The purpose of the research was to investigate the quality of C.V.O., in Secondary Vocational Education structures and Higher Vocational Training Schools. In more detail, teachers, students, graduates, and adult students were asked the research question regarding: “the degree of satisfaction with S.V.G. provided by the school”.

Teachers, students and graduates of vocational education and adult students from the prefectures of Attica and Cyclades participated in the survey.

The final sample consisted of a representative number of participants (men and women of different ages, geographic regions, specialties, etc.), to ensure the reliability and validity of the research. In total, 382 teachers, 594 students, 180 graduates, and 344 adult students participated in the survey.

During the sampling design, it was decided to apply mixed sampling, as the most appropriate method, for a more representative sample of the population (teachers, pupils, and adult students) ( Sharp et al., 2011 ; Collins et al., 2007 ), while for the graduates, simple random sampling was applied.

Subsets of the population were selected that met specific properties ( Panaretos & Xekalaki, 2000 ). For this reason, stratified sampling was used, to ensure a representative sample of each part of the population, to deal with the specificity of the population’s geographical distribution (existence of many small islands, some of which have absorbed part of the immigration wave through their schools) and to reduce the estimation errors ( Papoulis, 1966 ; Halikias, Lalou, & Manolesou, 2015 ).

Cluster sampling followed because the hierarchical sampling plan starts from the educational regions of the country, continues in the school units of each region and ends in the classrooms of the schools ( Papageorgiou, 2015 ).

The sampling design was completed by applying simple random sampling between each cluster as defined in cluster sampling, to select the final sample ( Turk & Borkowski, 1990 ).

The educational regions of Greece are 52 ( Ministry of Education, 2024 ). We chose the prefectures of Attica and Cyclades for the research (stratified sampling). Each constitutes a different stratum (layer) of research. The particularity of this study was the participation of populations from urban and semi-urban (island) areas. According to the sampling design, within the strata (layers), there was as little dispersion as possible (high homogeneity) but between them, as much as possible (heterogeneity) ( Halikias, Lalou, & Manolesou, 2015 ).

The stratified sampling design neither impose restrictions on how to sample from stratum to stratum, nor on the sample size of each stratum ( Papageorgiou, 2015 ).

Then, the simple random cluster sample of multiple stages, of equal size, was applied ( Panaretos & Xekalaki, 2000 ; Kamalu, Niyomwungere, & İlker, 2024 ). The term “random” does not mean that we selected whoever we wanted from the population. The sampling was random because in its various stages, the composite units were selected by simple random sampling, i.e. by lottery ( Halikias, Lalou, & Manolesou, 2015 ).

The researchers had direct access to the sample of the population, however, the members of the population were approached indirectly, through larger groups (clusters) to which each member belongs.

Equal-sized sample of the survey populations was selected. The proportional division between samples and populations was not applied because the expected samples (teachers, pupils, and adult students) for the population of the Cyclades were too small to draw conclusions and complete the statistical analyses ( Zafeiropoulos, 2010 ).

Cluster sampling involves several stages ( Farmakis, 2015 ). The design of the sample survey included four stages:

• 1^st stage, with simple random sampling, the selection of the primary sampling unit was carried out. From the eight regions of the prefecture of Attica, three were selected, the region of C’ Athens, D’ Athens, and A’ Piraeus. From the four regions of the prefecture of Cyclades, three were selected, the region of B’ Cyclades, C’ Cyclades, D’ Cyclades.

• 2^nd stage, with simple random sampling, the secondary sampling unit was selected, which concerned the municipalities that belong to these regions of the prefecture of Attica and the islands that belong to the regions of the prefecture of Cyclades.

• 3^rd stage, with simple random sampling, the tertiary sampling unit was selected, which concerned the schools belonging to the selected municipalities and islands.

From the prefecture of Attica the V.H.S. of Haidari, the V.H.S. of Renti and the V.H.S. of Sivitanidio school were selected. From the prefecture of Cyclades, the V.H.S. of Paros island, the V.H.S. of Naxos island and the V.H.S. of Thira island were selected.

