The Demographic Factors Affecting the Writing Skills of Slovenian Year 6 EFL Students

efL writing is a complex and difficult productive skill for young learners. National assessment of efL at the end of year 6 in Slovenia shows that additional research into the various variables affecting efL writing is needed. The aim of this study is to predict the relationship between gender, place of living, home possessions, years of schooling and parents’ education, and efL writing performance for year 6 students. a representative sample of 790 pupils completed a demographic e-questionnaire and two writing e-tasks. according to aNova, statistically significant factors are home possessions, number of Slovenian and english books, years of learning english and mother’s education. Statistically insignificant factors are gender, place of living and father’s education. Demographic factors predict efL writing performance by 9.9%.


Introduction
Writing in a first language has been researched deeply over the last 40 years and presented in various models as a complex skill (Weigle 2011, 23). Writing in a second language was first presented in the monitor model (krashen 1982). for young learners, aged five to 12, writing in a first and additional language is a complex skill regarded as particularly difficult, especially in contrast with adults, due to their age-related characteristics such as growth, literacy and vulnerability (Mckay 2011). early recognition of poor writing skills may be crucial information in predicting students' future career opportunities (Pretorius and Naudé 2002). The 2018 results of year 6 students at national assessment show wide variation in Slovenia. as stated in several previous reports, efL writing should be understood as a process that needs a lot of practice and, accordingly, should be developed more systematically and earlier, while appropriate feedback should also be given (rIC 2018, 189). Some students do not even attempt to write or they write in another language (rIC 2019, 189). In general, efL writing goals stated in the national plan are not achieved, which was also reported in 2017, with poor results observed in the case of all three criteria -content, grammar, and vocabulary and spelling (rIC 2022, 6-7).
Demographic factors, among others, are known to influence academic performance (tinto 1975). Some, such as the relationship between age, gender, and academic achievement are under-researched (Slaughter 2007, 52). Based on these findings, we conducted quantitative research to find out the relationship between the selected demographic factors such as gender, socio-economic status, region, years of schooling and parents' education, and efL writing performance. The factors are discussed in depth in the next section.

Theoretical Background
It is encouraging that scores on the efL national assessment at the end of year 6 and participation have been rising in recent years in Slovenia (rIC 2021). however, new research is required to determine why this is so. In contrast, there are still about a half of the students who do not master efL writing skills, or do so only partially. early detection of poor efL writing skills may be a crucial warning that points to the students' future career possibilities. Therefore, efL writing skill must be developed more effectively, as noted in the 2009 report. New research could also explain why the ranking of regions has been the same for several years now.
gender, as one of the key demographic factors, is one of the essential components in evolving the agency and identity of writers, but remains under-researched in both second language writing and second language acquisition as a larger field (kubota 2003). Competence development of efL writing skills from the point of view of gender has been studied, but mostly with students in upper secondary education (keller et al. 2020). female-male differences with primary school learners are usually studied from the perspective of CLIL and writing genres (graham et al. 2021). Steinlen (2018) examined the development of english writing skills of primary school pupils in years 3 and 4 attending a bilingual german-english school as opposed to mainstream programmes, and the results also acknowledged the gender dimension.

