Method
This research project consisted of two parts, a short survey and follow-up interviews with a select group of subjects.22 The survey consisted of fourteen items that were evenly split between recreational and educational items, and three additional items related to gender, type of reader, and preferred genre of reading material, as well as an open-ended comments item. Surveys were distributed at two medium-sized public libraries, each with a diverse patron base, serving towns of approximately 37,000 and 65,000 people respectively, as well as the local university community. Both are located in the Midwest, more than one hundred miles from any major city. Surveys, along with recruiting posters and consent letters, were placed at the circulation and reference desks, with envelopes for completed surveys kept by staff behind the desks. Patrons were encouraged to fill out and return surveys at the service desks. Sixty-two surveys were completed and returned over a ten-day period.
The survey items were developed by the author. Ideas for some of the items were drawn from Ross’s study, “Finding without Seeking,” using the five categories created from her interview data. After substantial revision, the items were initially validated in an informal pilot test. A draft of the survey was distributed to a group of LIS students who were also leisure readers, and they filled out the survey in the presence of the researcher. Comments and feedback were encouraged and oral and written feedback on many of the items was received. This feedback was then used for a final revision of both the survey items and design. This final version of the survey was then distributed to the public libraries as described previously.
After surveys were completed and collected from the libraries, survey results and items were validated using a statistical technique known as factor analysis. Factor analysis is a data reduction technique that can also be used for validating survey instruments. As part of the validation, factor analysis can be used to determine how many factors are present in an instrument.23 In this case it was used to determine whether or not the data had just the two factors of educational and recreational (reading), and eliminated the possibility of a third, unknown factor. Factor analysis also can be used to determine how well each item “loads” onto each factor. In this case it was used to determine whether the items that were intended to be part of the educational or recreational factor were actually a part of the intended factor. Within factor analysis there are two methods for interpreting the results, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). CFA was used because “CFA is typically a more useful way of testing whether a given test’s patterns of association with other variables correspond to what is expected. In CFA hypotheses are investigated by imposing restraints on certain factors so as to define more precisely the expected nature of the association between variables.”24 In this study it was more beneficial to have a better sense of the exact relationship between the variables of recreation and education based on the hypothesis of a definite relationship between the two.
The last item on the survey asked subjects if they were willing to participate further in the research project. More than fifteen respondents volunteered to do so. Subjects initially were selected by their ability to be contacted. E-mails were sent to all who left legible e-mail addresses, asking if they still wished to participate. Several left e-mail addresses that were illegible or were no longer working, and many never replied to the initial e-mail. Eight responded by e-mail, and interviews were scheduled with those who were able to participate within the time frame. Some of the respondents were unable to participate during this time period, and thus could not schedule meeting times. Six subjects agreed to be interviewed, but only four completed the interview process. Subjects who did not leave an e-mail address were contacted by phone, and three subjects were successfully contacted. All three agreed to be interviewed and completed the interview process.
Altogether seven subjects completed the interview process. Interviews were conducted at the public libraries, places that were both familiar and comfortable for the participants and author. Interviews took approximately thirty minutes and focused on the responses to the survey questions, further exploration of the educational and recreational outcomes, and the subjects’ individual reading experiences. All interviews were taped and transcribed. Transcripts were coded for educational and recreational factors, as well as for genres and other reading themes, such as whether the interviewee came from a family of readers, and whether they had been reading since childhood.
Survey Results
More than two hundred surveys were available at the reference and circulation desks of the two participating libraries for ten days in April of 2004. At the end of the collection period, sixty-two surveys had been completed and returned. Table 1 includes the text of items one through fourteen from the survey and a breakdown of the response data by item.25
Factor Analysis of Items 1-14
The author and a consultant worked together to code the data into a computer database, which in turn permitted analysis to be performed by means of a computer program. To confirm the supposition that these items measured a general factor, CFA was performed on items one through fourteen of the survey instrument (for a listing of all survey items and responses, see table 1) by means of the CONFA computer program.26 Subsequent CFAs were conducted to determine if a better fit might be achieved with a two-factor model. In both cases optimal model fit was achieved by dropping items with low factor loadings. Model fit was calculated for both the single- and two-factor analyses using Tanaka’s goodness of fit index (GFI) as calculated by the CONFA program. Reliability of the factors in each of the models was assessed by means of McDonald’s omega.27
Confirmatory factor analysis of items 2, 3, 4, 7, 8, 10, 11, 12, and 14 yields a single factor model with a good fit (GFI .97) and high reliability (McDonald’s omega = .91). Confirmatory factor analysis of the same nine items according to a two-factor model also yields a good fit (GFI .97). Reliability of factor 1, consisting of items 2, 4, 8, 10, 12, and 14, is good (McDonald’s omega = .89). Factor two, consisting of items 3, 7, and 11, also has high reliability (McDonald’s omega = .94). Conceptually, the two-factor model makes sense. Items 2, 4, 8, 10, 12, and 14 can be understood to be asking respondents about their perceptions of educational outcomes from leisure reading, while items 3, 7, and 11 can be thought of as asking respondents about recreational outcomes from leisure reading.
The substantial correlation between the two factors (r = .51) is a plausible value for the relationship between the two factors and lends additional support to the validity of the two-factor model. This result suggests that respondents perceive a relationship between educational and recreational outcomes of leisure reading. This was interpreted to mean that readers value both educational and recreational aspects of leisure reading. However, the correlation is not perfect; the leisure reading experience is about more than just education. This supports the hypothesis that while there is a relationship between education and recreation, readers are not choosing to read solely for the purpose of learning while reading, but that learning is often an unexpected benefit of leisure, albeit one that can be very important to the reader.
