Evaluation of the Life Skills of Students in Adolescence: Scale Development and Analysis

Adolescents have been the leading group for life skills (LS) education in the last decade. The primary purpose of this study was to develop a valid and reliable instrument for assessing the general LS of students in adolescence. In addition, the relation of adolescents' LS to gender, grade level, science/physics course scores, GPA, and socio-economic status variables was examined. This descriptive survey study included two data sets, one for the development of the scale part, including 692 students, and the other for confirmatory analysis and inferential statistics, including 887 students. LS were assessed using the Life Skills Scale (LSS), which consisted of 83 items in 10 sub-dimensions, and evaluated critical thinking, creative thinking, decision making and problem-solving, coping with stress and emotion, social responsibility, teamwork, self-esteem, self-awareness, empathy, and interpersonal relationship and communication skills of the students. The results of the exploratory and confirmatory factor analysis and the reliability coefficient of the scale provided solid statistical evidence, which was coherent with the 10-subdimension model of the scale. The findings revealed that the LSS is a valid and reliable tool for adolescents to evaluate their LS. Students scored highest on empathy, self-awareness, and self-esteem, whereas they scored lowest on teamwork, coping with stress & emotion, and critical thinking. There was a significant relationship between various sub-dimensions of LS and some variables such as GPA, grade level, gender, and science/physics course score. The results also indicated that the academically successful students were highly skilled in most LS sub-dimensions.


