Development of an instrument to measure NEET-youth self-directed learning skills

ABSTRACT

The purpose of this study was to develop a reliable and valid self-report instrument to measure self-directedness in learning for young people who are not in education, employment or training (NEET-youth). Accomplishing this purpose involved two stages: (a) formulating scale items that describe the factors of self-directed learning (SDL), and (b) validating the factor structure of the questionnaire. Exploratory factor analysis produced a six-factor structure of openness to experiences, resilience, attitudes, future orientation, metacognition, and responsibility. The confirmatory factor analysis revealed a satisfactory model fit and factor loadings. The reliability and convergent validity of the instrument were assessed and the results of the six factors were described and compared across different samples. The resultant 34-item SDL Scale for NEET-youth (SDL-NEET) is a highly reliable (α = 0.927) instrument for using with students and NEET-youth.

KEYWORDS:

• Self-directed learning
• skills
• measurement
• NEET-youth
• dropouts

Introduction

There is a significant number of young people in society who do not participate in working life, education or training (NEET-youth). NEET-youth are mostly characterized by a low level of education (Eurostat, 2022), lack of skills, lack of capacity (Bolhuis, 2003) and motivation (Goldman-Mellor et al., 2016). In order to increase the participation of NEET-youth in lifelong learning and help them realize themselves in the labour market, targeted support activities are crucial. Paabort et al. (2023) review about NEET-youth policies highlighted the need for a tailor-made approach and further research about youth’s skills to prevent their long-term NEET-status.

Self-directed learning (SDL) as a key component of lifelong learning has become increasingly important in a rapidly changing society. SDL competencies support continuous personal development and coping with change. Based on the concepts of SDL (Brockett & Hiemstra, 1991; Garrison, 1997; Knowles, 1951), the SDL skills contribute to develop better self-concept, potential, responsibility, motivation etc. These are the aspects that are found to be weak specifically in the case of NEET-youth.

Measuring non-cognitive skills allows for a more comprehensive evaluation of an individual’s abilities and facilitates providing the best possible targeted and tailor-made support. It helps to go beyond academic achievements and consider the broader aspects of the personality, character and potential of young people. Kõiv and Saks (2023) have found that SDL competencies have been studied very little in individuals who have dropped out of education, and SDL measurement tools have not been validated for this target group. In a situation where the existing measuring instruments contain questions that are not relevant for young people not participating in education, there is a need for a new instrument that takes into account the specific context of the target group.

NEET-youth do not participate in education and their skills, attitudes and everyday life management may include significant obstacles. Due to the specific qualities and conditions of NEET-youth (e.g. lack of interaction with teachers and the learning environment, personal characteristics and context), it is crucial to develop a tool that considers these aspects. Reliable data on the SDL competencies of NEET-youth will enable the creation of more precise, effective and targeted interventions to further enhance their learning skills. Therefore, it is important to know which factor structure best suits to measure SDL skills of NEET-youth and what are the specific SDL skills that need to be supported.

Concept of self-directed learning

SDL has been an important concept in adult education and lifelong learning. Knowles (1951) belief that adult people are expected to acquire a mature understanding of themselves, develop a dynamic attitude towards life, and acquire the skills necessary to achieve the full potential of their personalities, is still valid today. It must be considered that the learning path to adulthood begins before becoming an adult, and therefore SDL skills development has been highlighted at all educational levels. The need for SDL skills comes from the fact that SDL has a long-term positive effect on participation in education (Farrington et al., 2012), people with higher SDL skills level are more satisfied with their life and have a more purposeful direction regarding their future aspirations (Edmondson et al., 2012). Knowles (1975) has pointed out already decades ago that new tools and methods in education necessitate a higher level of SDL skills and this statement has become even more important since the developments in digital and remote learning. SDL is a fundamental competence for individuals living in our modern world, where social contextual conditions are changing rapidly, especially in the digital age (Morris, 2019).

According to Knowles (1975), SDL is a process in which individuals take the initiative, with or without the help of others, in diagnosing their learning needs, formulating learning goals, identifying human and material resources for learning, choosing and implementing appropriate learning strategies, and evaluating learning outcomes. Brockett and Hiemstra (1991) further develop the concept by emphasizing the interaction of context and the person’s own responsibility and their Personal Responsibility Orientation (PRO) – a model that is designed to recognize both the differences and similarities between self-directed learning as an instructional method and learner self-direction as a personal characteristic. Garrison (1997) raised concerns that SDL had largely been defined in terms of external control and facilitation, rather than internal cognitive processing and learning, and he proposed an SDL model which integrated external management (contextual control), internal monitoring (cognitive responsibility), and the motivational (entry and task) factors associated with learning in an educational context. Based on these approaches, it can be highlighted that all three factors support a person’s own internal readiness, and that context is especially important in the development of SDL skills. The SDL concept becomes more complicated in the case of a target group that is not in education or employment (NEET-youth) and whose previously mentioned SDL dimensions could be weak.

