In lieu of an abstract, here is a brief excerpt of the content:

221 Appendix A: Method for the Map Survey1 .................................................................. To obtain unbiased results with a known margin of error from a quantitative study, a sample should be randomly selected from the entire population, with every unit in the population having the same chance of being drawn. Hence, to sample the fantasy genre, one would compile a list of all the works of fantasy ever written and randomly pick as many works as are necessary to obtain the desired margin of error. This is simple in theory but impossible in practice. Even if a straightforward definition existed that would determine unequivocally whether a given text belonged to the genre or not, it would be impractical to check everything ever published in English (or even everything published in English after 1858 or 1954, or any other arbitrary start date for the genre) against that definition. If one accepts Brian Attebery’s suggestion that fantasy is a fuzzy set, where genre affiliation depends on similarity to a number of core works,2 listing the works of a genre becomes impossible even in theory, as works belong to the genre to some degree. A question that needs to be resolved when sampling a genre is what actually comprises the population.To address this question with some precision, I have borrowed terminology from the International Federation of Library Associations and Institutions . According to its Functional Requirements for Bibliographic Records, a work is “a distinct intellectual or artistic creation,” an abstract entity that is realized through one or more expressions. A manifestation is “the physical embodiment of an expression ” and can exist in a number of items.3 For example, as J. R. R. Tolkien made constant revisions of the various impressions and editions of The Lord of the Rings, this work exists in a number of expressions. Each such expression is physically embodied in manifestations such as, for instance, the 2002 HarperCollins hardcover edition or the Houghton Mifflin 2004 Fiftieth Anniversary edition, each of which exists in a number of copies (items). Using these terms, we may ask whether a genre is a group of authors who write in the same vein or a collection of works that fulfill a certain set of criteria, or perhaps the totality of all expressions, manifestations, or even items that embody these works. If it is the authors, each author should appear only once in the sampling frame, and the probability for selecting Hope Mirrlees should be the same as for Terry Pratchett. That a writer who has published a single novel of genre interest would have the same impact on the genre as someone who has been writing steadily for decades seems dubious. Also, since we are investigating maps, it is reasonable to see the genre as a collection of fantasy works; the presence of maps can vary between expressions, but as a rule, if a work includes a map, a map is found in all of its expressions (although there are exceptions). I therefore decided to consider the population (the genre) to comprise all fantasy works, and to give each work in the sampling frame an equal chance of being selected for the sample. Since listing every work in a genre is impossible, sampling the genre directly is likewise impossible. A popular method for a literary survey is to use an easily avail- 222 aPPENdIx a able sample. (Deirdre F. Baker, for instance, describes her sampling thus: “I did a casual survey of the maps in some of the many fantasies I have on my own bookshelves .”4) Such a convenience sample, however, is almost certainly biased in one or more ways (in Baker’s case, both by what books she has in her bookcase and by which of these she picked for her survey); there is no way of estimating the representativeness of the sample.5 The survey in chapter 2 draws its sample from a sampling frame, whose possible biases are detailed shortly. The largest database of fantasy books available to me was the inventory list of SF-Bokhandeln (Sweden’s main science-fiction and fantasy retailer), which contained titles currently or previously in stock, or ordered, at the time when the list was copied to me.6 I filtered the list for fantasy novels in English (according to the retailer’s classification, it should be noted) and edited it: each title (work) was retained only once (separate editions, for instance hardback and paperback, resulted in multiple listings for some titles in the original list). Works...

Share