INTRODUCTION
The life skills (LS) concept was first introduced in a psychological consultation program in the mid-'60s.This program was about the application of the life skills consultation model on disadvantageous groups and described LS as "…the psychological and social skills for mastering the interrelated problems in living encountered in training, on the job, in the home, and the community."(Adkins, 1970).Since then, the concept has gained its place in the general culture, and there has been an increasing interest in LS programs (Adkins, 1984;Bailey & Deen, 2002;Chauhan, 2016;Choudhary & Rani, 2020).The scope of these programs varied and covered topics such as adolescents' problems, youth problems, disadvantageous groups, protection from epidemics, occupational problems, career development in business and industry, anti-poverty programs, marriage/separation/divorce problems, health, death, teacher education, nutrition, and sports (Adkins, 1984;Bailey & Deen, 2002;Botvin & Griffin, 2004;Choudhary & Rani, 2020;Ghasemian & Kumar, 2017;Gould & Carson, 2008;UNICEF, 2012;WHO, 1994).
In recent years, adolescents have been the leading group for LS education (Balsano, Phelps, Theokas, Lerner, & Lerner, 2009;Durlak, Weissberg, & Pachan, 2010;Population Council, 2013;UNICEF, 2012;WHO, 1994).Adolescents experience a period of changes, which include biological, cognitive, and socio-emotional changes, during their transition from childhood to adulthood (Santrock, 2012).Joseph (2018) states that the probability of facing different issues and problems increases during this period.Building LS in this term helps them understand themselves and makes them feel more comfortable dealing with the problems.
Assessing adolescents' LS is essential to understand better their usage level of these skills in their own lives and the ones they need to improve.Many recent studies regarding LS emphasized the absence of valid and reliable instruments to measure them (Chauhan, 2016;Erawen, 2010;Green, 2008;Kennedy, Pearson, Brett-Taylor, & Talreja, 2014).Particularly, it is evident that there is a need for measurement tools that cover all of the general LS (Erawen, 2010;Vranda, 2009).This study aimed to develop and validate a survey instrument that could adequately evaluate the LS of adolescents.1.1 Scope of Life Skills WHO (1997) defines LS as positive attitudes and skills which help a person effectively cope with the problems of daily life.Core LS, which encourages adolescents to be healthy and positive, can be listed as decision making, problem-solving, creative thinking, critical thinking, effective communication, interpersonal relation skills, selfawareness, empathy, coping with emotions, and coping with stress (WHO, 1997).Other emphases about the definition of LS are as follows: 'The necessary skills to satisfy daily life requirements' (Kennedy et al., 2014;Khalil, 2018); 'basic development units of human existence' (Ginter, 1999).
There are no defined categories concerning LS, but they include various skills necessary for a successful life (Danish, Forneris, Hodge, & Heke, 2004).According to Hendricks (1998), it is crucial to find out and define the target LS for young people.Therefore, the LS categories may change depending on the target audience and their needs (Bailey & Deen, 2002).For instance, LS programs' target skills and objectives are different for adolescents (Magnani et al., 2005;Vranda, 2009) and adults (Diehl, 2004;Helfrich, Aviles, Badiani, Walens, & Sabol, 2006).Hendricks (1998) organized the 4-H LS model and categorized the LS into four primary groups: Head, Heart, Hand, and Health.The studies referenced above and the other studies in the literature (Erawen, 2010; Partnership for 21 st Century Skills, 2015; Tan, 2018) examined the LS by using different categorizations, but the sub-components of these categories are mostly the same with the core LS listed by WHO (1997) (Erduran Avcı & Kamer, 2018).
A literature review on LS instruments, developed for assessing LS of different age groups and characteristics, revealed specific instruments.Some of these instruments were used to evaluate LS programs with designated/special purposes (Bailey & Deen, 2002;Cronin & Allen, 2017;Dunn & Arbuckle, 2003;Kadish et al., 2001;Kennedy et al., 2014).The studies used other instruments shared the same focus as the present study and targeted to evaluate the general LS of adolescents (Erawen, 2010;Green, 2008;Özmete, 2008;Prasad, 2018;Vranda, 2009).Vranda (2009) and Prasad's (2018) LS scale factors consisted of the ten core LS stated by WHO (1997).Vranda (2009) proposed a 115-item scale to evaluate the LS of adolescents aged 13 to 16, and Prasad (2018) created a 50-item scale to evaluate the life skills of 11 th and 12 th -grade senior secondary school students.Özmete (2008) developed a scale in their research, which was performed on high school 1 st -grade students, and aimed to determine the LS using personal development, health, family life, consumer educationfinancial planning, and career planning factors.Although these scale development studies (Özmete, 2008;Prasad, 2018;Vranda, 2009) included some statistical efforts, they lack the statistical analyses to provide decent evidence (a good statistical source data) for a valid and reliable scale.Green (2008) developed a 'Youth Life Skills Scale' to evaluate the life skills of young people between 11 and 16 with 112 items and five factors (communication, time management, coping skills, goal setting, and leadership).Even though Green (2008) provided detailed and solid statistical evidence about the scale factors, it did not cover all of the core LS defined by WHO (1997).It thus had limited content regarding the LS factors.Erawen's (2010) scale covered the adolescent group, included all core life skills categories (WHO, 1997), and provided solid statistical evidence regarding the scale's validity and reliability.This scale consisted of 120 items and had nine factors, including 'self-respect' and 'selfresponsibility' in addition to the core skills (critical thinking, creative thinking, self-awareness, empathy, interpersonal relations, and communication, decision making-problem solving, coping with stress, and emotions) defined by WHO (1997).The structure of Erawen's scale leveraged mixed methods research, which comprised qualitative (affinity diagram technique) and quantitative (model of confirmatory factor analysis) methods.The research provided reliable statistical data about the structure of the scale by analyzing the data collected from 1,305 high school students (aged between 12 and 17) in Thailand.Therefore, it can be claimed that out of all life skills scales in the literature, the scale developed by Erawen (2010) is a proper, valid, and reliable instrument to evaluate the core life skills of adolescents.
The scope of life skills may differ according to some cultures or countries (Koen & Ebrahim, 2013;WHO, 1997), such as Thailand (Erawen, 2010), Japan (Kobayashi et al., 2013) or India (Bhardwaj, 2013;Chauhan, 2016).Nasheeda, Abdullah, Krauss, and Ahmed (2019) stated, "Examining adolescent experiences within the embedded culture of the individual is important to understand how individuals from different backgrounds construct life skill knowledge into reality" (p.376).In this context, the present study aims to examine the LS development of adolescents in Turkey.One of the primary objectives of the Turkish education system is "growing individuals who are integrated to our values & competencies and possess required information, skills, and behavior" (TMNE, 2018a).One of the purposes of education is to make the students embrace and apply the fundamental values, such as effective communication, collaboration, respect, and responsibility.At the same time, they research and question a piece of information.Also, the national science course curriculum (TMNE, 2018a) and the physics course curriculum (TMNE, 2018b) in Turkey emphasize the importance of developing some life skills in students' responsibility regarding daily life problems.For instance, these six LS are emphasized in the science course curriculum: analytical thinking, decision making, creativity, entrepreneurship, communication, and teamwork (TMNE, 2018a).No valid and reliable LS scale can evaluate the development of the core LS proposed by WHO (1997), and the development of the LS emphasized by the Turkish national education system and the curricula (TMNE, 2018a(TMNE, , 2018b)).
The results of Alaca, Yaman, and Nas (2020) 's study showed that the pre-service science teachers in Turkey did not have the cognitive structures to produce enough answer words about life skills.Furthermore, Nasheeda et al. (2019) stated that most developing countries' LS programs lack evaluation and monitoring.Such research results strengthen the necessity of evaluating the LS of students.

