MODEL OF AFFECTIVE ASSESSMENT OF PRIMARY SCHOOL STUDENTS

This study aims to develop an instrument of affective assessment to measure the social competence of elementary school students in the learning process in schools. This study used the development model of Borg & Gall’s approach which was modified into five phases, including the need analyses, developing draft of the product conducted by experts, developing an affective assessment instrument, trying out the affective assessment instrument conducted by teachers of primary education in Yogyakarta, and the dissemination and implementation of the developed affective assessment instrument. The subjects were elementary school students whose school implemented Curriculum 2013 in the academic year of 2013/2014. The validity and reliability of each construct of the affective instrument were established using the PLS SEM Wrap PLS 3.0 analysis program. The study finds the following results. First, the construct of Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence in the limited, main, and extended testing has been supported by empirical data. Second, the validity of Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence in the limited, main, and extended testing meets the criteria above 0.70 for each indicator of the loading factor and the criteria below 0.50 for each indicator score of the cross-loading factor. Third, the reliability of Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence in limited, main, and extended testing meets the criteria above 0.70 for both composite reliability and Cronbach’s alpha scores. Fourth, the number of indicators at preresearch was 53, and 10 indicators were rejected in the limited testing, and four indicators were rejected in the main testing, and one indicator was rejected in the extended testing.


