This is the "SHORT" description of what I did in my dissertation. (Reminder: diss was, like, 600 pages long. I'm competing with the fantasty people here!)

Before you start, let me apologize for the academic death prose. I don't always use it, but sometimes they make you. I don't like it any more than y'all. Here's the thing to remember: I'm studying the crucial question: Why do books sell? If we can answer that, it's worth a little death prose wading, don't you think?
____________________________________
Abstract
In order to determine the extent of book marketing on book sales and the principles of resonance that may be linked to sales beyond that predicted by marketing, a demonstrably random sample of 192 books from the population of Young Adult (YA) books available from 2003-2008 in the national U.S. market were scored with estimations of marketing influence and total sales.

Three data analysis methods were used: first, a consideration of simple linear regression and variable correlation, termed the ‘linear’ method; second and third, the ‘difference’ and ‘quadrant’ methods, in which books with sales beyond what was predicted by their marketing scores were compared against books with sales below what was predicted by their marketing scores. In the ‘difference’ and ‘quadrant’ methods (the differentiation between the two methods will be discussed in detail later), statistically significant differences between the ‘beyond’ and ‘below’ sets of books were established as possible points of resonance or dissonance—i.e., key aspects of audience appeal or distaste. These aspects were then compared to the initial results from the ‘linear’ method and probable patterns of resonance were established. These patterns had important implications pertaining to issues of race, gender, and the belief-systems particular to teenagers. The nine main patterns established were: patterns of meaning/sophistication, patterns of emotion, patterns of estrangement, self-importance, gender, race, socioeconomic status, divine potential and moral identity.

Three auxiliary studies were conducted first to identify the extent of cover-likeability as it pertained to sales, next to establish that young adults (and not their parents or other adults) accounted for the majority of resonance points observed, and finally to estimate the overall effectiveness of self-promotion among authors. Results indicated that cover likeability may increase sales by up to 14.5%, that young adults can reasonably be assumed to be the primary readers of texts, and that author self-promotion is not significantly connected to increased sales.

As a final exercise, the list of possible principles of resonance was used to evaluate the publishing industry’s effectiveness at assigning marketing/promotion. Numbers indicate that publisher book marketing is disconnected with principles connected to sales somewhere between 3% and 55% of the time and a more accurate estimate may not be possible without another study.

Outcome Variables: Marketing and Sales Scores
In order to be able to factor out the effect of marketing/promotion on a book’s sales, each book was given a marketing score and a sales score. The marketing score was calculated based upon the following criteria: 1) the actual price paid to the author for initial rights; 2) how many additional books the publisher committed to publish in the initial deal; 3) whether or not there was a bidding war among publishers that either ended in an auction or was circumvented through a pre-empt; 4) the relative fame and/or reputation of the author; 5) whether or not School Library Journal or Booklist (two of the most influential reviewers of YA fiction) gave a starred review; 6) Whether or not one of 50 major review sources gave the book a full-length positive review; and 7) whether or not there was potential for ‘carryover,’ e.g. books about vampires published shortly after the success of Twilight.

In order to make an estimate of sales, the following three variables were considered: 1) The book’s Amazon ranking compared to how many standard deviations from the mean for the book’s year of publication ; 2) The book’s BN.com rankings ; 3) The number of weeks the book spent on each of thirteen bestseller lists tracked by the PM database.

When each book had a marketing and a sales score, the Pearson’s Relationship Coefficient of the marketing-sales relationship was calculated to be 0.6, indicating a moderately strong linear relationship between marketing and sales. Linear regression provides the formula:

S = 0.63M +2.58

By comparing residuals, assembling quadrants of sales vs. marketing, and considering differences between sales and marketing scores, the data were assembled into three basic categories: books that sold better than their marketing predicted, books that sold worse than their marketing predicted, and books that sold as their marketing predicted.

Independent Variables
Each of the 192 books was analyzed based on 150+ questions pertaining to plot, literary merit, morality, race, gender, and other categories of interest to YA authors. The aim was to identify book traits that may result in greater audience resonance/dissonance—specifically looking for important patterns of appeal and distaste.


Auxiliary Study One: Cover Likeability
149 students from a California high school (chosen because its demographics were analogous to the U.S. population at large) were shown pictures of each of the book covers from the sample. They were asked to quickly score each picture on a scale of 1-5, with lower scores corresponding to distaste. The average cover-likeability scores of books that performed better than their marketing predicted were compared to books that performed worse than their marketing predicted. Books that sold better than their marketing predicted had a cover-likeability score 11% (+/- 5%) higher than books that sold worse than their marketing predicted. This difference seems to be largely driven by the opinions of girls. When split into gender groups, the girls’ cover-likeability scores for books that sold better than predicted were 14.5% (+/- 6.5%) higher than books that sold worse than predicted. The boys’ responses showed no statistically significant difference between the two groups of books. The data indicates that a likeable book cover seems to help sales, but only if it appeals to girls. Considering that the vast majority of the YA fiction audience is female, this is expected.


Auxiliary Study Two: Library Data
In order to confirm that young adults were the ones primarily responsible for the purchase of YA books and, thus, that any data about trait resonance corresponded with young adult (as opposed to adult/parental) taste, 37 libraries from around the country were sampled. From each library the following was recorded: the number of each of the 192 books that the library stocked and the number of copies of those books checked out. The reasoning was that librarians (adults) were responsible for the stocking of books, and that books checked out were more likely to be checked out by young adults, themselves, than by their parents or other adults. The results showed a close correlation between all relevant groups. The number of books stocked mirrored the number of books checked out and both of these mirrored the number of books sold, indicating that divergent audiences of YA lit (teens, parents, librarians) either do not appear to have fundamentally different tastes or that all groups defer to the taste of the young adult reader.
____________________________________
Overwhelmed much?

Tomorrow: I will talk like a NORMAL PERSON again. What should we talk about? I think marketing. Why? Because it is more important than anyone will ever tell you. For all practical purposes, it's the ONLY thing between your book and a bestselling book, accounting for as much as 80% of a book's success or failure.

To put it another way: you know how Donald Maas, in his book, Writing the Breakout Novel, says that there is a disconnect between marketing and sales? That the real reason books don't sell is their quality?

He is wrong.

I have equations to prove it. This one is my favorite: