Educational authorities invest large sums of money, offer courses, organize meetings, facilitate exchanges, and design programs to promote reading competency at schools on the basis that they “should provide all children, regardless of their achievement levels, with as many reading experiences as possible” (Cunningham & Stanovich, 1998, p. 8). However, these initiatives are often ineffective because after the financial investment is made, the results are not properly analyzed in serious statistical evaluations so as to implement the appropriate political measures and educational investment strategies.

With a view to mitigating this shortcoming, this paper presents the reading habits of secondary school students in Extremadura (Spain). To do this, 4,392 surveys were administered at a total of 53 schools throughout the region. Reading habits were analyzed in terms of time spent reading, books purchased, preferred genres and other variables related to the use of social networks, library visits, bilingual education, how reading ranks among favorite hobbies, languages read, and reading habits and rates among students of foreign origin. All these variables were evaluated by respondents’ sex, age, level of schooling and geographic area.

By determining the geographical location of the surveys, it was possible to map some of these variables. This could provide the starting point for implementing active policies to help improve reading competency rates in certain areas.

Certain hypotheses or research issues were taken as the central focus for presenting the results, and these are also discussed in the conclusions. First of all, there may be differences between areas in terms of reading habits in secondary school. Secondly, it is hypothesized that women read more than men, and thirdly, in principle, the specific measures offered by educational authorities (Spanish/English bilingual schools and departments based on AICLE [Content and Language Integrated Learning]), REBEX (School Library Network of Extremadura) and reading programs are expected to be effective. Other hypotheses include higher reading rates among students enrolled in ESO (compulsory secondary education) and in Baccalaureate (upper secondary education) courses¹ than among those enrolled in vocational training, and differences between the reading habits of students of Spanish origin and those with immigrant backgrounds. Lastly, it is likely that other factors, such as library visits or perceptions, attitudes or beliefs surrounding reading as a leisure activity, have a positive impact on reading habits (Applegate & Applegate, 2004; Asselin, 2000; Benevides & Paterson, 2010; Daisey, 2009).

This aim and these hypotheses are reflected in the following research objectives: 1) to present the state of the art of previous statistical research on reading habits in Extremadura; 2) to describe the current reading habits of secondary education students in the Autonomous Community of Extremadura based on age, educational stage, school and sex (evaluating students’ book purchasing habits, library use habits and literary preferences); and 3) to compare reading habits in secondary education based on sex, geographic area, educational stage, whether or not the education is at a bilingual school, school participation in regional or national reading programs, and nationality of origin.

To put the study area into its sociodemographic context, Extremadura is a region of southwest Spain, bordering Portugal, with an area of 41,635 km² and a population of 1,079,720 as of January 1,2017. Due to its low per capita income, Extremadura is the only region in Spain to remain among the least developed in the EU for the 2014-2020 planning period, with a GDP per capita below 75% of the average in Europe (Nieto & Cárdenas, 2015), which has encouraged investment in education and research (Grecu et al., 2015; Mate & Harris, 2013). Indeed, a PISA report (OECD, 2018) shows Extremadura is one of the regions with the worst results in the country.

For years, rural populations have been shrinking and losing infrastructure. Towns with fewer than 10,000 inhabitants account for 96% of the municipalities in Extremadura (Figure 1), while 50% of inhabitants live in towns of more than 10,000 residents.

Figure 1

Stratified two-stage sampling was employed to enable subsequent mapping of the variables. The study area was divided into two parts: on the one hand, the seven cities with populations of more than 20,000 inhabitants and, on the other, the municipalities were grouped into twelve main zones. In these twelve rural zones, the municipalities were classed in intervals by population size (< 1,000 inhabitants; from 1,000 to 2,000; 2,000 to 5,000; 5,000 to 10,000; and 10,000 to 20,000). Table I presents the study zones.

