Abstract

The purpose of our study was to determine what correlation, if any, exists between mean page length of books and genre for five different genres of fiction. The five genres that were studied were Romance, Historical Fiction, Science Fiction, Mystery, and Fantasy. The population of interest in our study was the population of all books of these five fiction genres at Westlake Porter Public Library in Westlake, Ohio. The data was collected using five separate random samples, distributed according to genre. The data in the samples was directly controlled by publisher; as all data collected came from publishers who were located on the Eastern coast of the United States. Between 80 and 95 data points were collected for each genre from the books available at WPPL.

           

The data was analyzed using three different statistical tests. Initially, we decided to run a chi-square test for homogeneity of populations because our data came from five different populations. We grouped the data in a categorical fashion, tabulating the frequency of five different bins of page lengths for each sample. We then ran a standard chi-square test for homogeneity of populations. The test was inconclusive, however, because some of the expected cell counts were less than 5, negating a primary assumption of the chi-square test. In order to conclusively compare the populations, we decided to instead generate 95% confidence intervals for the true population mean of each of the five populations from the sample data available.

 

The results of the interval generation were sufficient to largely order the genres based upon page length. Our intervals did not overlap and therefore nicely drew distinctions between the sizes of the mean page lengths for the five genres. The science fiction and fantasy intervals, which were the largest, overlapped, followed by mystery, romance, and historical fiction. In order to try and draw further conclusions, we performed a two-sample t-test on the data from the Science Fiction and Fantasy populations. However, the results of that test were insufficient to clearly determine that one genre was larger than the other. Therefore, the hierarchy that our confidence interval study produced remains as the conclusion for relative page lengths of different genres at Westlake Porter Public Library.