Introduction
Building national education is an activity of shaping the mentality of prime human individuals who are law-abiding and also respectful of the norms and customs rooted in Indonesian culture (Ekowarni, 2009, pp.2-3).In school context, shaping children's mentality implies accustoming them to behaving in such praiseworthy ways as being honest in speech and action, helpful to those in suffering, filled with perseverance in working, and tolerant of diversity in race, custom, religion, and socioeconomic status.It has become part of the most important aspirations of Indonesian nation as expressed in the statement of the national educational goal.
The national educational goal is to develop learners' potentials so that they become Indonesian individuals with faith in and fear of God, noble morals, good health, great knowledge, competence, creativity, and independence, and become individuals who are democratic and responsible citizens.Based on the national educational goal, the objective of education in Islam at SD (short for sekolah dasar or primary school) or MI (short for madrasah ibtidaiyah, or primary school counterpart affiliated to Islam) is stated as follows: (1) To develop learners' faith in and fear of Allah and (2) to cause the manifestation of learners who are devout and of noble morals (Center of Curriculum, 2007, p.9).
In Indonesia, religion education is the shared responsibility of family, school, and society (Dewantara, 2009, pp.103-107).The result of religion education in the family is love, that at school is science, and that in society is social interaction.This differs from the educational tradition in such progressive or developed countries as the United States of America (Gable, 1986, p.1), where the school handles learners' intelligence and character while religion education becomes the responsibility of the family and church or society.However, in Indonesia, religion education conducted by family and society is reenforced by religion education at school.
Religion education ruled by society is organized through such an activity as learning to read Al-Qur'an, the Islam holy book, at a mosque and other places of worship, namely, a mushola, a surau, and a langgar; an Islamic learning forum such as a majlis ta'lim, and a pondok pesantren -an Islamic boarding school.A person called guru ngaji -literally meaning a teacher of Al-Qur'an reading -or kiai or ajeungan, according to different regions, is a central figure in Islamic education in society.Being respectful in attitude and absolutely obedient to guru ngaji or kiai or ajeungan is the main value implanted in the consciousness of santris -students of pondok pesantren.The materials taught are reading and putting in writing the contents of Al-Qur'an, learning by rote the contents of Al-Qur'an or the hadiths (collections of reports of the Prophet's sayings), using basic Arabic grammar, and knowing the fiqh (derived laws) of ritual worship or mu'amalah (financial transaction).Learners or santris listen to the explanation given by guru ngaji or kiai or ajeungan about the content of Al-Qur'an related to the procedures of the ritual worship or normative principles in social worship or even direct practice in activities of living independently like planting rice in paddy fields or growing plants in the garden or breeding fish in a pond owned by guru ngaji or kiai or ajeungan.Guru ngaji or kiai or ajeungan acts as a role model of living within society in accordance with religious values so that individuals with noble personality and civilized society are formed (Van Bruinessen, 1995, pp.18-21).
The individual education implanted in the family to make one active, persevering, courageous, smart, alert, clear in thinking, and caring is reinforced by religion education at school.Likewise, social education habitually practiced in the family to make one frugal, dislike vain deeds or conditions, give help easily, and have empathy to someone suffering is also reinforced by religion education in school.Education of the family model works in synergy with education of the school model (Dewantara, 2009, pp.104-105).
The practice of religion education at school is conducted by religion teacher of the same faith as the learners and refers to ramburambu or directive of curriculum set by the government, while learning outcome is evaluated according to the critera derived from the objective of religion teaching at school.The process of religion education at primary school covers three domains: Cognitive, affective, and psychomotor.Therefore, the learning achievement evaluation should also cover cognitive, affective, and psychomotor abilities.However, Krathwohl, Benjamin, and Bertram (1973, p.15) state that affective learning achievement evaluation is seldom done by teachers.When it is done, it is something forced because there is some educational research being done or there are questions from some academicians.This is strengthened by Bagir (2003, p.6) that the practice of religion education at school focuses on cognitive ability only and does not touch learners' center of moral awareness of which the source is affective ability.As a result, learners could obtain a score of 80 in a cognitive test on religion but they could receive a score of 50 for their affective ability and behavior.They know the obligation of cleansing themselves from najis (ritually unclean object) after urinating, but cleansing their body parts from najis is more often simply not done.
Evaluation of learning achievement in religion education at school should emphasize wholeness and integration among cognitive, affective, and psychomotor domains.Thus, the evaluation result could give whole and integrated information of the learning activity within a certain length of time.However, the reality reveals that the learners' cognitive evaluation is more often done by teachers of religion subject.The technique of evaluation employed is testing.Meanwhile, non-test evaluations revealing the learners' affective ability and behavior are not yet fully done by classroom teachers, teachers of physical education, sports, and health, and religion teachers.
According to Tyler (1973, p.2), there are two reasons why affective domain is disregarded by educators.First, generally teachers have the view that the main job of the school is developing cognitive domain concerned while the development of affective domain becomes the job of the family or the church.Second, it is assumed that affective domain would develop along with the development of cognitive domain or, in other words, the more the learners' cognitive domain develops, the more their affective domain develops.The truth of the assumption has been refuted empirically (Bagir, 2003;Anderson & Anderson, 1982).
The two reasons of Tyler are based on an assumption that there is difficulty in constructing affective teaching and learning objective.However, the teacher's success in conducting affective teaching and learning, and the learners' success in achieving affective competence need to be attained though the attainment is assumed as a 'difficult' thing to do.According to Anderson (1982, p.524), affective evaluation is quite possible to do on condition that it meets five principles, namely: (1) The objective of the evaluation is clear; (2) what is to be evaluated is clear; (3) it uses the right instrument; (4) the quality of the instrument used is known; (5) the method of interpreting the data presented by the instrument is known.The affective factors relevant to the teaching and learning process are attitude, interest, values, tendency of choice, awareness of self-esteem, self-control, and anxiety.
The research specifically focuses on the affective behavior concerning social competence related to human or other objects and appearing from the teaching and learning process conducted to primary school students by their classroom teacher, physical education, sports, and health teacher, and religion teacher.The presence of social competence is measured by using an affective instrument.Because the affective instrument used by teachers at primary school is not yet adequate, affective instrument which is understandable to its users (i.e., teachers), objective, valid, and reliable in nature, and easy to interpret need to be made available.

Method
The research was research and development.It employed the approach of development model of Borg and Gall (1983).
The research was conducted from February to September 2014.The setting was the primary schools that have implemented Curriculum 2013 in Yogyakarta Special Region, Indonesia.
The research subjects were 245 primary school students in limited testing, 529 in main testing, and 887 in extended testing.The users of affective evaluation instrument in this case were 30 primary school classroom teachers.

Procedure
The try-out planning of the developed affective evaluation model on the social competence of primary school students went through the steps of need analysis, the formulation of conceptual definitions, their derivation into operational definitions, and the formulation of indicators of each research variable, resulting in a draft of the model of affective evaluation on primary school students' social competence, the process of expert judgement, the implementation of the development, and the dissemination of the research results (Borg & Gall, 1983, p.776).A flow chart of the testing plan is presented in Figure 1.