Table 1

Rural Zones	Natural and/or functional districts included
I	Las Hurdes, Sierra de Gata, Trasierra-Tierras de Granadilla, and Valle de Ambroz
II	Valle del Alagón, Rivera de Fresnedosa, and Riberos del Tajo
III	La Vera and Valle del Jerte
IV	Tajo Salor and Sierra de San Pedro
V	Las Villuercas-Ibores-Jara and Campo Arañuelo
VI	Comarca de Trujillo, Sierra de Montánchez, and Zona Centro
VII	Lácara Sur and Municipios Centro
VIII	La Serena-Vegas Altas and Guadiana
IX	La Serena and Siberia
X	Río Bodión, Tierra de Barros-Matachel, and Tierra de Barros
XI	Lácara-Los Baldíos, Comarca de Olivenza, and Sierra Suroeste
XII	Comarca de Llerena and Comarca de Tentudía
Urban Zones	Badajoz, Cáceres, Mérida, Plasencia, Don Benito, Villanueva de la Serena, and Almendralejo

In order to correctly conduct the study, a large number of information sources were consulted, ranging from censuses to stated preference surveys. The censuses were useful in defining the dimensions of the population being studied and properly stratifying the survey collection process.

The statistics office of the Educational Quality and Evaluation Department of the Secretariat-General for Education, pertaining to the Ministry of Education and Employment of the regional government of Extremadura, provided the spatial distribution of secondary school students in the region of Extremadura, which proved essential for the correct stratification of the surveys since it provided certain information not contained in the official censuses published by the Spanish National Statistics Institute (INE). Therefore, this is the main source of data used in the study to gather information about students enrolled in secondary schools, with a view to obtaining the sample size to conduct the surveys.

A census of students broken down by educational stage was needed to calculate the sample size to study reading habits among secondary school students. Likewise, to improve the mapping and graphical representation of the information extracted from the surveys, regardless of educational stage, information on the enrolled student population, broken down by town and student place of residence, was needed,. All information was also provided with a breakdown by sex.

The statistical unit for all variables was the individual secondary school student aged 12 or older attending the different kinds of schools at this educational level (compulsory secondary education institutes [IESO] and secondary education institutes [IES]). The biographical variables examined were sex, age, nationality, educational stage, school, province of residence and municipality of residence. The variables employed in relation to reading habits were the number of books read per month, reason for reading, time spent reading, frequency of library visits, favorite literary genres, reading format, reading in free time, language read, use of social networks, and book purchases.

The opinion poll was conducted by sampling the students present at the time of the survey in schools across the entire study area.

The data were collected directly from the sample units at the schools in hard copy format (on paper). An indirect collection method was also available, with data received in a worksheet generated by the Google Drive form application after the surveys had been correctly completed with guidance from teachers/survey takers, but this second collection method had very little impact. Ultimately, a total of 4,194 valid surveys were taken from a population of 75,670 secondary school students.

The survey was conducted from February to June 2017. The days were chosen based on the geographic location of the municipalities in the twelve rural zones drawn up within the study area and the municipalities not included in these zones (towns with more than 20,000 inhabitants). Within the twelve rural zones, one municipality was selected for each population interval, and students of all educational stages present were surveyed, provided there was a secondary school there.

The population figure in the region was taken into account and the corresponding number of surveys needed for each group of municipalities was calculated, as shown below.

The sample base was the total population of secondary school students in the 2016-2017 academic year, broken down by municipality of residence, age, sex, and school. In light of the sample size, the maximum error was estimated to be below 3% for a confidence level of 95%. In addition, it must be noted that no results with fewer than 25 responses were taken into account when displaying the data.

" n " = N ∙ Z 2 ∙ p ∙ q d 2 ( N - 1 ) + Z 2 ∙ p ∙ q

where

Table 2

Total in Extremadura			Respondents per age group								Mean	SD
Sex	Number	Total	12	13	14	15	16	17	18	+18
Females	37,396	2305	375	450	323	427	333	273	96	165	14.8	2.01
Males	38,274	1889	306	365	353	362	262	199	72	85	14.6	1.91

The questionnaire was drawn up based on the information needs defined by the research group and each survey was expected to take around 10 minutes to complete. The paper format questionnaire was tested in December 2016 on respondents chosen at random in two schools in the city of Navalmoral de la Mata (Augustóbriga and Albalat) with the aim of verifying the suitability of the questionnaire, in terms of identifying potential difficulties in understanding the questions, and confirming the average amount of time to schedule for each survey.

The surveys were conducted depending on the number of students present in the classroom during the survey takers’ working hours. The aim was that there should always be more than 25 students to complete the survey per grade or at least per school. Likewise, the students must also come from different grade levels.

After all the paper surveys were completed, the results were manually transcribed into the database. The data contained in the Google Drive worksheets were imported from Microsoft Access (used as the database manager) into the table containing the transcribed results. The information was analyzed using descriptive statistical measures.

Ultimately, surveys were taken at a total of 53 schools, taking into account only those that received more than 25 responses per item. The opinion poll was taken on a sampling of students at public and subsidized private schools in the two provinces, taking into account aspects of interest in the study, such as whether the school belonged to the school library network (Red de Bibliotecas Escolares de Extremadura, REBEX), had a bilingual department or immigrant students, or was implementing a reading program. The distribution per province was as follows:

Table 3

Municipality type	Province of Badajoz / Cáceres
	Schools surveyed	REBEX schools	Schools with bilingual department
< 1,000 inhab.	0 / 2	0 / 2	0 / 1
1,000-2,000 inhab.	1 / 1	1 / 2	0 / 0
2,000-5,000 inhab.	6 / 7	4 / 4	3 / 1
5,000-10,000 inhab.	6 / 8	4 / 4	2 / 5
10,000-20,000 inhab.	3 / 2	1 / 2	1 / 2
> 20,000 inhab.	10 / 7	6 / 4	9 / 6

Table 4

Municipality type	Province of Badajoz		Province of Cáceres
	Females surveyed	Males surveyed	Females surveyed	Males surveyed
< 1,000 inhab.	68	51	155	136
1,000/2,000 inhab.	65	57	169	135
2,000/5,000 inhab.	333	300	157	109
5,000/10,000 inhab.	318	270	214	154
10,000/20,000 inhab.	147	146	171	147
> 20,000 inhab.	489	361	200	186

A database was generated to process the surveys efficiently and effectively. Three entities were described in the database: students, municipalities, and zones. After defining the tables accordingly, queries were created using the SQL query language (the specific language that gives access to a relational database management system to specify different types of operations), and the database was implemented in Microsoft Access.

When the query process was completed and saved in the database, the desired outcome was to display data in relation to the geographical component - that is, produce a thematic map. To achieve this, a connection was needed between Microsoft Access, where the alphanumerical database was managed, and ArcGIS 10.2, the geographic information system in which the graphical outputs would be created.

The research described in the manuscript used a self-report instrument to examine the reading habits of secondary school students. The analysis of descriptive statistics employed in this research has been endorsed by other studies describing the same variables (Pérez-Parejo et al., 2018a; Pérez-Parejo et al., 2018b).

Our research took into account two aspects that are of particular relevance and interest in the literature consulted: disaggregation of results by sex (Logan & Johnston, 2009; McGeown et al., 2012) and the question of motivation for reading (Pitcher et al., 2007). In particular, this last aspect determines to a large extent the results obtained.

In this respect, our results are consistent with those of surveys from other parts of the world. For example, Fuentes et al. (2019) showed that in primary education in Chile, reading attitudes have proven to be relevant in reading performance (p. 225), which is in line with our results for secondary education in Extremadura. In conclusion, this subjective perception of reading is a highly reliable indicator that shapes the full set of results.