First steps toward developing an instrument to measure educational and recreational outcomes of leisure reading by adult public library patrons have been moderately successful. Its goodness of fit for the single- and two-factor models and its reliability measures are reasonable for a nine-item measure that is still under development. Clearly it would be logical to improve the instrument by writing additional items, a task that is now made easier given the knowledge gained from these initial analyses.
Results from Individual Items
Results from the individual survey items were also calculated.28 One of the most interesting results is the high percentage (87 percent) of respondents who agreed with the statement, “fiction reading serves as an escape.” Because of this high number, it can also be said that even the people who see fiction as an escape also feel that they learn from fiction reading, based on the responses to item 2, with 81 percent agreeing that they learned a lot by reading fiction.
In regards to specific types of learning, 77 percent of respondents felt that they had a better understanding of other countries and cultures. This was validated in the interviews where learning about another time or place was the most commonly mentioned educational outcome. In “Finding without Seeking,” Ross’s results emphasized learning about other people and self-growth through reading. Interestingly, in the survey responses, only 50 percent of respondents felt that reading better prepared them to understand and solve problems, although 71 percent felt that reading helped them better understand their world. This was also a theme of the interviews (see the section on interview results for a further discussion of this type of educational outcome). This difference can most likely be attributed to the sample size and the fact that Ross’s readers were selected by interviewers and were not necessarily public library patrons, as was the case for all the respondents in this survey. If this survey were to be repeated it would be important to look further at these specific types of learning, with more items addressing how leisure reading helps readers better understand others and gain new insights and perspectives, as well as how readers feel they grow and change through fiction reading, as these areas were emphasized by Ross’s respondents and were also important to the subjects of this study.
Genres
Among the readers surveyed, the most popular genres were literary, historical fiction, mystery, and spy/thriller/adventure, all ranked as well liked by more than 60 percent of respondents (see table 2).29 Religious and inspirational fiction was the least popular, ranking as well liked by less than 12 percent and disliked by more than 60 percent of respondents. Romance was a close second, with 59 percent disliking romance, although 23 percent of respondents gave romance a positive ranking. Due to the small sample size and the limitation of surveying only two libraries, it is likely that these results say more about the likes and dislikes of two specific communities. In this case, both communities have higher-than-average education levels due to the presence of a large local university, with both libraries serving members of the university community. Anecdotal evidence also suggests mysteries and thrillers as the preferred genre of academics, and mysteries are one of the most popular collections in the participating libraries. Literary fiction scored remarkably high (67 percent), well above more common popular fiction genres such as romance, fantasy, and science fiction, and this is also likely due to the university community or the small sample size. The popularity of literary fiction also may be due to the self-report nature of the survey with respondents feeling like they should report that they liked literary fiction.
Within the interviews, historical fiction was most commonly mentioned, closely followed by mysteries. This correlates well with the survey data, as mysteries were also very popular with survey respondents (see table 2). The high number of responses for historical fiction likely is related to the fact that the most common educational outcome mentioned by interview subjects was learning about another time or place, both of which are common parts of historical fiction. Historical fiction is also often thought of as part of other genres, as mysteries, thrillers, inspirational, and romances can be set historically. Therefore, while patrons may not have given historical a very high ranking as a genre of its own, based on the titles mentioned in the interviews, it can be deduced that historical fiction is fairly popular with patrons, though most prefer it mixed in with other genres such as romance or inspirational. A current trend in popular fiction is books that cross genres, or integrate more than one genre in a single work. A good example is the popular Outlander series by Diana Gabaldon, which integrates romance, historical fiction, and the fantasy element of time travel.
Gender
Twenty-seven percent of survey respondents were male and 73 percent were female. This is not a surprising breakdown, with evidence suggesting that women are more likely to read fiction and more likely to pursue leisure reading as a hobby.30 The fewer number of male respondents was also reflected in the interviews, where only one interview was successfully completed with a male subject. When compared with genres, these interviews produce interesting results. Romance and religious fiction are often thought of as women’s stories, while science fiction, fantasy, and spy/thrillers as more likely to be read by men. Mysteries, historical fiction, and literary fiction generally are considered gender neutral, though none of these groupings are mutually exclusive or scientifically proven. Based on the results of the female survey respondents, it is interesting that romance and religious fiction were not very popular. But perhaps that can be explained by the high proportion that picked the gender-neutral historical fiction and mystery stories.
Frequency
Readers were also asked to identify themselves by type of reader: (1) heavy reader, reading more than three books a week; (2) frequent reader, reading one to three books a week; (3) moderate reader, reading two to four books a month; and (4) occasional reader, reading less than two books a month.31 This was one of the most obviously successful parts of the survey instrument, with a good distribution of types of readers: 28 percent considered themselves heavy readers, 33 percent frequent readers, 27 percent as moderate readers, and 12 percent as occasional readers. Since this survey was conducted at the public library, with the recruitment materials asking for readers to participate, and most of the surveys were completed at the reference desk, the range of readers is satisfying and is helpful in supporting the rest of the data. It shows that the results come from all types of readers, whereas Ross’s results come only from self-identified “heavy readers.”32