Data, Intrument, and Technique of Data Compilation
In terms of data type, the research data concerned were quantitative data on students' social competence in a Likert scale with a continuum line in positive and negative directions.The instruments of data compilation were an observation sheet, a questionnaire on the readibility of the instrument for affective evaluation on social competence, and a list of documents.
The data were compiled with the technique of using the observation sheet, questionnaire, and documents concerning social competence in the process of students' social interaction with teachers, peers, and other people within the school area during teaching and learning activities.

Technique of Data Analysis
The data on the practice of affective evaluation at primary school were analyzed by using a technique of qualitative data analysis going through the steps of data reduction, data presentation, and data verification.Data on the quality of affective evaluation instrument were analyzed by means of confirmatory factor analyses (CFA) technique using the analysis program of SEM PLS Warp 3.0 to determine the convergent validity, discriminant validity, and reliability of constructs of Honesty, Decency, Responsibility, Discipline, Care, and Self-Confidence.

Findings and Discussion
There are two criteria for evaluating the outer model to fulfill the requirements for the convergent validity of reflective constructs, namely: The score of loading factor is ≥0.70 and the value of p is significant (<0.05) (Hair, Hult, Ringle, & Sarstedt, 2013).The reliability of instrument is indicated by two scores: Composite reliability and Cronbach's alpha coefficient of reliability.Both scores should be greater than 0.70 as required for reliability (Nunnaly, 1978).Discussion on convergent validity and reliability of the constructs of Honesty, Discipline, Decency, Responsibility, Care, and Self-Confidence is as follows.The latent sub-variable of Principle Adherence has four indicators: j1, j2, j3, and j4.In fact, j1 and j4 have loading factor scores below 0.70 so that they are not included in extended testing analysis.j2 and j3 have loading factor scores above 0.70 and there is no crossloading score above 0.5 in the limited try out, main testing, and extended testing.
The latent sub-variable of Universality has six indicators: j6, j7, j8, j9, j10, and j11.All of them have the loading factor scores above 0.70 and there is no cross-loading score above 0.50 in the limited try out, main testing, and extended testing.Data of convergent validity of the construct of Honesty are in Table 2.
The reliability analysis results of the construct of Honesty show: The composite reliability scores of 0.909 and 0.930 in the limited try out, 0.847 and 0.924 in main testing, and 0.862 and 0.896 in extended testing of the latent sub-variables of Principle Adherence and Universality, respectively.
The scores of Cronbach's alpha coefficient of reliability also show the levels above 0.70, namely, 0.866 and 0.910 in the limited try out, 0.729 and 0.901 in main testing, and 0.679 and 0.860 in extended testing of the latent sub-variables of Principle Adherence and Universality respectively.Data of the instrument reliability are in Table 3.

The Construct of Discipline
Theoretically, construct of Discipline consisted at first of four latent sub-variables with 13 indicators.After the limited try out, it consisted of two latent sub-variables with nine indicators.Meanwhile, after the main testing, it consisted of two latent sub-variables with eight indicators.After the extended testing, it consisted of two latent sub-variables with indicators which are remaining eight in number.The data are presented in Table 4.The latent sub-variable of Obedience has five indicators, namely: d1, d2, d3, d5, and d6.The indicator of d1 has the loading factor score below 0.70 so that it is not included in the extended testing analysis.Moreover, d2, d3, d5, and d6 have the loading factor scores above 0.70 and there is no cross-loading score above 0.50 in the limited try out, the main testing, and the extended testing.
The latent sub-variable of Integrity has four indicators: d8, d9, d10, and d11.All of them have the loading factor scores above 0.70 and there is no cross-loading score above 0.5 in the limited try out, the main testing, and the extended testing.The data of convergent validity are clearly presented in Table 5.
The results of the reliability analysis of the construct of Discipline present the composite reliability scores of 0.906 and 0.968 in the limited try out, 0.871 and 0.952 in the main testing, and 0.854 and 0.942 in the extended testing for the latent sub-variables of Obedience and Integrity.
The scores of Cronbach's alpha coefficient of reliability for the latent subvariables of Obedience and Integrity also show levels above 0.70, namely, 0.870 and 0.956 in the limited try out, 0.802 and 0.933 in main testing, and 0.772 and 0.918 in extended testing for, respectively, the latent subvariables of Obedience and Integrity.Data of the instrument reliability are in Table 6.