Kõiv and Saks (2023) analysed SDL concepts and NEET-youth characteristics, and defined SDL for NEET-youth as a supported process in which a person’s attitude towards learning improves, and the subject develops initiative, independence and abilities in shaping their educational path with a positive view of the future. In the context of NEET-youth, this definition highlights the change in a person’s internal monitoring and motivational factors, similar to Garrison’s (1997) concept. Learning as a crucial aspect in a person’s development process requires personal support skills and competencies to enable people to plan their learning paths and to learn by themselves (Candy, 1991). Kõiv and Saks (2023) found that the personal factors of SDL include self-concept, beliefs, orientation, motivation, process-related skills, learning skills and personal characteristics (See Appendix 1). According to Brockett and Hiemstra (1991), the external characteristics of an instructional process and the internal characteristics of a learner should match, so that the process fits the needs and desires of the learner and the social context in which the learning takes place. Kõiv and Saks (2023) have highlighted the more precise dimensions of the SDL concept, divided between processual and personal aspects: 1) preconditions, elements and tools as process dimensions, and 2) preconditions, skills and goals as personal dimensions. When supporting the development of self-directed learning, different SDL tools and elements must be applied to bring about changes in the person’s SDL skills.

In summary, it can be stated that SDL is a valuable concept that supports entry into and persistence in lifelong learning, but in the development of which different factors must be observed. Therefore, when aiming for the active participation of the population in lifelong learning, the needs and context of different target groups must also be considered, so that the support provided is available and appropriate.

The characteristics of NEET-youth

In 2021, 13.1% of the 15–29-year-olds in the EU were neither in employment nor in education and training (Eurostat, 2022). Young people who are not in education, employment or training are called NEET-youth (Furlong, 2006). According to Eurostat (2022), the highest number of NEET-youth is among young people with low education. There can be different reasons why young people drop out of education. These factors can be loss of interest in learning, lack of social support, or relationships with teachers and parents (Connell & Ryan, 1989; Duffy & Elwood, 2013; Vasalampi et al., 2018; Zimmerman & Pons, 1986). However, the listed reasons may be less important for those young people who have had to interrupt their studies due to illness or disability. Mascherini (2019) has pointed out that young people with poor health and disabilities are 38% more likely to be NEET-youth compared to those in good health, and 7% of the NEET-youth belong to this sub-group.

Low skill level and low educational attainments significantly increase the probability of reporting long-term NEET status (Barth et al., 2021; Jongbloed & Giret, 2021). The NEET-youth condition is associated with a negative vision about the future, and such a negative perspective involves difficulty in thinking about the future as beautiful and well planned, and setting goals and considering the specific means of reaching those goals (Parola et al., 2022). Young people who shared their thoughts about their futures were also likely to have a reduced risk of becoming NEET-youth (Cheng & Nguyen, 2022).

Emotional and ontological aspects can lead to a negative impact upon self-concept (e.g. low self-esteem and self-confidence, low aspirations, motivation and expectations, negative academic perception, fear of failure, stress and anxiety, low resilience) (Brown et al., 2021). Tayfur et al. (2022) highlight the findings that having lower self-esteem, external locus of control, and no job aspirations significantly increased the likelihood of being NEET by around 25 or 26 years of age. Duckworth and Schoon (2012) suggest that improving school motivation and educational engagement among young people, as well as school characteristics, may represent possible leverage for reducing the risk of subsequently becoming NEET.

The importance of soft skills has been emphasized in a number of studies, but few studies have been about the skills of NEET-youth. Goldman-Mellor et al. (2016) have found that NEET-youth are associated with lower personal and social well-being, and they often describe themselves as having few ‘soft’ skills and not feeling optimistic about their chances of ‘getting ahead in life’. Ellena et al. (2021) point out that soft skills retain a certain importance in fully understanding the NEET phenomenon; however, only a few researchers have focused on them specifically. Therefore, there is a gap in the literature both in terms of studying NEET-youth soft skills in general, but also measuring skills.

Young people who leave education and training prematurely are bound to lack skills and qualifications, and they face a higher risk of unemployment, social exclusion and poverty (Konle-Seidl, 2021). Therefore, NEET-youth find themselves in a trap from which it is increasingly difficult to escape, because low level skills bring difficulties in acquiring and developing skills, which also makes them less likely to engage in training or further schooling (Vugt et al., 2022). Prior achievement, as well as educational aspirations and school engagement, and school characteristics can help young people overcome disadvantage and avoid becoming and remaining NEET (K. Duckworth & Schoon, 2012).

As education plays a significant role as a predictor of NEET status, declining rates of early leavers from education and training can be considered an important factor in lowering the risk of becoming NEET-youth (Konle-Seidl, 2021). Therefore, it is not only important to get young people back to school, but also to support their willingness to stay on the education path. Rising SDL skills is a way of turning individuals into lifelong learners (Candy, 1991). Bringing NEET-youth back into education is a supported process that has important implications for continuing education. According to different characteristics both personal and contextual, it is necessary during the supporting process to assess and understand NEET-youth SDL skills in terms of various factors. Measuring SDL skills is critical for those individuals who need additional support to come back and/or participate in the learning process. A more precise understanding of their needs facilitates targeting the support as precisely as possible.

Overview of existing SDL instruments

Measuring non-cognitive skills can provide insights into the strengths of individuals and highlight areas for improvement. A. L. Duckworth and Yeager (2015) have highlighted that measuring these skills contributes to well-being and achievement in young people, providing an understanding of their development, and impact on life outcomes. By understanding which non-cognitive constructs have an impact, we can better understand which skills matter for success in life (Humphries & Kosse, 2017).

