DESIMAL: JURNAL MATEMATIKA

The purpose of this research is to find out the structural equation model (Structural Equation Model) on the perceptions of users of the graduates of the Teaching and Education Faculty (FKIP)


INTRODUCTION
Statistical methods that can be used to analyze causal relationships can be analyzed using structural equation modeling (SEM). SEM can accommodate the relationship of a variable that is not only direct but also indirect (Carrasco, 2010).
The direct and indirect relationships are analyzed using path analysis. while the relationship between latent variables and their indicators was analyzed by confirmatory factor analysis, thus it can be stated that SEM is a combination of confirmatory factor analysis and path analysis (Sri Indra Maiyanti, Oki Dwipurwani, 2008). The application of SEM allows researchers to be able to answer research questions that are regressive or dimensional, where SEM combines factor analysis with multiple regression analysis (Dirgantara & Suryadarma, 2014) which can be applied separately only in factor analysis or only in regression analysis (Carrasco, 2010). Furthermore, SEM in conducting confirmatory data analysis always requires various assumptions such as the theory must be sufficiently supported, the number of samples is large, and must be normally distributed (Prayitno et al., 2021). Partial Least Squares (PLS) is a method for building predictive models when the factors are multiple and highly collinear (Tobias, 1998). PLS can link a set of independent variables with several dependent (response) variables. On the predictor side, PLS can handle many independent variables, even when the predictor displays multicollinearity (Garson, 2016).
In this research, factors that influence performance will be analyzed as performance capabilities related to main competencies, and performance as an individual's ability to do something with certain expertise. Performance is achieved by a person or group of people in an organization, in accordance with their respective authorities and responsibilities, in an effort to achieve the goals of the organization concerned legally, not violating the law, and in accordance with morals and ethics. The above formulation explains that performance is the level of success of a person or institution in carrying out its work (Rahman, 2022).
Meanwhile, endogenous variables include behavioral ethics, communication skills, English language skills, and the ability to use information technology. Behavioral ethics is a collection of principles, values, or morals, which are used as a guide for someone to behave (Sultoni et al., 2018). Communication ability is a person's skill in interacting between individuals with other individuals (Pehrson et al., 2016), the ability to speak English skills in interacting using a foreign language (English) (Handayani, 2016), the ability to use information technology is the ability to use information technology that requires skills in mastering computer programs in utilizing available technology (Ceha et al., 2008), and self-development efforts are activities carried out to increase selfprofessionalism in order to have professional competence (Rahman, 2022).
The purposes of this research are: (1) to analyze the influence of ethical behavior, communication skills, English language skills, and the ability to use information technology on the ability to work in teams for FKIP UMPRI graduates; (2) to analyze the influence of ethical behavior, communication skills, English language skills, and the ability to use information technology on the performance achievements of the main competency abilities of FKIP UMPRI graduates through the ability to work in teams; and (3) to analyze the effect of the ability to work in a team on the performance achievements of the main competency abilities of FKIP UMPRI graduates.
This is an interesting research study because several research results state that the development of courses based on communication skills, foreign language skills, and applicable courses according to the skills and competencies of graduates are several points that are used as recommendations to improve graduate competencies (Setyaningsih, 2013). The ability for self-development and high integrity are expected to be possessed by graduates as an effort to deal with the professional needs required by graduate users (Santoso et al., 2019). The measurement of graduate performance needs to be carried out by universities as the basis for implementing higher education strategies for progressive improvement (Prasetyawati & Kosasih, 2021). Performance capability is supported by program administration and study program service quality (Sulistiana et al., 2016).
Based on the description, the measurement of factors that affect the ability to work in teams and performance abilities related to the main competencies of the Teaching and Education Faculty (FKIP) Muhammadiyah Pringsewu University (UMPRI) is the topic of the discussion in this research with its update by analyzing these factors using SEM PLS.

METHOD
This research is a type of quantitative approach research that explains the influence of research variables using Structural Equation Analysis (SEM), to be precise using PLS, because the research sample consisted of 27 graduate users who responded to graduates working in the institutions/institutions they led. The sampling technique is classified as purposive sampling, in which the questionnaire given is intended to be filled in by users who have graduated from FKIP UMPRI.