The differentiation by sex exhibits similarities with the results of previous studies in Finland and Estonia, with a lower reading index in boys and spaces with demographic characteristics similar to those of our study area (peripheral territories and a dispersed population), although their model of education and PISA results are very different (Säälik, 2015). We are aware that it would be desirable to analyze the socioeconomic variables of the studied population (per capita income) as this is a factor that has been taken into account in other studies (Yusof, 2010), and perform a national comparison with other regions that share the same educational model, and these are limitations of our work. Based on the interesting conclusions reached by other similar studies, our future work should cover, as a new line of research, other variables that allow us to make a deeper comparison with recent studies that point to the relevance of reading profiles and models (Errázuriz et al., 2019) and teachers’ own conceptions about reading (Errázuriz et al., 2020), variables that, as these studies show, condition the results of any survey on this topic. Likewise, it was found that teachers with an epistemic profile derive greater pleasure from reading and tend to teach subjects associated with language, while those with a more reproductive profile report reading for practical reasons (Errázuriz et al., 2019, p. 420).

On the other hand, we believe that a line of research has been opened by the geographical positioning of respondents and the generation of a thematic map of results, which makes it possible to visualize the impact of road and rail infrastructure on reading habits and literacy processes in rural contexts like the one studied.

The paper has great potential and we believe it could make an important contribution to the wider body of research. Reading habits are an important factor in literacy development and may therefore be of interest to readers. Novel aspects of this study include its focus on a minority population and the use of a large, stratified, randomly-sampled population.

The hypotheses or research issues taken as the central theme, which were presented in the introduction to this study and also served as a framework for presenting and analyzing the results, were used in drafting these conclusions.

First of all, the results shown in the maps indicate that the differences between geographic areas within the autonomous community were less conspicuous than expected, revealing progressive educational equity. Even so, the provincial capitals, largest towns, and best connected areas always achieved somewhat better results whereas the lowest scores were observed, with a few exceptions, in zones with the highest levels of rurality, specifically in the districts in northern Cáceres and along the border with Portugal. In this sense, a response is needed from the region’s educational authorities in order to achieve full equality.

It is worth noting that 21% of respondents (14% of women and 29% of men) report not reading any books in a month, although this figure is variable based on the determining factors examined in this paper. Similarly, 67% of students devote less than two hours per week to reading, which is a very low rate with a wide margin for improvement. The general trend is that reading rates decrease as age increases. In addition, students in vocational training read less than those in compulsory secondary education (ESO) or upper secondary school (Baccalaureate).

The differences in reading habits by sex confirm that women read at higher rates than men across all the age groups analyzed, with women reading an average of 1.6 books a month, compared to 1.18 for men. Despite the fact that 54% of students do not read in their free time (35% of women and 59% of men), the trend whereby girls read more than boys still applies. This trend is further echoed in spending on books (16% of women and 25% of men do not buy books), library visits and the subjective perception of reading as a leisure activity.

In terms of the variables related to school participation in special educational programs, the reading rate at bilingual schools is 5 percentage points higher, which is perhaps a smaller difference than might be expected. Whether or not the school belongs to the school library network REBEX does not have a marked impact on the statistics. The data do not vary depending on whether or not the schools participate in reading programs either, which should give the authorities cause for reflection when reviewing these educational programs.

Paper books are the preferred and most common reading format for all ages and educational stages, followed by mobile phones and tablets, which are preferred over computers. The vast majority of students report using mobile phones as a means of accessing social networks in both urban and rural settings. In this regard, mobile use is no lower in rural than in urban settings.

To conclude, it is important to note other data obtained that may affect reading habits. Nearly half of secondary school students (46%) admit that they never visit the library, 44% for women and 49% for men, and two in ten secondary school students do not purchase any books per year. The preferred genres are, in descending order, fictional works and textbooks. In addition, 35% of students admit that they read mainly because they are required to, so they do not consider this a common leisure activity outside the classroom. This subjective perception is always a highly reliable indicator that shapes all results on reading habits across different age groups and educational stages.