The tools used for measuring SDL in young people so far have varied. Kõiv and Saks (2023) have identified 16 different evaluation instruments used for young people in the form of self-report questionnaires and one interview. The main SDL measurement instruments, which have been used individually or in combination with other instruments, are SDLRS (Guglielmino, 1977), SDLRS (M. J. Fisher et al., 2001) and SDLS10 (Lounsbury et al., 2009), as well as SDLA (Chang, 2006), PRO-SDLS (Stockdale & Brockett, 2011), SDLI (S.-F. Cheng et al., 2010) and SDLI (Suh et al., 2015) (see Table 1).

Table 1. The main measurement tools used to measure SDL of young people.

Year	Author(s)	Scale (No of items)	Sample	Validity	Reliability
1977	Guglielmino, L.M.	SDLS (58 items)	307 students	Delphi method	0.87
2001	Fisher, M.J., King, J., & Tague, G	SDLRS (40 items)	130 students	CFA	0.87
2006	Chang, C.-C.	SDLA (37 items)	38 undergraduates	N/A	0.92
2009	Lounsbury, J.W., Levy, J.J., Park, S-H., Gibson, L.W., & Smith, R.	uni-dimensional instrument (10 items)	2,125 students	CCFA	0.82
2011	Stockdale, S.-L., & Brockett, R.-G.	PRO-SDLS (25 items)	190 undergraduates	CFA	0.91

None of these instruments were used to measure SDL in NEET-youth, the target group outside education and the labour market. In the case of the most used scales, the main characteristic is that multidimensional scales have between 10 and 58 items; in order to measure the validity, the confirmatory factor analysis (CFA) method has mainly been used, and the reliability of all the presented scales is high (0.87–0.92). These scales could also be used to create the SDL Scale for NEET-youth.

Research problem, goal and research questions

In order to effectively support SDL in young people, there is a need to assess their SDL competences. Previously, different methods and tools (e.g. Guglielmino’s SDLS 58-item scale, Fisher’s SDLRS 40-item scale etc.) have been used to measure SDL skills. Kõiv and Saks (2023) analysis, however, revealed that no specific efforts have been made to define or measure SDL in NEET-youth or other target groups not involved in education or employment. In all of the mentioned scales, there were statements that assumed that the respondent is participating in education or that components of institutions or guidance are involved (i.e. I set specific times for my study (M. J. Fisher et al., 2001); I am good at finding the right resources to help me do well in school (Lounsbury et al., 2009); I am an effective learner in the classroom and on my own (Guglielmino, 1977); I usually struggle in classes if the professor allows me to set my own timetable for work completion (Stockdale & Brockett, 2011)). Furthermore, the existing scales have not been validated for dropouts or NEET-youth as the target group. Therefore, in this study, a scale was created that does not assume participation in education or a specific learning environment, or communication with a teacher, but the possibility that informal learning still takes place has been considered. Even the PRO-SDLS scale (Stockdale & Brockett, 2011), included a few statements about the learning environment and guidance, and the majority of the statements expected respondents to be engaged in the learning process. Furthermore, according to the authors, PRO-SDLS (Stockdale & Brockett, 2011) holds considerable promise in the study of self-directed learning within the higher education setting. While the majority of SDL studies are for individuals studying in an educational institution, the uniqueness of this study lies in its target group, whose non-participation in education was specifically considered when creating the questionnaire. It is common to create situation-based or target group-based questionnaires on the basis of existing instruments. The main goal of this study is to compile and validate an instrument (SDL-NEET) for measuring self-directed learning skills in NEET-youth. In this study, it is assumed that informal learning takes place in every person’s life, regardless of their affiliation with an educational institution. In the case of NEET-youth, any instrument must consider the fact that the respondents cannot be expected to participate in any form of guided learning process. To create such an instrument, it is necessary to study the extent to which the questionnaire is suitable for measuring the self-directed learning of the sample group. Based on the goal, the following research questions were posed:

1. What factor structure best suits measuring self-directed learning skills in young people?

2. What are the specific self-directed learning skills of NEET-youth to be supported, in comparison to students or employed youth?

Methodology

Research design

Accomplishing the aim of the research involved four stages: (1) creating scale items according to the 7 personal factors of the theoretical model (Kõiv & Saks, 2023) (see Figure 1), 2(2) checking the comprehensibility and clarity of the statements with cognitive interviews, (3) collecting data from youth aged 18–29, and (4) conducting data analysis according to the research questions. Data management and statistical analyses were performed using SPSS version 28.01.1.1. and SPSS AMOS version 26. Participation in the study was voluntary, sensitive data was not collected and the research used only anonymized data. Therefore, the approval of the ethics committee was not sought.

Figure 1. Theoretical model measurement factors (authors, 2023).