The present study
To understand young people's skill development, it is crucial to establish valid and reliable instruments for LS (Cronin & Allen, 2017).The need for the tools to assess these skills is obvious, and this study has the potential to contribute to the literature by validating a new instrument, the Life Skill Scale (LSS), for adolescents.In some studies, the LS was examined regarding variables such as gender, grade level, and SES.Some studies investigated the relationship of the LS to gender, and a portion of them indicated that females scored higher than males (Kennedy et al., 2014;Kobayashi et al., 2013;Vranda, 2009).Another group stated that males scored higher than females (Ansari et al., 2016), and the remaining ones emphasized that there was no superiority in scores according to gender (Duerden et al., 2012;Ghasemian & Kumar, 2017).
Furthermore, Duerden et al. (2012) stated that the school level produced the most significant score differences across LS measures.They found that the LS scores of the primary school students were higher than secondary school and high school students except for communication and self-respect.The communication scores of the high school students were higher than the other students, and the self-respect scores of the secondary school students were higher than the other students.For the students' GPA, Cronin et al. (2021) and Currie et al. (2012) revealed that LS was associated with academic achievement scores in certain sub-dimensions such as goal setting, problem-solving, and leadership.Yayla Eskici and Özsevgeç (2019), in their study, examine the studies on the LS and underline that it is a research area that needs to be questioned according to socio-economic level.Therefore, the purpose of the study was to develop an instrument evaluating the general LS of adolescent students and to analyze them concerning certain variables.
The study's research questions were listed as follows: (1) Is the LSS developed in the present study valid and reliable according to the results of the exploratory and confirmatory factor analyses? (2) Are there any significant differences between the total and dimensions scores of the students' LS and the variables of GPA, gender, grade level, science/physics course score, and SES?

METHOD
This study was a descriptive survey study that aimed to develop an LSS for adolescent students-between the ages of 10 and 17 and to confirm and cross-validate the tenfactor structure of the developed LSS (Fraenkel, Wallen, & Hyun, 2012).Therefore, all of the procedures to support the construct validity of the LSS were justifications to validate the scale's ten-factor structure.Furthermore, LS sub-dimensions of secondary and high school students were examined in terms of different variables -namely GPA, gender, age, grade level, school type, science/physics course score, and SES.

Participants
The accessible population of the study was around 9,700 middle and high school students in Burdur, Turkey.Among this student population, we collected the research data from the two most crowded secondary schools (Secondary School A and B) and three high schools (High School A, B, and C).LSS was applied to 1,967 students, out of which 1,068 were secondary school students and 899 were high school students (approximately 20.3% of the population).At first, the 211 middle school and 177 high school students' data, including incomplete markings, were removed.Then, we used two data sets for the study: data set-1 (for EFA) and data set-2 (for CFA and the inferential statistics).We set the number of individual data for these two sets as close as possible.Data set-1 was created by combining the data from Secondary School A and High School B (692 students in total), and data set-2 was created by combining the data from Secondary School B and High School A and C (887 students in total).We combined data from the secondary school students with high school students because LSS covered the age group (ages 10 to 17) from both school types.The EFA, 13 students' data including outliers were removed from data set-1, was performed on the data obtained from 679 students [369 female, 54.3% and 310 male, 45.7% (secondary school students: 216 female and 196 male; high school students: 153 female and 114 male)].The CFA, 302 students' data including outliers were removed from data set-2, was carried out with the data of 585 students [360 female, 61.5% and 225 male 38.5% (secondary school students: 130 female and 120 male; high school students: 230 female and 105 male)].A sample size greater than 200 was considered adequate for CFA (Brown, 2012;Field, 2009).
The inferential statistics, 122 students' data including multivariate outliers were removed from data set-2, was carried out with 765 students [449 female, 58.7% and 316 male 41.3% (secondary school students: 198 female and 190 male; high school students: 251 female and 126 male)].In the sample of the inferential statistics part, the students' distribution of independent variables (Gender, GPA, science/physics course score, grade level, and SES) are presented in Table 1.