varIa
Minimal differences in efL writing proficiency, according to gender, are reported in the national assessment of efL in Slovenia, which is composed of listening, reading and writing tests at the end of year 6 (rIC 2019, 177-78;rIC 2021, 185). This finding is seen as extremely positive and taken as evidence that the topics selected are general enough and relevant to both male and female students. It is further assumed that neither male nor female students are discriminated against by efL teachers.
as reading is a skill strongly related to writing (krashen 1993), it is informative for us to find out the effects of demographic factors such as gender in reading, too. The PISa 2018 results for Slovenia (oeCD data, as reported by šterman Ivančič, štigl and Čuček 2019, 22) indicate that motivation for reading, as measured by the reading satisfaction index, is below average compared to other oeCD countries, which is -0.4 for boys and 0.3 for girls, while in Slovenia it is -0.55 and 0.11, respectively, on a scale ranging from -1 to 1.5. according to Duarete, raposo and alves (2012), a satisfaction index, often referring to customers and a specific moment, can traditionally be measured by a combination of statistical procedures such as regression analysis, descriptive statistics, comparison of means tests, structural equation modelling, analysis of deviations, performance analysis, correlation statistics, importance performance, path analysis and factor analysis.
Socio-economic status (SeS) is known to have an effect on students' academic performance. In many studies, learners coming from low-SeS households develop academic skills more slowly as compared to learners coming from high-SeS environments (Morgan et al. 2009). Serquina and Batang (2018) find that learners coming from families with a low SeS may have the ability to accomplish a task, but then may get interrupted by some social factors. one of the conclusions is that focused attention is impossible for learners whose families struggle financially. Socio-economic and cultural status was also measured in SurveyLang (european Commission and eSLC 2012), the first broad international study of language competences. Students in the last year of lower secondary education or the second year of upper secondary education were invited to participate (european Commission and eSLC 2012, 243). The study considered students' home possessions and parents' educational status. however, in the study's final report no specific demographic factors are discussed, as they are seen to be largely beyond the control of policy makers.
With regard to parents' education, a large correlation was also found in relation to children's academic achievement in Idris, hussain and ahmad (2020). tam (2009) compares the influence of mothers' and fathers' educational level on academic achievement with a sample of hong kong families children, with boys and girls treated as separate groups, based on the claim that the issue in question had not yet been studied systematically. The results show that mothers care about and promote academic achievement more intensively than fathers, which is said to be in accordance with Chinese culture. Boys benefitted more from maternal efficacy than girls, but were more hampered with regard to psychological governance. In contrast, the girls' academic performance was more enhanced by paternal academic efficacy than that of the boys.
Parents' education, as part of SeS, is known to influence the academic achievement and cognitive development of a child. This is achieved through a series of family environment variables such as parents' educational expectations, parenting ideas and behaviours, and the parent-child relationship (Bradley and Corwyn 2001;yeung, Linver, and Brooks-gunn 2002).
The importance of maternal education for children's academic achievements is widely recognized; however, the multiple potential mechanisms that explain this relationship are underexplored (harding, Morris and hughes 2015). The same authors explain that theories of human, cultural and social capital are integrated with two developmental psychology theories -bioecological theory and developmental niche theory. Magnuson (2007) discusses the role of maternal education from the point of view of its improvement in relation to young mothers with a low level of education, and older mothers with a high level of education, in terms of academic achievement and the quality of their home environments in middle childhood (age six to 12). Children of young mothers with low education who improve their education perform better in reading, but less improvement is observed in maths. No impact of education improvement was found with children of older mothers with previous high education.
In view of region as another demographic factor, statistical analyses derived from the results of Slovenian national assessment of efL in 2020/21 (rIC 2021, 185-86) indicate that the ranking of five most successful regions have remained the same for some years, with central region -containing the capital -being the best. as pointed out several times by the national assessment committee, a mixed-methods approach is needed to explain what factors lead to lower versus higher results, and thus what causes differences in the results of efL assessment in year 6 according to the various regions. The research could also explain the differences in results according to the students' place of living.
according to SurveyLang (european Commission and eSLC 2012), school time in relation to (formal) early language learning was a major issue in the eU in 2012, by which time the various countries had gradually increased the overall years spent teaching languages, which was mainly achieved by starting fLL earlier. The topic of early language learning is reviewed by Nagode and Pižorn (2016) in relation to some widespread myths and misconceptions. The myth that "earlier means better" seems to be true, but with certain limitations. The question of the proper age to start learning a foreign/second language is a complex one. When deciding on the starting age, the final aim of a language programme plays a crucial role. Learning conditions, e.g., enough exposure to the language, and possibilities for students to use the language in various contexts, have to be taken into account as well. eU citizens are strongly encouraged and even expected to learn more than one additional language, as stated on the official website of the european Union (cited on august 17, 2022) under the theme of the importance of multilingualism. Language competences are stated to be at the heart of building the european education area. In other words, they are indispensable for mobility, cooperation and mutual understanding across borders.

Method
This is a non-experimental quantitative study. Descriptive and causal-nonexperimental methods of pedagogic research have been used. The research has been conducted according to ethical rules.

Participants
Prior to the study, primary school pupils in general had been found to experience considerable difficulties in efL writing (Jashari and Dagarin 2019). The year 6 students were selected as the central primary school population and the upper year of the second triad of Slovenian primary education. We believed that the results of this study could help us better understand what demographic factors affect students' outcomes in writing in the year 6 National assessment of knowledge (Nak) in efL. In addition, the students participating in this study might have benefited from taking part in it because it would be great practice for them with regard to actually taking the Nak in efL at the end of year 6, as they were also asked to complete two standardized writing tasks. In selecting students, no limitations were applied in terms of gender, ethnicity, special needs or any other background relevant to the context of efL. We refer to year 6 students as younger learners, with the age cut-off of being (approximately) 12 (Mckay 2011, 1). a systematic random sample of 5% was planned from the statistical population of all year 6 primary school pupils taking the regular curriculum in the school year 2017/18 (N = 18.932). In order to select only students with english as their first foreign language, and not german, we have removed the latter group.
The required 5% corresponded to 22.7 primary schools. We concluded that the research should be conducted at 24 schools, 12 of them from larger cities, i.e., capitals of regions, and 12 from smaller towns and villages (abu-Bader 2021). The actual participation was as follows: 310 students from villages (40.5%), 350 students from smaller cities (45.7%) and 105 students from larger cities (13.7%). from the point of view of statistical regions valid in the time of the research, we planned an equable, systematic geographical distribution. The final number of participants amounted to 825 individuals.

Instruments, Materials and apparatus
The demographic questionnaire is based on a broader contextual questionnaire used before in SurveyLang (european Commission in eSLC 2012), and the related description of the results for Slovenia (rutar Leban 2013). The questionnaire is composed of 11 factors: school, region, student, gender, place of living, home possessions in terms of "what" and "how many", number of books in the official language of the country of research, number of books in english, years of previous learning efL, and parental educational level. Theoretically, the range of variables could be extended to include, for example, items signifying wealth specific to the research country. however, this would necessitate additional research to identify these items and an even longer contextual questionnaire. In SurveyLang (european Commission and eSLC 2012), parental educational level was measured on an eight-level scale. Several students found it difficult to answer all the questions. for year 6 students, the existing questionnaire might already be too long considering their concentration span as well as some cognitively demanding items, such as parental educational level. Moreover, it is difficult for schools to allocate (more) time for research. The questionnaire was converted into an electronic form. Some motivational symbols were added. to achieve a satisfactory level of anonymization, students had to enter their code first. as for reliability, it is not calculated for demographic data. a pilot study was conducted with 29 year 5 students from two classes at a primary school in May and June 2017. Students participated anonymously after receiving written permission from their parents or guardians. In general, students liked the electronic tasks more than the traditional paper version. Some technical problems were encountered, such as entering the task more than once. This led to a time-consuming extra effort verifying all entries and selecting the best version for statistical analysis. additionally, 1ka -the system used to collect the data -did not save the results if the students interrupted the task. The availability of the school's computer science teacher and the efL teacher proved to be most beneficial in getting the students to complete the task. Content-wise, the students expressed difficulties in comprehending some questions, and they were offered oral support. Prior to the main research, technical difficulties were addressed. Some instructions were corrected as well. The order (sampling) and timing of tasks was pre-determined.
We also used two writing tasks, which had been used before as standardized tasks for the national assessment of efL at the end of year 6. The paper version was converted into an e-version. Students typed the text into a special field on the screens as part of the selected 1ka programme. Both the written instructions and criteria were preserved unaltered. accurate oral instructions were given. The programme allowed unlimited text length. a child-friendly visual background for the e-tasks was selected, and all the tasks were peer-reviewed.
During the course of our pilot study the students were asked to complete three writing tasks. The task My Favourite School Friend was eliminated from the main study as being less suitable compared to the Monster and Winter tasks. In the task Monster, students were asked to describe the appearance, place of living and preferred activities of the monster in the picture. In the task Winter, they had to write to a pen-friend from an african country to explain what the weather in winter looks like in Slovenia, what you wear when going out and what you can do. The task Winter is more authentic than Monster, as it is set in a real situation and environment.
Both tasks are picture-based. however, from the point of view of predicting writing performance based on task characteristics, it would be more informative if we included two different tasks, such as a listen-and-write or read-and-write tasks. New tasks could also be created, but this was not a priority of this research.

Procedure
Contacts with schools were established by emailing the missive to headteachers. after their consent, the schools received another email with all the details regarding data collection as well as all e-tasks. Parents and guardians received a letter with key information about the research and a written consent form. Participation was announced as anonymous and voluntary, meaning that students could exit the study at any phase, which is a research ethics stipulation par excellence.
Students were asked to write as much as they could. If they could not recall an english expression, they were allowed to use a word in their first language, although this strategy was employed only exceptionally in order to keep the flow of writing. We emphasized that students should write in english. During the pilot phase some students wrote in Slovenian. varIa Nevertheless, the pilot-study-instructions did not explicitly state that the text should be written in english. Students were encouraged to use their imagination in the Monster writing task. as for grammatical accuracy, if they were unsure and hesitating they were told that it was better to misspell the word than not to write it at all. Students were told to read the instructions carefully, to copy the codes accurately, to adhere to the timing and order of tasks (the a/B system was used and each student received the corresponding a or B paper list and was asked to tick off the task after finishing it), the way of asking questions, the need/ possibility of writing paper tasks instead of e-tasks, the position of letters q, w, x and y on the keyboard, and the prohibitions with regard to using dictionaries and talking to schoolmates. all data were collected in an electronic format. Data collection lasted from the beginning of february to the end of april 2018. Schools allowed two class periods for this study. ICt classrooms were used, although of varying quality. Data were collected by the researcher at all schools with efL and ICt teachers assisting this process.
The researcher used about 10 minutes for an introduction as research procedures are a rather unusual situation for both students and teachers. an additional lesson, i.e., three instead of two, would be helpful in this context. Students could ask questions but were asked to do so at the beginning, if possible. efL teachers handed students' written consent forms to the researcher beforehand.
The students were asked if anyone wanted to do the tasks in paper form, providing they could justify their preference. Some students with various special needs were assisted by their teachers. all students preferred e-tasks, probably because these minimize the required input and correspond with greater levels of personal relevance.

Data analysis
to determine the relationship between demographic factors (each factor taken separately), frequencies, mode, skewness, kurtosis, minimum and maximum are calculated. to find out statistical significance, one-way aNova is used. homoscedasticity of residuals is checked by the Breusch-Pagan test of heteroskedasticity. graphically, the relationship between each demographic variable and efL writing performance is presented by histograms. one has been selected to be presented in this paper whereas others are explained in the text. to test the heteroskedasticity of the number of books in the country's official language, and the number of english books together, a two-way aNova is used.
to detect missing data, we could use Analyse > Missing Value Analysis as a standard function in SPSS. however, as we wanted not only to detect missing data but also eliminate all units with at least one missing value, we used a filtering procedure NMISS (v0, v93, v3 to W14) < 1. after filtering, 765 units remain. The final number is still higher than planned. In all further statistical analyses, only units without missing values are used. These are units with answers to all demographic questions and both writing tasks.
No attrition occurred, that is, no participant dropped out. There were also no cases of students' non-response.

variables in the Study
School is a simple nominal independent variable with 21 values, according to 21 participating schools. region is a simple nominal independent variable with 11 values, according to the fact that 11 regions participated. Student, as a code, is a simple nominal independent variable with 765 values, standing for all students whose data were included in the statistical analysis. gender is a simple nominal independent variable with two values. Place of living is a simple nominal independent variable with three values, 1 representing village, 2 town and 3 city. home possessions, in terms of "what", is a compound independent ordinal variable with values from 1 to 16. More than one response is possible. each answer is assigned 1 point. Mean value is calculated. according to SurveyLang (european Commission and eSLC 2012), the first eight home possession items are equal across countries, whereas those from nine to 16 are country-specific. home possessions in terms of "how many" is a compound ordinal independent variable with values from 1 to 4. Students had to select one answer for every home possession from the list. The answers are transformed into points (values) according to the following key: one point for the answer 0, two for 1, three for 2, and four for 3 or more. for all items together the mean value is calculated. Number of books in Slovenian, as well as number of english books, is a simple ordinal independent variable with six values corresponding 0-10, 11-25, 26-100, 101-200, 201-500, more than 500 books. years of learning efL is a simple ordinal independent variable with 10 values, 1 standing for zero years, 1 for two years, 3 for three years and so on, up to 10 for 10 years. Maternal and paternal education are two simple ordinal independent variables with the same seven values, according to the given seven levels of education, 0 for no schooling at all and 7 for the highest level of education. efL writing performance as the dependent variable is a compound with values from 0 to 10. as for the criteria, content (0-4 points), grammar (0-3) and vocabulary/ spelling (0-3) are taken into account. each criterion is an independent ordinal variable. efL writing performance is calculated as follows: ((rating 1 of writing task Monster + rating 2 of writing task Monster) / 2 + rating 1 of writing task Winter) / 2 overall, 2,304 writing tasks were assessed (768 x 3). The number does not take into account the tasks that we excluded from statistical analysis due to missing data. rating 1 was completed by the researcher for both writing tasks, Monster and Winter. finding english teachers for the compulsory rating 2 turned out to be difficult. Seven highly qualified raters -members of the national efL assessment team -helped with rating 2. The researcher also completed rating 2 of 118 Monster writing tasks, but only six months after rating 1. In cases of assessment discrepancy greater than two points, rating 3 was introduced. rating 2 for the writing task Winter was not carried out. In rating 1 the 2017 Nak scale was applied whereas in rating 2 the 2018 Nak scale was used. a comparison of the scales showed very few discrepancies. Intra-rater reliability (0.907) was calculated as 0.846 (the researcher as rater 1 vs. other raters). The logic of the formula leads to the average number of points achieved for both writing tasks.

Results and Discussion
The research question in this study was to identify the relationships between gender, place of living, home possessions in terms of "what" and "how many", number of books in the varIa official language of the research country, number of books in english, and maternal and paternal educational level as demographic variables, and efL writing performance with year 6 students. The statistically significant variables are home possessions, in terms of both, "what" (the most influential variable) and "how many", the number of books in the language of the research country, the number of books in english and maternal educational level. gender, place of living and paternal educational level are statistically insignificant. gender is statistically insignificant as a predictor of efL writing performance (p = 0.051). girls achieved a slightly lower result than 5 points (out of 10) and boys somewhat more than 4.5 points. however, the p value exceeds 0.005 by only 0.1%, so we are adding information on effect size, which is also frequently given in social science studies. effect size and consequently the proportion of variance explained in efL writing performance by gender can be calculated using partial η 2 = 0.005. accordingly, gender explains 0.5% of total variance in efL writing performance. Based on the partial η 2 coefficient, the effect size for gender as a predictor of efL writing performance can be calculated according to Cohen's criteria. It equals 0.07 and is regarded as small.
The result is in accordance with our expectations as the two writing tasks are standardized and, as such, were used for one of the national efL assessments for year 6 students. In the recent corresponding reports of the national assessment committee on the efL assessment at the end of year 6, very little difference in efL writing performance according to gender is mentioned, indicating that the tasks seem to be liked by both girls and boys, and pointing to non-discriminatory efL teaching in relation to gender. however, this finding has not yet been supported by research connected to the national efL assessment for year 6 students. We hypothesize that efL writing performance would be more varied if different types of writing tasks were used, such as a listening and writing task or reading and writing task, which we suggest for further research, taking into account also the findings of kormos and Wilby (2019) on task development. In practice, we would propose all existing ongoing support for (early) reading in the first language and efL to continue. The transition from efL reading to efL writing should be explored and, where possible, strengthened at the national level.
Place of living as a predictor of efL writing performance with year 6 students is statistically insignificant (p = 0.280). Under its influence, the average efL writing performance differs by approximately half a point (4.6 to 5, out of 10). a slightly positive trend in the direction village-town-city can be observed. years of learning english is statistically significant (p = 0.000), as expected. The Breusch-Pagan test confirms homoscedasticity with p = 0.391. a steadily increasing trend in the direction village-small town-large town can be observed. The vast majority of year 6 students (230) have been learning english for three years. The difference in efL writing performance ranges from about 4 to 7 points, which we estimate as large, as compared to the range of difference caused by other variables. home possession ("what") is a statistically significant variable with p = 0.001, as expected according to the known importance of socioeconomic status, to which the variable in question belongs. The Breusch-Pagan test of heteroskedasticity confirms heteroskedasticity with p = 0.036 (lower than 0.05). Nevertheless, this is most probably merely a consequence of the illogical and disproportionately high first (left) bar of the histogram (figure 1), which is represented by only one student. The other bars show a clear upward trend, and efL writing performance ranges from 2 to almost 6. frequencies show that most students fall into the following bars: bar 6 with 25 students, bar 7 with 37 students, bar 8 with 60 students, bar 9 with 67 students, bar 10 with 99 students, 11 with 119 students, 12 with 105 students, 13 with 81 students, 14 with 64 students, 15 with 46 students, and 30 with 16 students. as for the list of 16 wealth indicator items, these should be verified in any further research endeavour for their country-specific appropriateness. The list is based on SurveyLang (european Commission and eSLC 2012). Home possession in terms of "how many" turns out to be a statistically significant variable (p = 0.011). The Breusch-Pagan test confirms homoscedasticity (p = 0.448). The majority of students (553) possess two (same), 90 students claim to possess one item, and 122 claim to have three or more items of the same type. The difference in EFL writing performance is about one point. It arises mainly among students who generally own one item of the same type, as opposed to those who possess two, such as two smartphones. There is almost no difference in EFL writing performance if we compare students with two, and those with three or more items of the same type, according to the list. We can observe a slightly rising trend, that is, more home possessions results in higher EFL writing performance.
In SurveyLang, the item "bathrooms", in terms of how many, suffered from severe misfit in the case of Slovenia so it was excluded from further statistical analysis (European Commission and ESLC 2012, 245). This raises the question of whether the selection is appropriate for a specific country, as there might be some (new) country-specific wealth items. However, the main aim of this research was not to develop a new demographic questionnaire.
The number of books in the official language of the research country is a statistically significant variable (p = 0.015) in predicting EFL writing performance. The difference between its minimum and maximum value is about one point. However, the dispersion, i.e., the results of 95% of the sample, is about two points. A rising trend can be observed, that is, more books in Slovenian leads to higher EFL writing performance. Frequency statistics show that 92 students opted for answer one (lowest number of books), 183 stated two, 287 pupils reported three, 110 four, 67 five, and 26 six. The majority of students possess between 26 and 100 books in the country's official language, Slovenian.
Number of books in English is a statistically significant predictor of EFL writing performance (p = 0.041). A figure 1. The relationship between efL writing performance and home possessions, in terms of "what", with year 6 students. home possession in terms of "how many" turns out to be a statistically significant variable (p = 0.011). The Breusch-Pagan test confirms homoscedasticity (p = 0.448). The majority of students (553) possess two (same), 90 students claim to possess one item, and 122 claim to have three or more items of the same type. The difference in efL writing performance is about one point. It arises mainly among students who generally own one item of the same type, as opposed to those who possess two, such as two smartphones. There is almost no difference in efL writing performance if we compare students with two, and those with three or more items of the same type, according to the list. We can observe a slightly rising trend, that is, more home possessions results in higher efL writing performance.
In SurveyLang, the item "bathrooms", in terms of how many, suffered from severe misfit in the case of Slovenia so it was excluded from further statistical analysis (european Commission and eSLC 2012, 245). This raises the question of whether the selection is appropriate for a specific country, as there might be some (new) country-specific wealth items. however, the main aim of this research was not to develop a new demographic questionnaire.

varIa
The number of books in the official language of the research country is a statistically significant variable (p = 0.015) in predicting efL writing performance. The difference between its minimum and maximum value is about one point. however, the dispersion, i.e., the results of 95% of the sample, is about two points. a rising trend can be observed, that is, more books in Slovenian leads to higher efL writing performance. frequency statistics show that 92 students opted for answer one (lowest number of books), 183 stated two, 287 pupils reported three, 110 four, 67 five, and 26 six. The majority of students possess between 26 and 100 books in the country's official language, Slovenian.
Number of books in english is a statistically significant predictor of efL writing performance (p = 0.041). a slightly rising trend can be observed. More books in english result in higher efL writing performance. The dispersion, however, rises intensively, ergo, from about one point to as much as five, as the number of english books rises. The difference between the minimum and maximum value of efL writing performance is about 1.5 points, that is from about 4.5 to slightly more than 6. a total of 421 students selected the answer one, 221 selected two, 88 three, 23 four, 9 five, and 3 six (as the largest number of english books). frequency statistics show that most students possess 0 to 10 books in english.
If we compare the impact of books in Slovenian to the impact of english books, we can see that the difference in efL writing performance is approximately the same. however, the dispersion of data differs considerably (much larger in the case of english books). heteroskedasticity is tested for both variables (by a two-way aNova) and rejected (p = 0.517).
The number of books in the household, as a traditional socioeconomic measure, as well as the number of children's books as its extension, may be a robust predictor of academic achievement. Nevertheless, the statement refers to parents' questionnaires on the number of books at home (heppt, olczyk and volodina 2022).
Maternal educational level as a predictor of efL writing performance is statistically significant (p = 0.001). It causes a difference of approximately 1.5 points in efL writing performance. a downward trend can be observed, that is, higher maternal education leads to higher efL writing performance of year 6 students. Most mothers (253) attained higher education certificates, 165 mothers (upper) middle school, 139 higher vocational, 111 middle or lower vocational education, 73 students a vocational course or "matura" course, 16 primary school in terms of year 7 to year 9, and eight have an education of year 1 to year 6. There are no cases of mothers without any education.
The growing trend of efL writing performance due to maternal education is in accordance with our expectations, as the importance of maternal education for children's academic achievements is widely recognized (harding, Morris and hughes 2015). We surmise that the extent of 1.5 out of 10 points is considerable. In SurveyLang, the relationship of maternal education and efL writing performance is not discussed, but it could be constructive for comparative analyses. however, the result of this higher maternal education is not necessarily and straightforwardly a steady linear rise of the efL writing performance of their children. factors like mother's age and previous educational level when starting additional education play an influential role. Paternal educational level turns out to be statistically insignificant (p = 0.474). It causes a difference in efL writing performance of about one point. a fluctuating course can be detected. Most fathers (253) have the highest education possible, that is, university degree or higher, 169 (upper) middle school, 253 general middle school including grammar school, 163 higher vocational education, 133 middle or lower vocational education, 85 a vocational course or a course for graduation before leaving secondary school, 20 are reported to have year 7 to year 9 of primary school, three with year 1 to year 6 of primary school, and two with unfinished primary school or no education. heteroskedasticity was tested for maternal and paternal educational levels together and rejected (p = 0.231). The results regarding the relationship between paternal education and efL writing achievement are as anticipated vital for its established charge. That being said and contrary to the sum of our expectations, paternal education had no palpable effects. according to tam (2009), the influence of mothers' and fathers' educational level on children's (boys and girls studied separately) academic achievement should be studied more systematically.
The question about parents' education is relatively difficult for children (year 6), as it might be cognitively too demanding, in other words, too abstract. Moreover, it is difficult to ensure a high degree of anonymity which would be ethically suitable as the question is highly personal. These circumstances reveal that our results regarding paternal education may be less reliable. however, students answered the question on maternal education under the same conditions. Moreover, the question on parents' education is only one component variable of SeS.
overall, statistically significant demographic variables explain 9.9% of the whole variation in efL writing performance (r 2 = 0.099).
Data were also collected for school and region, as another two demographic variables, but not analysed as we are mostly interested in the role of the selected demographic factors as a whole. for the sake of statistical analysis, the student/code is also considered as a demographic variable. Student profiles could be identified.