Operational Definition of Variables
The following is presented in Table 1 to Table 7, namely operational descriptions of each research variables:       The ability to provide learning services Y20 Likert

Conceptual Framework
An overview of the path diagram of the hypothesized proportions in the endogenous and exogenous variables connectedness model can be presented in Figure 1.

RESULTS AND DISCUSSION
Based on the results of data analysis using software R, the following results are obtained:

Measurement Model Assessment a. Convergent validity
In evaluating convergent validity, it can be seen from the value of the loading factor.

b. Composite Validity
Composite validity is used to measure the stability and consistency of combined reliability measurements. Internal consistency reliability (composite reliability) with RhoA data, average variance extracted (AVE) is expected to be above 0.5, and RhoA above 0.7. Table 8 shows the results of the composite validity calculation.  Table 8 shows that self-development and the ability to work variables have an AVE value of less than 0.5. This can be interpreted that the two variables are not stable and consistent enough. It would be better if the indicator was reduced again.

c. Discriminant Validity
Discriminant validity with crossloading data, where the value for an indicator on one variable must be greater than the cross-loading value for another variable indicator in the same column. To find out whether the latent variable has sufficient discriminant, that is by comparing the intended cross-loading value, it must be greater than the other latent variables. The results obtained show that not all indicators meet discriminant validity, which means that not all indicators have sufficient discriminant. Because there are latent variables that are actually more explained by other indicators. This seems to support composite reliability in that it is necessary to reduce the indicators again, namely by eliminating the indicators that have the smallest loading factor values.

Determination of Appropriate Indicators
Due to the results obtained do not meet the indicators that deserve to be included in the model, by using R software, new indicators will be obtained to be used in the SEM model.

Table 9. Acquisition of Indicators in the SEM Model
The results of the loading factor values shown in Table 9 show that all indicators have met convergent validity which exceeds the value of 0.6. Following are the results of obtaining composite validity on indicators that are feasible to be included in the model.  Table 10 shows that each indicator is well-represented or reliable for measuring each latent variable. Furthermore, based on the indicator data that can represent each variable, the value of the loading factor will be determined. If a loading factor value is obtained that is more than 0.6, then the acquisition of each indicator will be continued by determining the structural model equation.  Table 11 shows that the loading factor obtained from each indicator on the attribute obtained a loading factor value of more than 0.6. This shows that indicators can measure their latent variables well.

Structural Model Assessment a. Collinearity Test
To see if there is multicollinearity between indicators can be seen by the collinearity test shown in Table 12.   Table 12 shows that the acquisition of VIF values for all indicators is less than 5, this means that all indicators do not experience multicollinearity.

b. Coefficient of Determination
The coefficient of determination R 2 , R square value 0.75; 0.50, or 0.25 respectively which means that the proportion of indicators is substantial (good), moderate (moderate), or weak (less), the percentage in measuring endogenous variables can be seen in Table  13.  Table 13 shows that the coefficient of determination is more than 0.75. This result means that the proportion of indicators is substantial; or good and fit for use.

c. The Effect of Independent Variables on Dependent Variables
The f square gain will indicate the magnitude of the effect of independent variables on the dependent variables with a category value range of 0.02; 0.15; and 0.35 which indicates a small-medium effect, or a large effect. In the following, the obtained f square calculation results are presented using R software:

SEM Modeling
SEM modeling with PLS uses indicators that have fulfilled the test. After bootstrapping, the SEM pathway model is obtained as in Figure 2.