Variables
The mean scores obtained for the ten sub-dimensions of the LSS were the dependent, and students' gender, GPA, science/physics course score, grade level, and SES were the independent variables for the inferential statistics of this study.The properties of variables, and item numbers from which the variables were derived or taken, are presented in Table 2.In the regulation on secondary educational institutions (TMNE, 2020), the categorical transformations of the students' scores as below 49.99 'does not pass ', 50-59.99 'passes'; 60-69.99 'average', 70-84.99'good', above 85 'very good.The students' science/physics course score (SPCS), representing the overall score for the previous term, and their general GPA are transformed into categorical variables concerning this transformation.The SES variable also indicates a composite measure of the SES index.The SES score in this study was a combination of students' scores for answers to the family's seven demographic questions related to education, occupation, and income variables as described in the Turkish socioeconomic status index document from the Social Structure Studies Program (SSSP, n.d.).The item for mothers' and fathers' education status was scored between 1 and 5, occupation status was scored between 0 and 1, total income was scored between 1 and 5, the number of people in the family was scored between 1 and 5, and ownership of their houses was scored between 1 and 3.The total SES score varies between a minimum of 7 and a maximum of 23 points.Its mean is 15.1, and the standard deviation is 3.1.The distribution of calculated SES scores was then divided into tertiles according to the percentage distribution of Turkish SES groups in the SSSP (n.d.).Therefore the scores between 7-14 are defined as low, between 15 and 18 are defined as medium, and between 19 and 23 are defined as high SES (see Table 1).

Data analysis
The LSS was applied by including 14 demographic questions on the scale.The demographic questions were used to attain the independent variables of the analysis described in Table 2.In the LSS, higher scores for students' answers to five-point Likert-type items were interpreted as a higher perception of LS.
Before EFA, CFA, and inferential statistics calculations, the data were analyzed for accuracy, missing data, outliers, normality, multiple accuracy, and singularity.Since missing values lower than 5% could be replaced by any method that handles the missing values, according to Tabachnick and Fidell (2007), they were replaced with the means of series.All outliers detected for single and multiple variables were removed from the data set.The multicollinearity within the items inspected with tolerance and VIF was not observed.
The shape of the scatter-plot diagrams was also checked for homoscedasticity, and this assumption was also validated.For the normality, as Byrne ( 2010) suggested, at first univariate normality (skewness and kurtosis values were within the range of -2 and +2) was validated.The CFA was conducted by running the baseline model; only the factor loadings are constrained to be equal across the data for the original 10-factor structure.The common fit indices are given in Table 3, together with their critical value ranges.
In addition to the values in Table 3, Hu and Bentler (1999) indicated a combination of rules for a reasonably good fit between the model and the source data to increase the acceptability of Type I and Type II error rates.a) SRMR value close to or below .08;b) RMSEA value close to or less than .06,and c) CFI and TLI values close to or greater than 0.95.There is no apparent difference between the combined values given in Table 3 and the critical limits of this study.Therefore, the statistically significant χ2 is not sufficient to imply that the data is incompatible with the model.In such a case, model fit can be evaluated by comparing the value of χ2 to twice the degree of freedom  (Byrne, 1989), analyzing critical limits for the value of χ2 / df, and analyzing the other fit indices.The dependent variables were the scores assigned through data imputation from AMOS for the ten subdimensions of the LSS, as explained in Table 5.The assumptions of the MANOVA were checked that the observations were independent and the sample size was sufficiently large for the groups of MANOVA.The assumptions of the absence of multiple variable outliers, the normal distribution of the dependent variables for each independent variable, linearity, multicollinearity, homogeneity of variances, and homogeneity of variable matrices were also met.The age and type of the schools were not included in the further analysis since they do not meet the assumptions of homogeneity of normality and variance.