According to the sampling design of cluster sampling, between two clusters there must be as little dispersion as possible, i.e. great homogeneity of the group, but within the clusters greater dispersion ( Halikias, Lalou, & Manolesou, 2015 ).

This condition is satisfied because in V.H.S. (e.g. of Naxos) students from all over the island attend. Some of them are engaged in tourism and belong to the most affluent groups of the population, while others are engaged in agriculture and belong to economically disadvantaged groups. Generally, V.H.S. are large schools, each municipality has a small number of schools, therefore, there is a large dispersion (heterogeneity) among the population. It is stated that a cluster with heterogeneity implies a representative sample of the population and therefore a more accurate estimate of the population.

• 4^th stage, the quaternary sampling unit was selected, which involved a section from each class of the school. This section was selected by simple random sampling. These departments were the sample of the research. The selection of the sample was carried out at the group level and not at the level of individual members of the population ( Papageorgiou, 2015 ). In this phase, a census of the members of the final clusters was carried out and not a sampling of individual members.

The sample number of students was determined in advance. A sample was selected, representative of the population corresponding to 300 students from the prefecture of Attica and 300 students from the prefecture of Cyclades. Specifically, an attempt was made to collect 33 questionnaires from each class (A, B, C) of each school. In total, 297 student questionnaires were collected per cluster. This number corresponded to 3.4% of the total population of Attica and 11.9% of the population of the Cyclades ( Greek Statistical Authority, 2024 ). A total of 594 students participated in the survey. There is no absolute rule stating the exact percentage of each sample (e.g. data must be collected from 10% of the population).

The survey sample size was closely related to ( Gray, 2023 ; Rahman, 2023 ): 1) the research question, 2) the population (in terms of size and characteristics), 3) the sampling method, 4) the sampling error, 5) the level of confidence defined, 6) the techniques of data analysis. For the reliability of the research, the number of 600 students was considered sufficient to reduce errors.

At each stage of sampling, simple random sampling is applied, so we can use its formula to estimate the maximum error. In the formula:

$e=\frac{1}{\sqrt{n}}$,

where: “n” is the sample size we used. A sample size of n = 594 students corresponding to:

$e=\frac{1}{\sqrt{n}}=\frac{1}{\sqrt{594}}=0.041$

Therefore, the maximum error equals 4.1% at the 95% significance level. A similar sampling design was applied for all the other participants.

The sample number of teachers was determined in advance. A sample was selected, corresponding to 200 teachers from the prefecture of Attica and 200 from Cyclades. The survey was not limited to the teachers of the schools designated in the third stage of sampling, in order to have a representative sample of the population.

A total of 3.347 V.H.S. teachers are employed in the prefectures of Attica and Cyclades ( Greek Statistical Authority, 2024 ). 382 teacher questionnaires were collected. This number corresponded to 9.5% of the total population.

The sample number of adult students was determined in advance. A sample was selected, corresponding to 300 adult students from the prefecture of Attica and 100 from Cyclades. In the third stage of sampling, the tertiary sampling unit was selected, which concerned the S.H.V.T. who participated in the research.

From the prefecture of Attica, S.H.V.T. of Peristeri, Nea Ionia and Sivitanidio school were selected. From the prefecture of Cyclades the S.H.V.T. of Naxos island was chosen.

An effort was made to have adult students from all majors participate. For this reason, the participation of 20 adult students per specialty was sought. A total of 3.960 S.H.V.T. adult students study in the prefectures of Attica and Cyclades ( Edu.klimaka.gr, 2022 ). 344 questionnaires were collected. This number corresponded to 8.68 % of the total population.

Regarding vocational education graduates, there were limitations in the selection of the sample since there was no direct contact in the population. As a result, simple random sampling was applied, despite the disadvantages of the method. The participation of as many graduates as possible was sought. After permission, questionnaires were sent to their personal emails and a total of 180 questionnaires were collected. In the prefectures of Attica and Cyclades, a total of 8.531 students graduated ( Greek Statistical Authority, 2024 ). 180 graduate questionnaires were collected. This number corresponded to 2.44% of the total population.

