Participants

The population of the research is composed of 410 principals. Therefore, as the population of the research were reached, there was no need for sampling (Gay and Airasian, 2000; Burke and Christensen, 2004; Cohen et al., 2007).

Of all the participant principals, 158 (55.2%) were from a primary school, 88 (30.8%) were from secondary schools, 11 (3.8%) were from district directors of national education,  11 (3.8%) of them were district deputy principals of national education and 18 participants (6.3%) were at other positions. Of all the participants, 96 (33.6%) principals were in social branch and 188 of them (66.2%) were in quantitative branch. The average age of the participants is 46 years and 5 months; the avarage of years spent as principal is 13 years and 6 moths; their total length of service is 23 years and 3 months; the length of being principal in current institution is 3 years and 8 months on average.

Data collection tools

In this research, the “Loneliness at Work Scale” (LAWS) which was developed by Wright et al. (2006) for measuring loneliness at work; and the “Employee Performance Scale” (EPS) consisting of 8 items to measure the individual performance of employee which was firstly used by Kirkman and Rosen (1999) and later by Pierson and Sigler (2000) were used. EPS scale was adapted to Turkish by Çol (2008), consisting of 4 items.

Loneliness at Work Scale (LAWS)

The “Loneliness at Work Scale” is a Likert-type measurement tool consisting of 16 items in total. The orginal form of LAWS is a 7-point Likert type, but for the adaptation to Turkish, 7-point grading is not significant. Consequently, for similar researches in Turkish, 5-point Likert type  scales  are  mostly  used  (Ozgüven,  1994;  ?eker  and Gençdo?an, 2006; Erku?, 2012).  In  this  research,  the  scale  was graded as strongly disagree (1 point), disagree (2), agree at mid level (3), agree (4), strongly agree (5). Three different total points are received through this scale. The total score received varies from 16 to 107 and high level of scores indicate the increasing level of loneliness at work. The original scale is composed of 2 dimensions; emotional deprivation and social companionship. The sub-dimension of emotional deprivation measures the quality of employee's relation with collegues at work; comprising the items like 'exclusion', 'friendship' and 'feeling of disconnection'. This sub-dimension can be described as the perception of the emotional quality of relations at workplace. Social Companionship sub-dimension is about the quantity of the relations at work; comprising the items like 'Participation to activities' and 'sharing the problems'. This sub-dimension could also be described as the perception of the emotional quantitiy of relations at workplace. The correlation between these two sub-divisions was determined as 0.63. The 5th, 6th, 10th, 11th, 12th, 14th, 15th, 16th items of LAWS were coded reversely.

The validity and reliability studies for the original form was conducted by Wright et al. (2006). According to explanatory factor analysis (EFA); the items between 1 and 9 constitute 'emotional deprivation'; between 10 and 16 'social companionship' factor. The confirmatory factor analysis (CFA) was applied on a new sample of 363 employees. Within CFA, the fit index of the scale's two-dimension structure was examined and the chi-square value (χ2=300.4, p< .001) was determined as significant. The fit index values were found as RMSEA=.09, GFI=. 87, AGFI=.83, CFI=.93, CI=.08-.10, BIC=582.99.

LAWS was adapted to Turkish by Do?an et al. (2009). In this research, the reliability of LAWS was examined by internal consis-tency and test-retest methods. The Cronbach α coefficient that was gained from the sample was .91 for the scale in total,  .87 for emotional deprivation sub-dimension and .83 for social relationship sub-dimension. The results about item total correlation vary between .47 and .76 for emotional deprivation, and between .49 and .67 for social companionship. The factor loads of the emotional deprivation sub-dimension emerging as a result of EFA vary from .39 to .83 and this sub-dimension explains the 27.51% of the total variation. The factor loads of social companionship vary from .54 to .79 and this sub-dimension explains the 23.88 % of the total variation. Coefficient of correlation between sub-dimensions was found as .68.

The model fit index of LAWS acquired from CFA was analysed and the minimum chi-square value (χ2=182.56, s=436, p= .00) was found significant. The fit index values were found as RMSEA=.047, NFI=.98, CFI=.99, IFI=.99, RFI=.97, GFI=.95 and AGFI=.93. These fit index values indicate that the modal is coherent.  The factor loads of the model vary from .45 to .85 regarding the emotional deprivation sub-dimension and from .49 to .90 regarding the social companionship. The correlation between factors was .79 (Do?an et al., 2009).

In the current research, as the sample group consists of only the school and institution principals from the field of education, the validity and reliability analysis for LAWS were repeated. In order to determine the factor structure of LAWS, EFA was applied. For the factor loads, the values over .30 was found adequate. In order to assess the appropriateness of the sample to the factor analysi, Kaiser–Meyer-Olkin (KMO) measure of sampling adequacy and Barlett test of sphericity were applied. Kaiser-Meyer-Olkin (KMO) coefficient of sampling adequacy was found as .85 and χ2 value of Barlett Sphericity test was found as 1373.762 (p< .05). These results indicate that the data are adequate for factor analysis. As a result of Principal Components Analysis and Varimax rotation with EFA, a two-factor structure was obtained which has an eigenvalue above 1 and explains the 44.1% of total variance; and does not take part in the factors of the scale's original form. In consequence, it was decided to apply CFA.

The fit index of the model obtianed from CFA was examined and the minimum chi-square value (χ2=226.903, CMIN/DF: 2.315, p=.000) was found significant. The fit index values were found as RMSEA=.068, NFI=.83, CFI=.90, IFI=.90, RFI=.80, GFI=.91 ve AGFI=.88. These fit index values indicate that the modal is coherent. The factor loadings vary from .24 to .67 regarding the emotional deprivation factor and from .35 to .82 regarding the social companionship factor. The correlation between factors was found as .53.

The internal consistency which is the reliability indicator of LAWS was calculated by  Cronbach Alpha and it was found as .84 for the scale in total, .70 for the factor of emotional deprivation and .82 for the factor of social companionship.

Employees’ Performance Scale (EPS)

For measuring the individual performance of employee, the Employee Performance Scale, consisting of 4 items and unidimen-sional, which was used firstly by Kirkman and Rosen (1999) and later by Sigler and Pierson (2000) was used in current research. In this research, EPS was graded as strongly disagree (1 point), disagree (2), agree at mid level (3), agree (4), strongly agree (5). The minimum point to be received from EPS is 1 and the maximum is 5. It was determined that EPS was adapted to Turkish by Çol (2008) and by Gürbüz et al. (2010) separately. Çol (2008) found its reliability as .82, while Erku? et al. 2010) found it as .89.

In this research, the result of Keiser- Meyer- Olkin analysis was .74 and Barlett test was found significant (p= .000). Coefficient of correlation was 0.81. The factor loads of   EPS   as a result of EFA vary from .867 to .746 and it explains the 64.13 % of the total variation.

Personal Information Form

The Personal Information Form was prepared by the researcher in order to determine the age, gender, educational status, branch, length of service, length of principality service of the participants, level and the socio-cultural status of the school to be worked as a principle.

Data collection process

Data collection tools were applied to the participants as in two groups on a volunteer basis in the morning and in the afternoon at the place where they receive their in-service training. After giving information to the participants about the aim of the research, the process was applied. The participants were answered and informed through their questions. No ID information was required. The application process of EPS and LAWS took about 10 to 15 min.

Data analysis

Within the scope of research, in order to identify the loneliness at work and the level of principals' performance, average and standard deviation were determined. When interpreting the averages, for determining the performance/loneliness level, the scores in the lowest one-third of the distribution (1-2.33) were considered as low; those in the middle one-third of the distribution (2.34-3.66) were considered as middle and those at the top one-third of the distribution were considered as high. Hierarchical regression analysis was used in the analysis of the correlation between variables. The significance level in analyses was accepted as .05. These analyses were realized by means of IBM SPSS Statistics Version 18 (2009) and IBM SPSS Amos 21.

The first phase of statistical studies conducted as part of this research was examining the descriptive statistics of dependant and independant variables. Within this scope; the average scores and standard deviation values of variables like performance, social companionship, emotional deprivation were determined. Furthermore, the internal consistency coefficient of these variables were estimated and reported. As the primary sub-purpose of the researcher, the descriptive statistical findings of individual performance of principals in their workplace -institutions- and the level of their loneliness were stated (Table 1).

Principals' performance level (X=4.32 ss.=.55) is high. The principals suffer from loneliness in the medium level (X=2.54, ss.=1.20) considering the social relationship dimension of Loneliness at Workplace Scale, in the low level considering the emotional deprivation dimension (X=1.95, ss.=.61)  and the total scores (X=2.13 ss.=.76) received from the scale.

As the secondary sub-purpose of the research, the findings of hierarchical regression analysis considering the question of “Together with the parameters of total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal; does the level of loneliness at work (sub-dimensions of emotional deprivation and social companionship) predict positively the individual performance of principals?” were stated in Table 2.

In the first phase, total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal; in the second phase the sub-dimensions of loneliness at work, emotional deprivation and social companionship were counted into the model. The results are shown in Table 2.

In accordance with the secondary purpose of the re-search, the hierarchical regression analysis was applied concerning the prediction of principals’ performance by the level of loneliness at work (dimensions of emotional deprivation and social companionship), total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal. In the first phase, total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal; in the second phase the dimensions of emotional deprivation and social relationship were counted into the model (Table 2).

As stated in Table 2, in the first regression model total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal predict the performance scores significantly (R²= .042, F_(4,270)=2.99, p<.05). The contribution of these four varia-bles to the total variance is 4%. The results obtained from the analysis indicate that; in the first control set, total working years as principal in the current institution (β=-.080, t₍₂₇₄₎=1.331, p>.05),  the manner of placement as principal (β=-.047, t₍₂₇₄₎=.709, p>.05) and total working years as principal (β=-.040, t₍₂₇₄₎=.597, p>.05) has no significant relation with performance. Accordingly,  working years as principal at current institution, the manner of placement as principal and total working years as principal does not cause any change in their performance. However, the socio-economic status of school (β= -.159, t₍₂₇₄₎=-2.626, p<.001%, 95 CI -.33 between -.05) has negative significant correlation with performance. Accordingly, it can be stated that if the socio-economic status of school is high, the individual performance of the principal is low; if the socio-economic status of school is low, the individual performance of the principal is high.

In the second regression model, it was observed that “total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal” predict the performance scores significantly (R²=.193, F_(2,268)=25.064, p<.05, Durbin-Watson= 2.039). Accordingly,  the total working years as principal at current institution (β= .114, t₍₂₇₄₎= 2.049, p>.05 %95 CI .01 between .03), the socio-economic status of school (β= -118, t₍₂₇₄₎= 2.094, p>.05 %95 CI -.27 between .01), the emotional deprivation (β= -.20, t₍₂₇₄₎= 2.965, p>.05 %95 CI -.30 between -.06) and the social companionship (β= -.24, t₍₂₇₄₎= 3.589, p>.05 %95 CI -18 between -.05) have important correlation with principals' individual performance at work. As emotional deprivation and social companionionship were added to independent variable set in model 1, it was observed that “the total working years as principal at current institution” became an important independent variable while it is unimportant in model 1. As being important explanatory of model 2 the total working years as principal at current institution, the socio-economic status of school, emotional deprivation and social companionship explain the 15% of total variance of performance which is dependant variable. The total explained variance change between model 1 and model 2 is in the level of 11% (15%-4%). This result shows that model 2 is more functional in the explanation of the principals' performance. In other words, counting the variables of emotional deprivation and social companionship as independent variables in the model provides advance in regression model. There seems to be a relatedness between the total working years’s as principals at current institution increase, the socio-economic status of that school rises; as emotional deprivation and social companionship decrease, the performance of principals increases.

In this work, principals' performance is high according to their perceptions.  Erku? et al. (2010) in IT sector, Çelik and Ç?ra (2013) in tourism sector and Tutar and Alt?noz (2010) in industry sector found the individual performance as high. As seen, employees generally tend to evaluate their performance as high. In fact, in business life, employers complain about the low performance level of their employees. This fact rises as one of the main reasons of conflict at workplace. The effect of culture should not be ignored in the assessment of the individual performance of the employees. In socio-cultural environments with low level of welfare and democracy, employees are accused of having low performance (Ouchi, 1989;  Sigler and Pearson, 2000).

Principals in education sector suffer loneliness at medium-level in the social relationship sub-dimension of Loneliness At Work Scale, at low level according to the emotional deprivation sub-dimension and the scores received from the scale in total. In their work on different professional branches, Wright et al. (2006) and Iecovich (2013) found that the participants suffer from loneliness at medium level. For Y?lmaz and Alt?nok (2009), shool principals' loneeliness level is low, for Izgar (2009),  however, thier loneliness level is at medium level. In our research, principals in education sector experience low level of loneliness. These different results may arise from the socio-cultural development level of the area in which the principals work. Hence, as the research area of this study, Aydin is socio-economically developed and has a population in which parents have a high interest in education and the students are successful in national university entrance exams (Güçlü, 2013). Thus, the fact that principals do not feel lonely in such an environment can easily be comprehended.

The total working years as principal at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal predict the performance scores at significant level. No significant correlation was found between the total working years as principal, the manner of placement as principal and the total working years as principal, which are in the first control set and performance. Accordingly, the total working years as principal, the manner of placement as principal and the total working years as principal do not cause any change in their performance. Turunç and Çelik (2010) state that individual paramaters like age, gender, marital status do not have any significant relation with performance. Cacioppe et al. (1998) dedicate that performance deve-lopment studies which focus on individual parameters like age, gender, marital status and length of service do not give consistent results. However, the socio economic status of the school shows a significant negative correlation with performance. Upon these results, it can be stated that as the socio-economic status of the school is high, the performance of the principal gets low; as the socio-economic status of the school is low, the performance of the principal gets high. Principals of socio-economically low level schools are obliged to show more effort and performance in order to maintain the development of school and routine workflow. In socio-economically low level areas, the parents, local principals and teachers contribute less to the development of schools (Kaya, 1993). Thus, it is highly comprehended that the principals of such schools claim to show high individual performance.

As the independent variables “the total working years as principals at current institution, the socio-economic status of the school, the manner of placement as principal and the total working years as principal” are controlled, it was clearly observed that the emotional deprivation and social relationship scores predict the performance scores significantly. Accordingly, emotional deprivation and social relationship show a negative correlation with individual performance at work. As the principals' emotional deprivation and social relationship scores decrease (high point means high level of loneliness), the performance at work increases. It is understood that the principals’ being emotionally deprived and having poor social relations and support cause their performances to be low. Similarly, in their research Golden et al. (2008) indicate that there is a significant negative relation between professional isolation and performance; as the isolation increases, performance decreases accordingly. As for Campbell et al. (2006), the lack of coworkers' support is one of the main reasons of loneliness at work.

According to findings of the research, principals in education sector experience loneliness at medium level within the scope of companionship sub-dimension of loneliness scale. In this case, in order to establish effective relations with their coworkers at school, the principals can be provided with trainings focusing on the development of their leadership skills. Besides, cowork-ers at school may be encouraged to participate in the school administration process and also the legal obstacles against their participation should be abolished. Considering the fact that emotional deprivation and companionship dimensions affect principals' perfor-mance, they should be given psychological support in order to eliminate the feeling of loneliness. Rather than the personal characteristics as age, gender, length of service, it should be highly considered that organizational factors (organizational environment, loneliness, organiza-tional commitment, citizenship etc.) and environmental factors (socio-economic status, cultural patterns etc.) are more important. The principals should perform studies in order to manipulate the effects of organizational and environmental factors positively on performance.

This study is limited to the principals in educational organizations in Aydin. Furthermore, another limitation of this research is evaluating mangers' performance from the point of their own perceptions. In further studies, while analysing the relationship between loneliness and performance, the principals' performance may be evaluated from the points of views of techers, students and parents.

The author has not declared any conflict of interests.