Framework for development of the LSS and results for the EFA and CFA
The five-stage model for scale development proposed by Hinkin (1998) was accepted as the framework.These stages were item generation, scale management, initial item reduction, confirmatory factor analysis, scale evaluation, and convergent/discriminant validity.The five-stage process stated by Hinkin was analyzed within the five subtitles of this section.
In this context, researchers formed a table including definitions, variables, and indicators of sub-dimensions by mostly revising Erawen's (2010) study (see Supplementary Material Table A).The researchers formed an initial item pool (including 189 items) by accepting and/or revising the items of the other scales in the literature and creating new ones (see Supplementary Material Table B).Six faculty members who were experts in educational fields discussed these items in three sessions.Note: χ2 = chi-square, RMSEA = root mean square error of approximation, RMR = the root mean square residual, SRMR = the standardized root mean square residual, GFI = goodness of fit, CFI = comparative fit index, and IFI = incremental fit index.

Scale management
Items were selected according to the comments of the experts.The researchers and experts agreed on the draft LSS containing 140 items in 10 dimensions.A pilot study was conducted on 40 students aged between 12 and 15.The pilot study aimed to collect the students' opinions about the scale items' clarity and the application duration.In addition, the scale was presented to 13 science teachers for evaluation.One of the special purposes and learning fields of the Turkish Science Course curriculum (TMNE, 2018a) is LS.The science teachers in Turkey are experienced and knowledgeable in making the students gain these skills.Therefore, the researchers consulted experienced science teachers to obtain their assessment of the draft LSS.The teachers' opinions were evaluated together with the application results in the pilot study, and four items were removed from the draft LSS.

Initial item reduction: Providing construct validity
After the first application of the LSS, construct validity of the LSS was carried out with explanatory factor analysis (EFA) by principal component analysis with the extraction method and varimax with Kaiser normalization as the rotation method.The KMO value was found to be .957,accepted as 'very good' for extracting factors from the collected data (Field, 2009).Bartlett's sphericity test result was χ² = 27350.787,p < .001.Therefore, this sample size was adequate for factor analysis (Field, 2009;Henson & Roberts, 2006).There were 136 items on the scale for the first application of the study.The researchers considered three criteria in item reduction: (1) The factor loadings of the items should be above the critical limit of .320(Tabachnick & Fidell, 2007), and (2) The Cronbach's alpha reliability of the scale should remain as "very good" (> .90)after the removal of the item, and (3) there should be another item that measures the same skill with the removed item, so the removal should not result in any measuring deficiency regarding the skill.For this process, the researchers carried out a factor analysis and analyzed these three criteria five separate times, repeatedly (16 in the 1 st cycle, 13 in the 2 nd cycle, 9 in the 3 rd cycle, 9 in the 4 th cycle, and 6 in the 5 th cycle, in total 53 items were removed).Then, principal components analysis was performed with 83 items during the EFA process.Since the scree plot advised that the factors after 11 th and 12 th had very close values, the scale was considered a 10-factor scale.
In the 'Total Variance Explained' table of the SPSS, ten factors accounted for 51.07% of the variance, which is acceptable (Henson & Roberts, 2006), with eigenvalues of 1.0 or higher.Furthermore, the items in these ten factors had loadings between .326 and .805.Therefore, the final factor groups and factor loadings for the EFA are given in Table 4.
The factors were named as given in Table 5 using the category names from the literature (Erawen, 2010).The internal consistency of the ten factors was analyzed by Cronbach's α values as given in Table 5 (ranging between .717 and .916).Therefore, the internal consistency of the whole test ( = .964)was considered acceptable, and the results of the LSS are reliable (Pallant, 2007).
In the LSS, the score for students' answers in five-point Likert type items was assigned the following codes: 1 'Totally disagree', 2 'Disagree', 3 'Neutral', 4 'Agree', 5 'Totally agree'.Higher scores are interpreted higher perception of LS.Table 2 represents the mean scores of the participants' perceived LS concerning sub-dimensions obtained from the LSS.The first three sub-dimensions with the highest mean scores in descending order are empathy, self-awareness, and self-esteem, and the subdimension with the lowest mean score is teamwork.