The survey was conducted from 04/2019 to 04/2021. During the study, research teams were asked to complete appropriately structured 30-question questionnaires. The questionnaires contained common questions in order to be able to make comparisons between populations and to understand the degree of satisfaction with S.V.G. provided by the school. In writing the questionnaires, clear closed-ended questions were used, using simple language and having appropriate explanations ( Papanastasiou & Papanastasiou, 2014 ).

In terms of structure, the questionnaires consisted of three parts:

• 1^st: Demographics.

• 2^nd: Satisfaction questions regarding the quality of education provided by the school.

• 3^rd: Satisfaction questions around S.V.G.

Responses were given on a ten-point Likert scale, graded with “1” corresponding to “not at all satisfied” and “10” corresponding to “completely satisfied”. The questionnaires took about 15 minutes to complete. Where necessary, the Likert scale was converted into a three-point scale with values from “1 - 4” corresponding to “not at all satisfied”, “5 - 6” to “satisfied” and “7 - 10” to “completely satisfied”. The questionnaires in the pilot phase were distributed to 10 students and 10 teachers. They were answered in the presence of the researcher after the necessary clarifications were made. Data were collected either directly in the Lime Survey or in paper form.

Data were analyzed using SPSS 29.0 (academic license). The analysis includes:

1) Descriptive statistical analysis to calculate mean and percentages of variables ( Vetter, 2017 ).

2) Checking the reliability of the measurement scale with the reliability index Cronbach’s Alpha ( Iseris, 2016 ).

3) Chi-square test to check the correlation of the variables with the variable concerning the research question. The test was carried out with the correlation coefficient Kendall’s Tau b ( Emerson, 2017 ).

4) Normality check using Kolmogorov-Smirnov and Shapiro-Wilk test ( Heliotis, 2015 ; Sedgwick, 2010 ).

5) Two-way ANOVA. It was checked if the independent variables (teachers, pupils, graduates, and adult students), affect the dependent variables (quality indicators) concerning the research question ( Hitti et al., 2022 ).

St. Post Hoc Test with (L.S.D.) to check between which population groups there was a statistically significant difference in responses ( Papakonstantinou, 2024 ).

Methodological limitations related to sample size, lack of prior research on the topic, as well as time and bias limitations were presented ( Dhanya, 2023 ). The ideal would be to collect as many questionnaires as possible from all the population groups of the survey. To graduates who could not directly access the population, 700 questionnaires were sent by e-mail but only 180 people fully participated in the survey. An attempt was made to apply a proportional division between the strata of research for teachers, pupils and adult students. It was found that the numbers of the samples concerning the prefecture of Cyclades were so small that their opinions were not heard compared to the opinions of all those of Attica. In addition, several statistical tests required a larger sample size to ensure the reliability of the research. During the literature review, there was a lack of topics related to C.V.O., and how it works among the four different population groups, so the results could not be compared with the results from the international literature. The time constraints that existed did not concern the researchers or the education structures, but all the education systems worldwide. The research started shortly before the onset of the pandemic (suspension of schools) and ended with the reopening of these structures. In addition, there was a bias in the answers given because there is an impression for years that the S.V.G. is not taught in schools as it should be ( O.E.C.D., 2004 ).

The results of the descriptive analysis concerning the values of the sample means are gathered in Table 1 . The participants’ satisfaction rates are presented in Table 2 .

Table 1 is structured in such a way that the reader can easily check and compare the average values of the participants’ opinions (satisfaction scores) of the research groups.

Continued

The last two columns in Table 1 are for Two-Way ANOVA to check:

1) The coefficient of importance (Sig.) regarding the effect of the independent variables (teachers, pupils, graduates, adult students) on the dependent ones (quality indicators of the S.V.G.).

2) The coefficient of significance (L.S.D. Sig.) concerning the statistically significant difference of the mean values of the opinions of the research groups.