Figure 2. SEM-PLS Pathway Model
The visualization in Figure 2 is the SEM-PLS path model by bootstrapping 50 times. Furthermore, to see whether the influence is significant, it can be seen by the absence of a value of 0 between the upper and lower limits. The upper and lower limits are the confidence intervals of 0.025 or 2.5% and 0.975 or 97.5%, shown in Table 15.  Table 15 states the effect of each indicator on the latent variables. The estimated value is indicated by the column Original Est. X12 has a value of 1 because BI development only has 1 indicator, namely X12. Next, Table 16 is the estimation of the structural model.  Table 16 shows that ethics has a positive effect on the ability to work and ethics also has a positive effect on main competencies. This means that ethics will improve the ability to work as well as improve the main competence. Meanwhile, communication skills have a negative effect on the ability to work and also have a negative effect on main competencies. Another thing is the development of English language skills has a positive effect on the ability to work and main competencies. This means that improving English skills will improve the ability to work as well as main competencies.
The opposite happened to the influence of IT skills which had a negative influence on the ability to work and main competencies. While self-development has a positive effect on the ability to work and main competence with a fairly large ratio. This means that increasing selfdevelopment will also improve the ability to work and main competencies. The same is followed by the positive influence of the ability to work on the main competencies. Where the increased ability to work will increase the main competence.
Thus, from the results obtained, it can be stated that (1) ethical behavior, English language skills, and selfdevelopment have a positive effect on the ability to work in teams for FKIP UMPRI graduates and the achievement of the performance of the main competency abilities of FKIP UMPRI graduates; and (2) communication skills and the ability to use information technology have a negative effect on the ability to work in teams and the performance achievements of the main competency abilities of FKIP UMPRI graduates. Based on these results, the user's perception of the factors that support the competency of FKIP UMPRI graduates is still in ethical behavior, English skills, and self-development. On the other hand, for communication skills and the ability to use information technology from processed data it can be stated that when a graduate's communication skills and ability to use technology are good, it does not guarantee that the ability to work in teams and the performance achievements of the main competency abilities of FKIP UMPRI graduates will be good too, and vice versa.
Furthermore, if viewed from the results of the PLS process, where the bootstrapping results are carried out 50 bootstrapping times it seems that the latent variables influence each other but are not significant. However, if it is done more than 50 bootstrapping, no results (errors) are obtained. As is the case in the initial model where indicator reduction has not been carried out that the latent variables have no significant effect. The recommended solution is to increase the amount of bootstrapping, but adding bootstrapping does not produce anything, by adding samples.
The implications of the research results indicate that ethical behavior, English language skills, and selfdevelopment efforts have a positive effect on the ability to work. In contrast, the results of other research show that the components of communication skills and the ability to use information technology have not had a positive effect on the ability to work in teams and the performance achievements of graduates' main competency abilities. Based on these results, the user's perception of the factors that support the competency of FKIP UMPRI graduates is still in behavioral ethics, English skills, and selfdevelopment efforts. This is in line with several research results which state that good ability to work is associated with high-quality work and high productivity as well as the enjoyment of remaining in one's job (Tuomi et al., 2001). Professional graduate competence is accompanied by the fulfillment of attitudes and values toward work (Van Den Berg et al., 2009). The measurement factor of the available ability to work highlights the facts that have an impact on the ability to work possessed by workers in accordance with the purpose of the measurement (Fadyl et al., 2010). Thus, the recency in this research is the measurement of two endogenous variables, namely the competence of graduates which is influenced by the ability to work, where the ability to work is influenced by five exogenous variables simultaneously the five exogenous variables (ethics, selfdevelopment, IT skills, English skills, and communication) do not jointly have a positive influence on the ability to work and the competence of graduates. Research data and results show that the interrelationship of factors that influence graduate competence and ability to work is in accordance with the social and environmental conditions where graduates come from and where graduates work, where the ability to work is in accordance with the factors that influence physically, mentally, socially, environmental, and organizational demands on work and capacity (Duivenbooden & Burdorf, 2015).

CONCLUSIONS AND SUGGESTIONS
Based on the results of data analysis and discussion in this research, it can be concluded that not all indicators have sufficient effect on latent variables so that indicator reduction is carried out until sufficient indicators are obtained, the acquisition of variables that give effects, namely ethical behavior, English language skills, and self-development efforts. It can also be shown that there is a positive influence from the ability to work on the main competencies, where increasing the ability to work will increase the main competencies of FKIP UMPRI graduates.
A suggestion for future researchers based on the results of this research is the expansion of factors that affect the ability to work and competence of graduates based on physical, mental, social, environmental conditions and organizational demands.