% (the trend line with book 1 excluded) than to the 9.27% estimate based on all six books, then we will have to seriously consider whether the first book is a model outlier, and if so, what that means.

In any event, the rising rate of adverb use in Rowling’s novels is surprising. It might suggest that, rather than being a stage of immaturity in a writer’s growth process, the use of adverbs is a stylistic device that Rowling turns to with increasing frequency.

Now, it is true that the type of adverbs that Mr. King is focused on in J. K. Rowling’s writing are a small proportion of the total set of adverbs used in English. He is specifically concerned with attributive adverbs that describe how characters speak in dialogue (he said forcefully, she said happily). Still, he characterizes this happening “8 to 10 times a page,” which would clearly be a significant amount. The data, however, do not support this.

If incidental adverbs, which make up the majority of adverb cases and are the kind no writer can avoid, were the only ones that Mr. King used normally in his writing, while Rowling was excessive in her use of the part of speech, then there should be a notable difference in the proportion of adverbs in each author’s work. Yet, when we use graphs to compare the two, we do not see such a difference. The aggregate proportion of adverbs used in all six Harry Potter books nearly identical to (in fact, slightly lower than) the sample from King’s Dreamcatcher. On top of that, book five in the series, the one King was reviewing for Entertainment Weekly when he made his comments about Rowling’s use of adverbs, was also within two hundredths of a percentage point of his own sample proportion (0.0816 to Kings’ 0.0814). 8 to 10 extra adverbs a page would certainly show up under these conditions. That would mean 3 to 4 per sample page, or 30 to 40 more adverbs per book, which would be enough to raise the proportion of adverbs 2 to 3 percent. In no book does Rowling’s proportion even go higher than 9.5% percent, 1.5% higher than King’s.

All this adds up to suggest that King perhaps does not practice what he preaches. This is the most interesting result yielded by the study. For all of his very public blustering about Rowling’s nervousness as a writer, King’s work has not been shown by the data to be any different.

It would seem, in fact, that the study best supports a third writer, the famous literary critic Harold Bloom. In an essay in the op-ed section of the Boston Globe criticizing King’s selection for a lifetime achievement award, the long-time Harry Potter critic wrote:

“Later I read a lavish, loving review of Harry Potter by the same Stephen King. He wrote something to the effect of, "If these kids are reading Harry Potter at 11 or 12, then when they get older they will go on to read Stephen King." And he was quite right. He was not being ironic. When you read "Harry Potter" you are, in fact, trained to read Stephen King.” (Bloom, 2003)

Lastly, as was noted briefly in the data analysis section, this study suffered from the inability to test for the statistical significance of its results. Note, for example, that the six data points used in the scattergraph of Harry Potter adverb proportions had a spread of only 3.2%, with a high of 9.5% and a low of 6.3%. 3.2% is a very small margin, can we be sure that it is really the result of actual change instead of sampling variability? The answer is no, we cannot be reasonably sure. The sample size was made large to deal with this exact issue, out of a desire to reduce the effects of random variation. Still, if the study could presume independence, then the margin of error yielded from the data at 95% confidence would be ±1.5% — a spread of 3% in total. That would obviously include almost all of the data in the interval.

(Assume a sample size of 1200 for each book, and set the p-hat proportion to .08 from the aggregate score. Thus ME = 1.96√(.08)(.92)/1200 = 0.0153)

What’s more, because we discussed how the study should not assume independence in its data, we can’t even make that assessment. We are left somewhat in the dark about the meaning and validity of our results. This unfortunate truth undercuts much of what the study seeks to assert. The only assertion we can confidently make is that the data does not support the hypothesis, proclaimed by Mr. King himself, that Rowling uses more adverbs than he does.

It is common at the end of discussions to suggest that further studies would benefit from larger sample sizes, but in truth, in this case the sampling size was likely more than sufficient – to the point where collecting data was quite taxing and time consuming. Instead, we would suggest that promising next steps would be studying a wider range of other books, to serve as a sort of control against which to measure Rowling. Perhaps the study of more Stephen King books would yield more interesting results. We could track the evolution of both writers to see if any differences emerge. Another interesting aspect of writing would be to examine more authors from both Great Britain and America (Rowling is British), to see if nationality plays into the way an author writes. This way, if differences emerged between two authors like King and Rowling, we would be better able to isolate the effect of nationality as a variable.

What the study really needs, however, is a reliable way to test the significance of the data. The techniques to do this, unfortunately, are still beyond the grasp of this statistics course.


Bloom, Harold. “Dumbing down American readers.” The Boston Globe, Sept. 24, 2003.

King, Stephen. “Harry Potter and the Order of the Phoenix, Book Review.” Entertainment Weekly, July 2, 2003.,6115,462861_5||251753|0_0_,00.html

King, Stephen. Dreamcatcher. Simon & Schuster, New York, 2001.

Rowling, J. K. Harry Potter and the Sorcerer’s Stone. Scholastic, New York, 1998.

Harry Potter and the Chamber of Secrets. Scholastic, New York, 1999

Harry Potter and the Prisoner of Azkaban. Scholastic, New York, 1999

Harry Potter and the Goblet of Fire. Scholastic, New York, 2000.

Harry Potter and the Order of the Phoenix. Scholastic, New York, 2003.

Harry Potter and the Half-Blood Prince. Scholastic, New York, 2005.

HP discussion[1].doc

New Page - Edit this Page - Attach File - Add Image - References - Print
Page last modified by Peter Y Tue Nov 29/2005 00:02
Post your comments:
Site Home > STAT W1111: Statistics Projects > Harry Potter and the Adverb of Doom > Discussion