The research design process, including data collection and analysis, is described in Figure 2 (See Figure 2). In the first phase of the research process, a measurement tool was created based on previously used measurement tools, their factors and statements, which were adapted according to the specifics of NEET-youth. Ultimately, a 56-item questionnaire covering 7 factors was used to collect the data. A sample of young people aged 18–30 years was recruited through educational institutions and specialists working with NEET-youth. Purposive sampling (NEET-youth or studying/working youth) was used to select participants based on the specific criteria relevant for this study (young people who were not in employment or education, and young people who were studying or working). A statistical description of the sample was prepared from the obtained results. The 7-factor instrument was analysed using CFA based on the collected data. According to the CFA results, it was necessary to decide whether to continue with the existing factor structure or find a structure with better results. As the model results were not good enough, an exploratory factor analysis (EFA) was conducted to find an alternative factor structure. The resulting 6-factor, 34-item questionnaire was in turn analysed using CFA. The results obtained were significantly better than the first structure. In order to assess whether the instrument is useful in measuring the self-directed learning skills of NEET-youth, the average results of the two target groups were compared using a Mann-Whitney U test.

Figure 2. Research design process.

Data collection instrument

Creating the data collection instrument, SDL-NEET, was theoretically grounded on the results of the systematic literature analysis (Kõiv & Saks, 2023), which aimed to define the construct of SDL in the context of NEET-youth and investigate the dimensions of the construct of SDL. The current study uses prior knowledge about how existing measurement tools can be useful in developing a new scale. In the first phase of developing an SDL-NEET instrument, the researchers analysed the content of five existing SDL instruments with greater reliability (0.82–0.91) (see Table 1), counting a total of 321 items across 30 factors. Items with similar meanings were merged and items deemed irrelevant in the NEET-youth context were then deleted (e.g. interaction with teacher, learning environment etc.), resulting in 56 items. The items were then aggregated according to the distribution of the 7 factors (see Figure 1) relevant for NEET-youth: self-concept, beliefs, orientation, motivation, process-related skills, learning skills and personal characteristics. In the adaptation process, the items were translated into Estonian and back-translated into English, the back-translation was compared to the original version. The statements of the final version were linguistically edited by two language experts.

Before sending the questionnaire to the sample, three cognitive pretesting interviews were conducted to determine whether the respondents interpret the items consistently. Cognitive interviewing identifies various difficulties that respondents may experience when attempting to answer survey questions, and identifying these difficulties allows the questionnaire to be improved before distribution (Miller et al., 2014). The individual face-to-face interviews were conducted using the think-aloud method, using 2 interviews with NEET-youth and 1 interview with a student from school. As the result of the cognitive interviews, three items were modified to make the content clearer and more understandable to young people.

For the proposed scale, a 5-point Likert-type format was chosen as the format that best reflects the respondents’ degree of agreement or disagreement with statements (Jebb et al., 2021) pertaining to self-perceptions of their actions and beliefs in self-directed learning practices. The questionnaire started with a brief introduction where the aim of collecting the data was explained to the respondents. The online questionnaire consisted of 9 background questions (e.g. on age, status, living place etc.) and 56 statements on the respondents self-directed learning practices.

Participants

The questionnaire was sent to 421 young people, of which 316 completed the questionnaires in full. The following table (see Table 2) presents the demographic background of the sample.

Table 2. Demographic data of the participants.

Variable	Sample total (n = 316) %, SD	Students/employed youth (n=250), %, SD	NEET-youth (n=66), %, SD
Gender
Male	128 (40.5%)	105 (42.0%)	23 (34.9%)
Female	185 (58.5%)	144 (57.6%)	41 (62.1%)
No answer	3 (1.0%)	1 (0.4%)	2 (3.0%)
Age
Age	19.92 (SD = 2.79)	19.31 (SD = 2.29)	22.23 (SD = 3.42)
Living place
Urban	199 (63%)	158 (63.2%)	41 (62.1%)
Rural	117 (37%)	92 (36.8%)	25 (37.9%)

The sample was divided into two sub-groups – NEET-youth and students/employed youth – with the sample size of 66 and 250 respectively. Comparing this to the data on NEET-youth in Estonia held by Statistics Estonia (2022), the sample of NEET-youth in this research is similar to the NEET-youth profile in Estonia in the 15–29 age group: rural NEET-youth 33.4%, urban NEET-youth 66.6% and the incidence of females (51.7%) is a bit higher than males (48.3%) (Statistics Estonia, 2022). The demographic data for the NEET-youth sub-group was completed by respondents from urban (62.1%, n = 41) and rural (37.9%, n = 25) areas, divided by gender so that 34.9% were male (n = 23) and 61.2% female (n = 41), and the average age was 22.23 years (SD = 3.42).

The age of NEET-youth is specified only within age groups in the statistics database and so it is not possible to find the exact average age of NEET-youth in Estonia. In recent years, the number of young people aged 15–24 with NEET status has grown rapidly from 8,600 in 2019 to 13,400 in 2021, making up 60% of the number of young people with NEET status aged 15–29 (Statistics Estonia, 2022). The same trend has been detected in Europe, the share of NEET-youth decreased considerably in 2021 compared to 2020 for all age groups except those aged 15–19 (Eurostat, 2022). Considering that the highest rate of NEET-youth in Estonia is in the age group 20–24 years (Eurostat, 2022) and the NEET-youth average age in this study is 22.21 years, this confirms that the sample group of NEET-youth is similar to the profile of the whole NEET-youth population.

Data collection

The data were collected electronically using the LimeSurvey environment in March – June 2022. The SDL-NEET, together with the participant background questions was distributed to a convenience sample by selected youth institutions and specialists.