Confirmatory factor analysis (CFA)
CFA with AMOS was performed in order to determine the model-fit whether the LSS derived from EFA.Before conducting CFA, data from 826 students were analyzed for outliers using Mahalanobis distance at p < .001(Tabachnick & Fidell, 2007, p. 99).The outliers were removed, and the CFA was conducted with 585 students' data.The related fit indexes (CFI, GFI, and IFI) were within the accepted thresholds in Table 3.All fit indices were observed to be within the acceptable value range (χ2(3268, N = 585) = 5953.186;χ2/df = 1.822;RMSEA = .0038,SRMR = .049;CFI = 0.900; IFI = 0.901), which corresponded to a fair fit.Therefore, the LSS corresponds to the same 10-factor structure model over the new student group in the Burdur sample as it adequately fits the model reached in the EFA.The path diagram of the 10-factor structure is presented in Figure 1, and the factor loading values are presented in Table 4.

Convergent/discriminant validity
At this phase, as stated by Hinkin (1998), reporting the discriminant, convergent, and concurrent validity clearly can prove the validity of theoretically justified relation between the variables.The construct validity was also controlled by convergent and discriminant validity.The Figure 1 The path diagram of the ten-factor structure of the LSS factor loadings were above the critical value of .50 for CFA (see Table 1), ranging from .354 to .839within the tenfactor structure.Regarding the ten factors, the computed average variance extracted (AVE) values ranged between .385 and .694and was calculated as .457for the whole scale.Moreover, the composite reliability (CR) was calculated for the ten factors ranging between .752 and .942,and the CR was .986for the whole scale.The AVE is higher than .400,which is acceptable when the CR is higher than .600(Fornell & Larcker, 1981).This is an indicator of acceptable convergent validity (Ullman, 2007).Furthermore, the factors measured different concepts, and the intercorrelation coefficients among them varied between .584 and .758.These values, which were lower than the critical threshold of .800,supported the discriminant validity (Brown, 2012).
The reliability coefficient (Cronbach's alpha) of the scale concerning the whole was calculated as  = .964.Table 5 shows the reliability coefficients for the ten factors calculated from the data collected for the study's EFA, CFA, and inferential statistics.The values indicated high reliability of the test results (Fraenkel et al., 2012).
When MANOVA results are examined in terms of SES of the students, the differences between the mean scores obtained for the sub-dimensions of the LSS were not statistically significa nt [F(20, 1508) = 1.212, p = .234;Wilk's Λ = 0.969, partial η 2 = 0.016].

DISCUSSION
One of the main objectives of this study was to develop an LSS for adolescents.Therefore, different stages of this study provided evidence for the structural validity of the LSS and the reliability of the results obtained from adolescents aged between 10 and 17.These stages were conducted as follows: (i) 140 items in the draft scale were selected by a team of experts from an item pool that was built after a comprehensive literature survey, and (ii) The structural, semantic, and conceptual equivalence of the scale items were developed by consulting to 6 literature and educations experts.13 science teachers inspected the validity of the first version of the scale.The clarity and the application time duration of the items were controlled during the pre-application on 40 students, (iii) While increasing the coefficients of factor loadings on the subdimensions, EFA was carried out repeatedly to keep the internal consistency of the items at a good level, and the items were removed under the supervision of the experts.
After repeated EFAs, an 83-item scale with 10 dimensions was obtained, (iv) The CFA, which was performed on a different group, confirmed the proposed structure and provided evidence for convergent and discriminative validity, (v) The internal consistency of the scale items were confirmed by Cronbach's Alpha coefficient for both EFA and CFA data.These analyses revealed that the reliability coefficients were high for the scale and the subdimensions, indicating a high degree of internal consistency.It can be stated that the well-planned and systematic aspects of the five-stage scale development process proposed by Hinkin (1998) proved effective in obtaining valid and reliable results.The study put forth evidence for the factorial validity, convergent validity, construct validity, discriminant validity, predictive validity, test-retest reliability, and internal consistency reliability of the LSS (Brown, 2012;Field, 2009;Tabachnick & Fidell, 2007;Ullman, 2007).The factor loadings in the previous studies (e.g., Cronin & Allen, 2017;Erawen, 2010) were calculated between .39 and .90 for both EFA and CFA, similar to the present study.The internal consistency reliability coefficients of the present study were higher than .70 and at a good level (Pallant, 2007), similar to studies by Cronin and Allen (2017) and Erawen (2010).In this context, this study proved that the LSS was a valid and reliable scale for adolescents, which evaluated critical thinking, creative thinking, decision making and problemsolving, coping with stress and emotion, social responsibility, teamwork, self-esteem, self-awareness, and empathy, interpersonal relationship, and communication.So, LSS could be used in studies aiming to determine adolescents' life skills.In addition, the inferential statistics part of the study revealed the relationship of adolescents' LS with some variables.They provided support for the discussion on the development of their LS.
The LS scores of the adolescent were relatively high in all three parts of the study (EFA part, CFA part, and inferential statistics part), and the score rankings of the subdimensions were similar.These parts indicated that they scored highest on empathy, self-awareness, and selfesteem, whereas they scored lowest on teamwork, coping with stress & emotion, and critical thinking.There are both similar and different results in the literature regarding these parts.For instance, Vranda (2009) found that the LS scores of the adolescents were average, the highest scores were on interpersonal relationships, and the lowest was on coping with stress.In Erawen's (2010) study with high school students in Taiwan, the highest scores for the high school students were on decision making and problem-solving, interpersonal relation and communication, and social responsibility; where the lowest scores were on creative thinking, critical thinking, coping with stress, and emotion.In studies conducted with university students, different results were obtained regarding the highest and lowest LS (Cronin et al., 2021;Ansari et al., 2016).The different  (Çardak, 2013;Santrock, 2012;Slavin, 2013;Statistics Canada, 2005), school curricula (Yuen et al., 2010), geographic region/country (Statistics Canada, 2005), or cultural differences (Santrock, 2012;WHO, 1997).Based on our cross-sectional findings, future research may examine how and why the students' LS development changes regarding other variables such as age, culture, development levels of the countries, curriculum, teacher education policies, parent relations, and so on.
According to the results from the inferential statistics, there were statistically significant differences between the mean scores obtained for some sub-dimensions of the LSS in terms of gender, grade level, GPA, and science/physics course score.However, no statistically significant differences existed between the LSS scores and SES.Looking at the results in terms of gender, the significant difference was found to favor females for empathy and the males for coping with stress and emotions.Santrock (2012) indicated that stress was prevalent in adolescence, females and males experienced stress in various types, and there were no differences regarding gender.However, females experienced more stress in peer relations.Although some findings in the literature generalized some of the skills according to gender, it was considered essential to handle the gender roles according to the culture and the context of the behavior (Santrock, 2012).Therefore, the studies in the literature indicate that there is a need for more studies that are supported by dimensional qualitative & quantitative findings and examine the differentiation and development of LS of adolescents according to gender roles.
Our results indicated statistically significant differences in self-esteem, self-awareness, coping with stress and emotions in favor of 5 th graders for other grade levels, and creative thinking in favor of 5 th graders concerning 6 th graders.Adolescence is a period of socio-emotional changes and is the transition from childhood to adulthood (Çardak, 2013;Santrock, 2012).The changes are quite fast in the first period of this term (ages 10-13) and slow down in the middle and last periods (ages 14-20) (Çardak, 2013, pp. 62-64).It is claimed that self-concept, defined as the view, attitude, and emotional perceptions of a person about themselves, differs more in adolescence (Terzi, 2013).Slavin (2013) stated that self-concept and self-respect changed during the transition into adolescence and during adolescence, where self-respect was at the lowest degree at the beginning of puberty, namely when children started secondary school or high school.In this study, the statistically significant difference for 5 th graders (ages 10-11) in self-respect and self-awareness indicated that these students were not adolescents yet, or could be at the beginning of pre-adolescence.
In Turkey, 5 th grade is the start of secondary school, and 9 th grade is the start of high school.The 5 th graders had the highest scores in coping with stress and emotions, and there was a general decline in the scores of this subdimension as the grade levels increased.One of the reasons for this could be related to the processes in adolescence.According to Slavin (2013), individuals are more optimistic and happier before adolescence.However, academic achievement becomes a more serious issue, the academic struggle increases, and young adolescents may experience more emotional reflections compared to childhood.Another reason might be the negative emotional conditions caused by central examinations, which are conducted at the end of secondary school and high school in Turkey (Atlı Özbaş, Sayın, & Coşar, 2012;Bacanlı & Sürücü, 2006;Ünalan, Çifçili, Dinç, Akman, & Topçuoğlu, 2017).In order to develop adolescents' ability to cope with stress and emotions, it can focus on studies in cooperation with guidance and psychological counseling centers, teachers, school administrators, and parents, and the longterm effects of these studies can be examined.
The results of our study revealed significant differences between most of the LS sub-dimensions in favor of the students with both high GPA and high science/physics course scores.These results indicated that the academically successful students were also highly skilled in most LS subdimensions.Cronin et al. (2021) stated that goal setting, time management, leadership, and total LS were positively associated with higher-education students' predicted academic grades.Similarly, Currie et al. (2012) stated that four life-skills categories (interpersonal communication, problem-solving, physical fitness, identity development) predicted an additional 9.4% variance in cumulative GPA beyond high school GPA and SAT scores.In addition, Currie et al. (2012) suggested that the ability to handle the associated stress would be a good predictor of initial academic success for students starting university.Chien et al. (2012) emphasized considerable evidence that the skills that predict success in elementary school also predicted social and academic success in adolescence and adulthood.Future research may focus on learning strategies that affect students' academic performance in classroom settings.
The present study showed no statistically significant differences between the LSS and SES.An international research report on adult skills indicated that for most countries, skills significantly decreased one's chances of earning less than half the median earnings (OECD, Statistics Canada, 2011).In addition, according to the results of Cronin et al. (2021), the LS of the students positively related to their social, physical, emotional, and school/work functions.
Like any study, there were several limitations to the present study.First, any action or instrument to assess the LS has its boundaries or constraints (Jacobs Foundation, 2011).In this study, these effects tried to reduce by (i) conducting pilot studies, (ii) providing enough time to the participants, (ii) telling them to make their markings after reading all of the items, and (iii) providing researcher support during the application, and (iv) conducting outlier analyses on the obtained data.Second, although the intersectional quality of our research provided statistically significant evidence regarding the differences between the LSS sub-dimensions and the variables, they were not enough for a relation of causality.'Life skills' should not be treated as isolated but as a composition of manual skills and psycho-social abilities (UNESCO, 2004).Third, future research can evaluate the LSS in adolescents from other countries/cultures or perform inter-cultural comparisons.Fourth, this study investigated the relation of a limited set of variables with the LSS.More research can be conducted on different variables associated with adolescents' LS.

CONCLUSION
This scale development study put forth two main conclusions.The first is that the LSS is a valid and reliable scale to determine adolescents' general life skills.The scale's CFA and EFA results, together with the convergent/discriminant validity and reliability coefficients (Cronbach's alpha), provided strong statistical evidence coherent with the 10-subdimension model.In this context, the LSS, with its 10 sub-dimensions covering 83 items, can be used as a valid and reliable scale to diagnose complex psychological concepts such as the students' perception of LS and identify their skills.
The second result is that this study provided evidence about how the life skills of the adolescents were related to several variables, such as grade level, science/physics course scores, GPA, and SES.According to the results of the present study, there were statistically significant differences between the average scores for some of the life skill sub-dimensions of the LSS and gender, grade level, science/physics course grades, and GPA variables.However, no statistically significant difference was found for the SES variable.

Table 1
Number of students and their percentages in terms of independent variables

Table 2
Description of the variables

Table 3
Fit indices with suitable thresholds used in this study

Table 4
Factor loadings of items within the EFA and CFA of the study

Table 5
The number of items, mean scores, standard deviations, and alpha coefficients of the study

Table 6
MANOVA post hoc findings for grade level concerning the sub-dimensions

Table 7
MANOVA post hoc findings for science/physics course score for the sub-dimensions

Table 8
MANOVA post hoc findings for GPA concerning the sub-dimensions