The Construct of Responsibility
Theoretically, the construct of Responsibility consisted at first of four latent sub-variables with eight indicators.After the limited try out, the construct of Responsibility consisted of two latent sub-variables with seven indicators.After the main testing, the construct of Responsibility consisted of two latent sub-variables with seven indicators.After the extended testing, it still consisted of two latent sub-variables with seven indicators.The data are presented in Table 7.  Concerning the convergent validity of the construct of Responsibility, the latent subvariable of Obligation has three indicators: t1, t2, and t3.The three indicators have loading factor scores above 0.70 and there is no crossloading score above 0.50 in the limited try out, main testing, and extended testing.
The latent sub-variable of Endurance has four indicators, namely, t4, t5, t6, and t7.All of them have the loading factor scores above 0.70 and there is no cross-loading score above 0.5 in the limited try out, the main testing, and the extended testing.The data of convergent validity are presented in Table 8.
The reliability analysis results of the construct of Responsibility show composite reliability scores of 0.904 and 0.927 in limited try out, 0.897 and 0.909 in main testing, and 0.874 and 0.892 in extended testing for latent sub-variables of Obligation and Endurance.
The scores of Cronbach's alpha coefficient of reliability for the latent subvariables of Obligation and Endurance also show the levels above 0.70, namely, 0.840 and 0.894 in the limited try out, 0.827 and 0.866 in main testing, and 0.783 and 0.838 in extended testing for latent sub-variables of Obligation and Endurance.The data are in Table 9.