Institutions from different regions of Estonia were asked to share the questionnaire to the target group – to those who were in education or of NEET status. According to the Estonian Social Welfare Act (Riigi Teataja, RT I, 2023) §15, the conditions and age of young people in need of assistance in social services is set as 16–26. The recommendation of the European Commission (2020) is to widen the age bracket to include young people until the age of 29. Considering the adult status and the age definition for NEET-youth, the sample for this study is young people aged 18–29.

Data analysis

An exploratory factor analysis with principal components method was conducted to find the factor structure for the SDL-NEET scale. In the current study, χ2 statistic, CMIN/DF, Comparative fit index (CFI), Normed fit index (NFI), Tucker-Lewis index (TLI), and the root mean square error of approximation (RMSEA) were used to assess goodness-of-fit of the factor structures. The indices are acceptable at the following levels: CMIN/DF < 3.0, CFI ≥ 0.95, NFI > 0.9, TLI > 0.90, and RMSEA ≤ 0.06 (Hu & Bentler, 1999; Kline, 2005). As the two sample groups were not equal and the NEET-youth group had only 66 respondents, the non-parametrical Mann-Whitney U test was used. The Mann-Whitney U test is a suitable analysis method to compare two independent groups (NEET-youth and students/employees) that do not require large normally distributed samples (Nachar, 2008). The test result will help to answer the second research question – which are the specific SDL skills of NEET-youth to be supported.

Results

Factor structure of the measure of youth self-directed learning skills

In order to answer the first research question – What factor structure best suits measuring the self-directed learning skills of young people? – confirmatory and exploratory factor analyses were performed. Factor analyses were employed to assess the fit of the data with the 7-factor structure of the theoretical model in the personal dimension (, 2023).

Confirmatory factor analysis

First, the factor structure of the theoretical model (Figure 1) derived from the systematic literature analysis by Kõiv and Saks (2023) was tested using a confirmatory factor analysis (CFA).

The CFA was performed to assess the structure of the instrument and identify the optimal model. The theoretical model contained seven factors: self-concept, beliefs, orientation, motivation, process-related skills, learning skill, and personal characteristics. These factors can be divided into the following dimensions: skills, characteristics and attitudes. In order to test the factor structure empirically, a CFA with a maximum-likelihood method of estimation was employed with the sample of NEET-youth and employed (learning or working) youth.

The goodness-of-fit indices of the primary 7-factor and 56-item model indicated a poor fit (χ² = 3591.58, df = 1463, CMIN/DF = 2.455(<3), CFI = 0.722(>0.95), NFI = 0.610(>0.90), TLI = 0.708(>0.90), RMSEA = 0.068(<0.08)). Examining the factor loadings and deleting twelve items with loadings lower than 0.4, the model fit indices were still poor.

Model modification is often performed after discovering a poorly fitting model. Modification indices (MI) are the statistics that can be used to select parameters to add to a model to improve the fit (Whittaker, 2012). Modification indices indicate how much the model fit would improve if the parameter were free instead of constrained. When testing the modification indices, more than half (35) of the items had an MI > 10. If two items strongly correlate, it means they are probably measuring the same thing or these items should be located under factors differently. After freeing highly correlated items, the model fit indices improved (χ² = 1481.587, df = 841, CMIN/DF = 1.762, NFI = 0.802, CFI = 0.902, TLI = 0.890, RMSEA = 0.049) but not sufficiently.

The hypothetical model was created on the basis of measurement tools used for full-time students and primarily studying and working people, considering the specifics of the situation of NEET-youth. It appeared that a measurement instrument aimed at young people, including NEET-youth, needed to have a more specific factor structure. A hypothesized factor structure may perform better for other target groups, such as older age groups.

Exploratory Factor Analysis

Due to the poor fit of the 7-factor structure of the theoretical model revealed by the CFA, an exploratory factor analysis (EFA) was conducted to find an alternative factor structure. Understanding that the items may load differently in the factors, and another factor structure may be more relevant, a principal components analysis was performed. The EFA as a statistical technique was used to explore the underlying factor structure for the SDL-NEET instrument, restructuring the items and factors in the model and reducing the number of items in the SDL-NEET. EFA is a method that attempts to identify the smallest number of hypothetical constructs that can parsimoniously explain the covariation observed among a set of measured variables and to identify the common factors that explain the order and structure among measured variables (Watkins, 2018).

In order to find a factor structure with a better fit, an exploratory factor analysis using principal component analysis factoring with oblique rotation (Direct Oblimin) was conducted. The oblique rotation method produces a factor structure that is more flexible and allows for the possibility of correlated factors (Costello & Osborne, 2005). This can be useful when there is a theoretical expectation or empirical evidence suggesting that the factors may be correlated.