Conclusion
Demographic variables have an impact of 9.9% on efL writing performance among the year 6 students examined in this study. The variables represented in this percentage are home possessions in terms of "what", home possessions in terms of "how many", number of books in the language of the research country, number of books in english, years of learning english, and maternal educational level. gender, place of living and paternal educational level were found to be statistically insignificant. varIa Considering the study's limitations, educational attainment could be measured at three levels exclusively, according to the 2011 International Standard Classification of education, since the eight-level scale used might be too difficult and abstract for year 6 students. The variables related to books could feature supplementary questions on children's books, as has been found in some other studies. The list of home possessions should be revised, and new country-specific wealth items could be added, whilst certain others deleted. We interpret the insignificance of paternal education as idiosyncratic, but in order to resolve this conundrum an interdisciplinary approach would be needed to transcend the limitations of what we as linguists can provide.
Concerning pedagogical recommendations, both efL teachers and policymakers should continue all activities to promote efL reading, commencing with an emphasis on reading in the first language.
Similarly, the continuity of efL learning in connection with the education system should be further supported in the hopes of maintaining the positive effect on efL writing achievement. Proper beliefs must be thus preserved. The transition from efL reading to efL writing should be explored and, where possible, strengthened at the national level.
amongst other things, writing in a foreign language is indeed a crucial skill enabling students to have better career opportunities, and thus it is important to observe and explore this at an early stage. This study fills the void in the research for this issue in Slovenia. Internationally, there is also less research on efL writing focused on young learners than on adults. We hope that this study offers an interesting and valuable new insight into the nature and importance of various demographic variables as efL writing predictors of year 6 students, both in Slovenia and beyond.