The Construct of Decency
Theoretically, the construct of Decency consisted at first of two latent sub-variables with four indicators.After the limited try out, the construct of Decency consisted of one latent sub-variable with four indicators.After the main testing, it remained consisting of one latent sub-variable with four indicators.After the extended testing, it continued to remain consisting of one latent sub-variable with four indicators.The data are presented in Table 10.
Concerning the convergent validity of the construct of Decency, the latent subvariable of politeness has four indicators, namely, s1, s2, s3, and s4.All of them have the loading factor scores above 0.70 and there is no cross-loading score above 0.50 in the limited try out, the main testing, and the extended testing.Data of the convergent validity are presented in Table 11.The results of the reliability analysis of the construct of Decency show the composite reliability scores of 0.968 in the limited try out, 0.966 in main testing, and 0.960 in extended testing for the latent sub-variable of politeness.The scores of the Cronbach's alpha coefficient of reliability for the latent sub-variable of politeness also show the levels above 0.70, namely, 0.956 in the limited try out, 0.953 in main testing, and 0.945 in extended testing.The data of the instrument reliability are presented in Table 12.
The Construct of Care Theoretically, construct of Care consisted at first of four latent sub-variables with nine indicators.After the limited try out, it consisted of two latent sub-variables with nine indicators.After main testing, it consisted of two latent sub-variables with eight indicators.After extended testing, it consisted of two latent sub-variables with seven indicators.
The data of the construct of Care are presented in Table 13.
Concerning the convergent validity of the construct of Care, the latent sub-variable of Attentiveness has four indicators: p2, p3, p4, and p5.The indicator p2 has the loading factor score above 0.70 and there is no crossloading score above 0.50 in the limited try out but its loading factor score in the main testing is below 0.70 so that it is not included in the analysis.p3, p4, and p5 have the loading factor scores above 0.70 and there is no crossloading score above 0.50 in the limited try out, main testing, and extended testing.
The latent sub-variable of appreciativeness has four indicators: p6, p7, p8, and p9.All of them have the loading factor scores above 0.70 and there is no cross-loading score above 0.50 in the limited try out, main testing, and extended testing.The data of convergent validity are presented in Table 14.The results of reliability analysis of the construct of Care show composite reliability scores of 0.880 and 0.922 in the limited try out, 0.887 and 0.904 in the main testing, and 0.860 and 0.895 in the extended testing for the latent sub-variables of attentiveness and appreciativeness.The data of the instrument reliability are in Table 15.
The Construct of Self-Confidence Theoretically, the construct of Self-Confidence consisted at first of four latent sub-variables with eight indicators.After the limited try out, construct of Self-Confidence consisted of two latent sub-variables with five indicators.After the main testing, it consisted of two latent sub-variables with four indicators.After extended testing, it remained consisting of two latent sub-variables with four indicators.The data are clearly presented in Table16.
Concerning the convergent validity of construct of Self-Confidence, the latent subvariable of assertiveness has one indicator: c2.c2 has loading factor score above 0.70 and there is no cross-loading score above 0.50 in the limited try out, main testing, and extended testing.
The latent sub-variable of tenacity has four indicators: c5, c6, c7, and c8.The indicator c5 has the loading factor score above 0.70, but it has a cross-loading score above 0.50 in the limited try out so that it is not included in the main testing analysis.The indicators of c6, c7, and c8 have the loading factor scores above 0.70 and there is no crossloading score above 0.50 in the limited try out, main testing, and extended testing.The data of convergent validity are in Table 17.
The results of analysis show composite reliability scores of 1.000 and 0.954 in limited try out, 1.000 and 0.939 in main testing, and 1.000 and 0.911 in extended testing for latent sub-variables of assertiveness and tenacity.
The scores of Cronbach's alpha coefficient of reliability also show the levels above 0.70, namely, 1.000 and 0.935 in the limited try out, 1.000 and 0.902 in the main testing, and 1.000 and 0.853 in extended testing for the latent sub-variables of assertiveness and tenacity.The data of the instrument reliability are clearly presented in Table 18.Of the 53 indicators which are originally formulated, 38 indicators meet the criteria for convergent validity, discriminant validity, and reliability as indicated by two score types, namely, scores of composite reliability and Cronbach's alpha coefficient of reliability, in the limited try out, the main testing, and the extended testing.After the extended testing, the number of the indicators of the construct of Honesty, which is originally eleven, has become eight; that of the construct of Discipline, which is originally thirteen, has become eight; that of the construct of Responsibility, which is originally eight, has become seven; that of the construct of Decency, which is originally four, remains four; that of the construct of Care, which is originally nine, has become seven; and that of the construct of Self-Confidence, which is originally eight, has become four.
Information of convergent validity of social value constructs is obtained from the second order analysis results using SEM-PLS WarpPLS 3.0 as basis.There are eight indicators of latent variable of Honesty,namely: j2,j3,j6,j7,j8,j9,j10,j11; three indicators of latent variable of Discipline, namely: d9, d10, d11; five indicators of latent variable of Responsibility, namely: t2, t3, t5, t6, t7; four indicators of latent variable of Decency, namely: s1, s2, s3, s4; one indicator of latent variable of Care, namely: p9; and three indicators of latent variable of Self-Confidence, namely: c6, c7, c8.Loading factor scores above 0.70 are achieved by all these indicators and there is no cross-loading score above 0.50.The data of convergent validity are shown in Table 19.The reliability of the instrument which is measuring social competence is indicated by two score types, namely: Those of composite reliability and Cronbach's alpha coefficient of reliability.The results of analysis indicate that the composite reliability scores for Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence are already above 0.70.The scores of Cronbach's alpha coefficient of reliability for Honesty, Discipline, Decency, Responsibility, Care, and Self-Confidence are also above 0.70.The data of the instrument reliability are presented in Table 20.