Proceeding from the theory, a five, six and seven-factor analysis was chosen to test the structure. When testing the five-factor structure, factor loadings greater than or equal to 0.4 accounted for 46.8% of the cumulative variance that explained less than half of the strategy use being represented by the items in SDL-NEET (Eigenvalue 1.649). In the case of the five-factor structure, the groups that formed were mixed, having 3 to 17 items in one factor. When testing the six-factor structure, factor loadings greater than or equal to 0.4 accounted for 49.42% of the cumulative variance that explained a little less than half of the strategy use (Eigenvalue 1.469). In the case of the six-factor structure the groups that formed were also mixed, having 3 to 10 items in one factor, including 3 factors with 3 items. In the case of the 5 and 6-factor structure, the statements gathered under the factor did not form a single, clear whole. For example, a 17-item factor in the 5-factor structure should simultaneously explain both attitudes towards learning and the learning process. In the case of the seven-factor structure, factor loadings greater than or equal to 0.4 accounted for 51.86% of the cumulative variance, which explained over half of the strategy use being represented by the items in SDL-NEET (Eigenvalue 1.368). The structure of the 7-factor was clearly the best compared with the other structures and the further development and validation of the scale was based on this structure.

Confirmatory factor analysis

Continuing the validating process with a CFA one more item revealed a low factor loading (lower than 0.4). Removing this item left the fifth factor with only one item; therefore, this factor was removed and the final number of factors was decreased to 6. So, the CFA resulted in a final SDL-NEET model with 6 factors and 34 items. According to the new loadings, the factors were renamed. The six factors (see Appendix 2) are: openness to experiences (6 items), resilience (4 items), attitudes (8 items), future orientation (3 items), metacognition (6 items), and responsibility (7 items). Only one item (I am opened to new ideas) appeared under two factors (responsibility and openness) with almost equal factor loading. If we remove this item from the openness to experiences factor, the alpha coefficient for that factor would decrease to 0.782, and by placing the item under the responsibility factor, the coefficient alpha would increase to 0.858. Due to the fact that the responsibility coefficient is already higher than the other factors, the item was left in the openness to experiences factor.

The internal consistency of the items within each factor was assessed using Cronbach’s alpha. The results revealed that the alpha coefficients in all factors were above the acceptable level of .60 (.62–.87) (see Table 3). The low coefficient in the future orientation (0.62) factor may be caused by the small number of items (n = 3) in that factor.

Table 3. Cronbach’s alpha coefficients and variance of the SDL factors.

Factor	Variance	Cronbach’s α	Number of items
Metacognition	30.75%	0.80	6
Resilience	8.36%	0.70	4
Attitudes	6.62%	0.87	8
Future orientation	4.27%	0.62	3
Responsibility	3.49%	0.85	7
Openness to experiences	3.46%	0.82	6
Total	56.95%	0.927	34

To verify the factor structure, the CFA was performed again and the goodness of fit was still a bit lower than expected but after freeing the association of highly correlated items, the model goodness-of-fit increased (χ² = 833.673, df = 498, CMIN/DF = 1.674, CFI 0.924, NFI = 0.832, TLI = 0.914, RMSEA = 0.046). The model fit indices of the new 6-factor and 34-item SDL-NEET are now acceptable for use in the research process.

In summary, the factor structure of the SDL-NEET scale created on the basis of the theoretical model was not confirmed. The new structure explored using an EFA and confirmed via a CFA reveals a different division of factors. When we compare the factors of the theoretical model with the factors of the SDL-NEET (see Appendix 3), we can highlight that most of the items from the factors self-concept and beliefs were loaded into the attitudes factor, and personal characteristics into the responsibility factor, orientation into the openness factor, and learning skills into metacognition factor. Items in the process skills and motivation factors were distributed more widely and at the same time two new factors – resilience and future orientation were formed using different items. The biggest change was that the items decreased in the process skills and motivation factors of the theoretical model and the new factors resilience and future orientation emerged. In addition, the orientation factor in the theoretical model is not so much oriented towards learning as openness to new experiences in the new structure. The personal characteristics factor is considered more specifically as the capacity for responsibility in the new structure. The process skills factor is considered more specifically as the resilience factor in the new structure.

The emergence of such new factors can be attributed to the circumstances that characterize the targeted group of young people as coping with stressful situations and having an unclear view of the future.

The resulting factors were renamed according to their content – openness to experiences, resilience, attitudes, future orientation, metacognition, responsibility. The next step in the research process was the comparative analysis of the target groups based on the means of the factors.

Comparison of the SDL skills held by NEET-youth and students/employees

In order to answer the second research question – What are the specific self-directed learning skills of NEET-youth compared to students/employed youth? – the means of the measured SDL factors were compared between the two subgroups: NEET-youth and young people who are studying and/or working.

Comparing these subgroups makes it possible to identify any significant differences between the groups and evaluate whether the scale adequately represents the content domain it is intended to measure for the relevant group. After checking the variance homogeneity and normal distribution, the non-parametrical -Whitney U test was conducted to examine the difference in the mean scores of the self-directed learning skills between NEET-youth (n = 66) and young people who are students/employees (n = 250). The results indicated that NEET-youth generally have lower results in all SDL factors compared to students/employed youth. The three SDL factors that revealed statistically significant differences between the two subgroups were future orientation [z =-2.237, p = 0.025], responsibility [z = 2.976, p = 0.003], and openness to experiences [z = 3.976, p = 0.001] (see Table 4).

Table 4. Differences between NEET-youth’ (n = 66) and students’/employees’ (n = 250) SDL factors.