In the last column where: “T: Teachers”, “S: Students”, “G: Graduates”, “A.S: Adult Students”. Therefore, when the table shows the notation: “T-S” then there was a statistically significant difference in the average values between Teachers and Students. It applies to the remaining observations.

The reliability coefficient (Cronbach’s alpha) was determined in order to check if the questions present a high coherence among themselves ( Markos, 2012 ). Table 3 shows that the reliability indices were high for all categories of respondents and for all axes of questions. Cronbach’s alpha values range from 0.732 to 0.958. High values of the index mean that the questionnaires fulfill the purpose for which they were constructed.

It was carried out to check the correlation of the quality indicators with the variable concerning the research question (Satisfaction with the S.V.G. provided by the school). The values of the Chi-square test for the quality indicators concerning: 1) the degree of satisfaction with the S.V.G. provided by the school (S7) and 2) the contribution of the S.V.G. to the choice of profession (S2) are presented as an example in Table 4 while the values for all other indicators are gathered in Table 5 .

Continued

Continued

The majority of the research groups were satisfied with the S.V.G. provided by the school and the information they received. At the same time, all those who were satisfied with the S.V.G. expressed the opinion that it helped them in choosing a profession. Kendall’s tau-b correlation coefficient showed a high correlation between quality indicators and research groups. Responses are statistically significant at a significance level of p < 0.001. By applying an additional 22 similar tables, Table 5 was created for the remaining questions concerning the quality of C.V.O.

The test was carried out using the Kolmogorov-Smirnov and the Shapiro-Wilk test. It was found that most variables follow the normal distribution with p > 0.05. According to the Central Limit Theorem, if the random variables (X₁, X₂, …, X_n) are independent and equal to Ε(Χ_i) = µ, V(X_i) = σ² < ∞, i = 1, 2, …, n, then, the random variable asymptotically follows for large n (typically equal to or greater than 30) the standard normal distribution N (0, 1).

${\rm Z}=\frac{{\displaystyle \underset{i=1}{\overset{n}{\sum }}{X}_i}-n\cdot \mu }{\sqrt{n\cdot {\sigma }^2}}$

Therefore, in our research, we can consider that the rest of the distributions also approach the normal ones.

It was checked whether the independent variables (teachers, students, graduates and adult students), affect the dependent variables (quality indicators) concerning the research question ( Table 1 ). The results of the Two-Way ANOVA (Tests of Between-Subjects Effects) are presented in the column (Sig.) of Table 1 . The following is an interpretation of the variable VAR S2, which concerns the S.V.G., and its contribution to the choice of profession. The level of study (students, adult students), the age (students, graduates), and the status of the participants (teachers, students, graduates) influence the opinion that S.V.G. contributes to the choice of profession. The effect was statistically significant (p < 0.001) ( Table 1 ).

St. Post Hoc Test and use of Least Significant Difference (L.S.D.)

It was tested whether the differences in the mean values of the views of the four independent populations were statistically significant. Control assumes normality and common dispersion of the populations.

Kolmogorov-Smirnov and Shapiro-Wilk tests were used for the normality of the distributions. The distributions approximate normal. A small deviation from normality does not significantly affect the use of the ANOVA test and does not affect the type I error rates ( Sedgwick, 2010 ).

Levene’s test was performed to determine the equality of the population’s dispersion. The non-significance of Levene’s test (p > 0.05) led to the conclusion that the variances were homogeneous and the ANOVA analysis can be used ( Siomkos & Vassilikopoulou, 2005 ).

Comparison of the means of two independent populations.

For the variable concerning teachers’ knowledge in S.V.G., (VAR S1), p < 0.001 applies. Therefore, we reject the null hypothesis.

The sample means of the populations differ significantly were a difference between: 1) teachers, they rated their knowledge in S.V.G. with (6.79/10) and students, they rated the knowledge of teachers with (5.89/10), 2) students and graduates, the former rated the knowledge of teachers with (5.98/10) while the latter with (6.68/10), Table 1 . A similar analysis followed for the remaining variables.