	NEET-youth		Students/employees		z	t	df	p
	M	SD	M	SD
Metacognition	3.56	0.87	3.75	0.74	−1.378	1.74	314	0.168
Resilience	2.78	0.88	2.93	0.85	−1.397	1.33	98.92	0.162
Attitude	3.04	0.91	3.16	0.86	−0.939	0.97	97.67	0.348
Future orientation	3.19	0.99	3.49	0.85	−2.237	2.26	91.89	0.025
Responsibility	3.76	0.94	4,16	0.58	−2.976	3.28	78.46	0.003
Openness to experiences	3.29	0.93	3.70	0.73	−3.243	3.13	87.24	0.001

Note: Statistically significant results given in bold.

These results indicate that the newly-developed scale (SDL-NEET) reliably measures the SDL skills and makes it possible to distinguish the estimates of NEET-youth and students/employees. By comparing the means of the different factors, it can be concluded that there is a statistically significant difference between the two groups in three factors: responsibility, openness to experiences, and future orientation. Therefore, these factors should be given special attention when supporting the development of SDL skills in NEET-youth.

Discussion

The main goal of this study was to compile and validate an instrument (SDL-NEET) for measuring SDL skills in NEET-youth. The newly formulated factor structure creates a specific concept of SDL, while it also confirms the definition of SDL for NEET-youth established by Kõiv and Saks (2023). The validity study indicated that the theoretical SDL factor structure did not measure young people’s self-directed learning skills reliably. The new validated structure, SDL-NEET, with 6 factors (metacognition, attitudes, resilience, future orientation, responsibility and openness) and 34 items had the best fit to measure the SDL skills of young people. The results indicated that NEET-youth have generally lower results in all SDL factors compared with students/employed youth and the three SDL factors that revealed statistically significant differences were future orientation, responsibility and openness to experiences.

The creation of a new factor structure provided a more accurate and specific factors. In the new structure, metacognition is considered as knowledge or beliefs about oneself as cognitive agents, about tasks, actions or strategies, and about how all these interact to affect the outcomes of learning (Flavell, 1979). This approach is coherent with learning skills, as half (3) of the items in the metacognition factor were related to the factor of learning skills in the theoretical model. In the process of developing SDL skills, attitudes are important, both as preconditions of the process and developing a goal during the support. Self-concept and beliefs as factors in the theoretical model were concentrated in the attitudes factor in the new structure, while attitudes had the strongest position in all 5, 6 and 7-factor structures analysed. As attitudes is considered a person’s degree of favourable or unfavourable conviction about a psychological object and can be changed based on beliefs and experiences (Ajzen & Fishbein, 2000), the process of supporting the development of SDL skills has a great impact on changes in attitudes to learning. The orientation factor in the theoretical model may be connected to openness, while in the new structure the majority of these factor items, together with some items from other factors, more firmly formed a factor whose common denominator is openness to new experiences. Openness to new experiences refers to one of the Big Five personality traits and is seen as a recurrent need to expand and explore experiences (McCrae & Costa, 1997).

The factors resilience and future orientation did not find any clear overlap with factors from the theoretical model and can be considered as new factors in the structure. Future orientation is the image individuals have about their future and provides the grounds for setting goals, planning, exploring options and making commitments that guide the person’s behaviour and developmental course (Seginer, 2008). Resilience is a broader construct that encompasses different aspects of adaptation and coping (Wu et al., 2013), while it is also seen as a process-based approach including various self-regulatory efforts in the domains of affect, cognition and behaviour (D. M. Fisher et al., 2019). Resilience refers to a dynamic process encompassing positive adaptation within the context of significant adversity (Luthar et al., 2000). In the context of NEET-youth, this is an important aspect. According to Santilli et al. (2017), resilience and future orientation have importance in adolescent career development, as they are associated with career adaptability and life satisfaction and play a relevant role in life designing. Responsibility in SDL can be defined as an individual’s ownership of their thoughts and actions, having control over how to respond to a situation, and accepting the consequences of their thoughts and actions as a learner (Brockett & Hiemstra, 1991). In creating the new factor structure, the personal characteristics factor in the theoretical model was specified, highlighting responsibility as the main characteristic.

Based on the above, it can be pointed out that the creation of the new factor structure provided a more accurate and specific factor structure than the theoretical model. The results of the current study are consistent with previous studies that suggested that NEET-youth may have a negative future vision (Parola et al., 2022), loss of interest in learning (Connell & Ryan, 1989) and a low level of soft skills (Goldman-Mellor et al., 2016). Comparing the results obtained from the current study and the characteristics of NEET-youth, it can be confirmed that the new scale for measuring SDL skills in NEET-youth is in line with the characteristics and definition of NEET-youth.

Limitations

In collecting data from NEET-youth, the authors considered the idea that NEET-youth is the hard-to-reach target group, and the members of this population may be less willing (or less able) to respond to the survey than those in the general youth population. Hard-to-reach is a term used to describe those sub-groups of the population that are difficult to reach or involve in research due to their physical and geographical location or their social and economic situation (Shaghaghi et al., 2011). As this target group uses specialist support, facility-based sampling was crucial for making contact with NEET-youth.

Given that it is a hard-to-reach group, and the data were collected from those who were ready to volunteer, the data analysis must also take into account that responses from young people in an even weaker position may have been missing. Due to the small sample size the EFA and CFA were made with the same sample. The Mann-Whitney U test was used because of the unequal distribution of the two groups. The validity studies should be replicated with various samples, including NEET-youth, and conducted internationally to allow a larger sample to be included.

Conclusion

SDL-NEET was an instrument developed to measure SDL skills in NEET-youth. The validity of the scale is supported by the results of this study. In conclusion, it can be said that the current study contributed in two ways to supporting NEET-youth returning to education.

First, it gave a clearer understanding of the SDL concept in terms of NEET-youth as the target group. During the process, the factors of the SDL concept were specified and the consistency of the factor structure with the definition of SDL for NEET-youth was also confirmed. The study confirmed that the SDL skills of NEET-youth are lower than student/employee youth, and significantly lower in those factors that overlap with the characteristics of NEET-youth identified in previous studies. Based on the collected data it can be said that the SDL skills of NEET-youth are at a low level, especially in terms of their future orientation, openness to experiences and responsibility. In the case of NEET-youth, the significance of a tailor-made approach providing support has been emphasized (Mascherini, 2019). The use of the scale facilitates the examination of sub-group differences. For instance, due to illness or disability, NEET-youth may have higher SDL skills but they have more limited opportunities. Applying a scale allows these variations to be discovered and highlighted.

Second, it provided a valid self-report questionnaire (SDL-NEET) for measuring the SDL skills of young people. Due to its good validity indicators, the instrument could also be successfully applied to other groups of young people. Its structure, which somewhat differs from the structure of the theoretical model, reflects the more detailed nature of NEET-youth and the SDL concept. Further studies could test whether this instrument could also be applicable with younger age groups and whether its factor structure could be valid with different samples.

The practical value of the implementation of the SDL-NEET scale lies in its application in working with young people, especially NEET-youth, in order to clarify which aspects of the case management process to focus on according to the needs of young people. A. L. Duckworth and Yeager (2015) pointed out that measuring personal qualities is only the first step, the collected data must be used to inform action. Pre-intervention measurements provide insights into the specific areas or factors that require attention or improvement. By identifying existing challenges or gaps, it can be possible to design interventions that target those specific areas and focus on bringing about positive change. The intervention may consist of different factor-specific practical activities, for example, reflections during the intervention support metacognition, creation of experiential learning possibilities helps to increase openness to new experiences, positive new learning experiences enable the formation of attitudes, career planning activities support the formation of a future vision, participation in various initiatives of the distribution of roles during the intervention can improve the sense of responsibility and new unfamiliar situations during the intervention support resilience.

Measuring before the intervention makes it possible to evaluate the effectiveness of the intervention.

The SDL-NEET scale will continue to be a useful tool in the diagnosis of the SDL abilities of young people, especially those not working or studying. As the statements of the self-report scale do not address the experiences of NEET-youth in a social context, the use of the scale is suitable in different societies. However, attention should be paid to social norms, as they may influence both the concept of the NEET status and the individual’s opinions and attitudes. In their research on NEET-youth in welfare regimes, Jongbloed and Giret (2022) have also highlighted the significant impact of social norms.

The results of the research can be implemented in the creation of intervention tools or the general support process for NEET-youth, and also for other young people in order for young people to continue in education and the whole lifelong learning process.

Disclosure statement

No potential conflict of interest was reported by the authors.

Appendix 1:

SDL dimensions and factors (Authors, 2023)

Appendix 2.

6-factor and 34-item SDL-NEET questionnaire

Table

	Metacognition	Resilience	Attitude to learning	Future orientation	Responsibility	Openness to new experiences
I like to evaluate what I do	0.67
When I’m reading I sometimes stop and check my understanding.	0.60
I need to know why	0.56
It is important for me to understand the essence of things	0.54
I like to gather the facts before I make a decision	0.49
I want to learn more so that I can keep growing as a person	0.42
If learning is hindered, I would rather give up.		0.71
When I see something that I don’t understand, I stay away from it		0.69
I often have a problem motivating myself to learn		0.61
I do not manage my time well		0.49
I love to learn			-0.71
I have a need to learn			-0.69
Learning is fun			-0.68
Constant learning is a bore			-0.59
I´m looking forward to learning as long as I´m living			-0.55
Learning is a tool for life			-0.54
The more I learn, the more exciting the world becomes			-0.49
The people I admire most are always learning new things			-0.41
I know what I want to learn				0.47
If there is something I need to learn, I find a way to do so right away.				0.44
I try to relate what I am learning to my long-term goals				0.42
I prefer to plan my own learning					-0.72
I am responsible for my own decisions/actions					-0.69
I prefer to set my own learning goals					-0.68
I prefer to set my own goals					-0.67
I am responsible for my own learning					-0.64
I like to make decision for myself					-0.55
I am capable of learning for myself almost everything I might need to know					-0.50
I enjoy a challenge						-0.67
I like to try new things, even if I´m not sure how they will turn out						-0.52
When presented with a problem I cannot resolve, I will ask for assistance						-0.47
I am open to new ideas						-0.46
I think of problems as challenges, not stopsigns						-0.45
I´m happy with the way I investigate problems						-0.42

Appendix 3.

Factors and items of SDL-NEET compared with the theoretical model