Discussion
The data resulting from the limited try out, main testing, and extended testing on the constructs of Honesty, Discipline, Decency, Responsibility, Care, and Self-Confidence in the section of the research findings above are further discussed according to the theoretical review on each construct and the requirements for quality in the affective evaluation instrument according to the criteria for the convergent validity, discriminant validity, and reliability of each construct.
Theoretically, someone who is honest has the characteristics of being able to be trusted by others, speaking in line with reality, being faithful to promises, being firmed in accomplishing tasks, giving fair witness (Izutsu, 1993), showing care of others (Lickona, 1991), basing oneself in accordance with selfidentity (Amin, 1988), treating others according to the same standards (Suseno, 1993), and also holding firmly to social competencies that are universal in nature (Schiller & Bryant, 2002).
Speaking in line with reality as a characteristic of being honest is more often done by students when the information concerned comes from a friend of the same age group than when it comes from a teacher.Indicator j1, which says 'the student does a task in accordance with information from the teacher', does not pass the main testing.Being able to be trusted as a characteristic of being honest is seldomly applicable on students.Indicator j4, which says 'the student tells a lie', does not pass the extended testing.Treating others in accordance with the same standards is seldomly done by students.In addition, indicator j5, which says 'the student treats fellow students differently', does not pass the limited try out.
Theoretically, some characteristics of people with discipline are being independent in taking care of themselves (Suseno, 1991;Mill, 1996), demanding mutual trust, equal rights, and fair treatment, controlling their own behavior out of their own will, planning their activities, and working together with their peers (Watson, 2008).
The characteristic of having the discipline of obeying such a social rule as playing truant while still wearing the school uniform is not suitable to be an indicator of discipline in the area of primary school because playing truant while still wearing the school uniform almost never happens at the schools becoming the research objects.The characteristic of having the discipline of not cheating in doing school work by copying a classmate's work or by giving notes for cheating to a classmate is also unsuitable to be an indicator of discipline at school because probably there is no culture of cheating or, on the contrary, cheating in doing school work is considered as a moral violation that is ignored at the schools concerned.The characteristic of having the discipline of self-control by, for example, being careful in doing activities at school is not suitable to be an indicator of discipline at primary school, either.Theoretically, someone with a sense of responsibility has the characteristics of having the courage of giving an explanation of an act one has done or a word one has uttered (Suseno, 1993;Lickona, 1991), a sense of duty, an awareness of risking being sued, and a feeling of possible appreciation (Bagus, 1996), being the cause of an event, doing an act at one's own free will (Bertens, 1994), being conscious of one's obligation to oneself, to society, and to the environment (Ministry of National Education, 2013;Albertus, 2012), and being dependable due to one's possession of the knowledge and ability of acting at one's own free will (Shihab, 1999).Having the courage to bear the risk for one's utterance or act is not suitable to be an indicator of a sense of responsibility in primary school students because their social responsibility is not yet as much as that of grown-ups.
Theoretically, a characteristic of someone with a sense of decency is showing respect in the form of words, deed, or sign or body language to someone else who has done something for the good of society in general (Izutsu, 1993).Respect is shown in the form of words when, for example, thanking someone for giving one something.Respect is shown in the form of a deed by, for example, being decently dressed when honoring guests.Respect is shown in the form of sign or body language by, for example, smiling or nodding one's head to someone whom one knows and even to a stranger (Miskawaih, 1994;Schiller & Bryant, 2002).
Those characteristics of decency were formulated into four indicators of being polite in social interaction at school, and none of the four indicators has failed in the limited try out, main testing, and extended testing.Theoretically, someone with care or concern has the characteristics of respecting the quality of someone or something (Lickona, 1991), treating others as treating oneself (Feldman, 1985;Schopenhauer, 1997), being filled with care or concern about other's sufferings (Suseno, 1991), understanding other people's point of view, being full of tolerance, empathy, and sensitivity to other's feeling (Elias, Parker, & Kash, 2007), showing care or concern for all creations of God by means of nurturing and preserving them (Shihab, 1999), and wanting what is good for everyone (Miskawaih, 1994).
Helping other people in learning difficulty as a characteristic of caring people is not suitable to be an indicator of caring in the context of education at primary school.Theoretically, someone with self-confidence has the characteristics of having a positive view of one's own ability to achieve a goal with the ability one has as basis (Catalano, Hawkins, & Toumbourou, 2008), an ability to consciously and freely choose (Schiller & Bryant, 2002), a strength influencing motivation and action (Bandura, 1989), and courage and patience in accomplishing a task (Bagus, 1996;Izutsu, 1993).Having the courage to express an opinion to others, being able to communicate in the presence of peers, working hard to attain achievement, and undergoing sufferings to achieve a goal are not suitable to be the indicators of self-confidence in the context of primary school.

Convergent Validity
The requirements of convergent validity of the constructs of Honesty, Discipline, Responsibility, Decency, Care, and Selfconfidence are already fulfilled according to the criteria.There are two criteria for the evaluation of the outer model to fulfill the requirements for the convergent validity of reflective constructs: The score of loading factor is ≥0.70, and the value of p is significant (<0.05) (Hair, Hult, Ringle, & Sarstedt, 2013).The score of the convergent validity of the latent sub-variables should be >0.50 as the minimum criterion (Fornell & Larcker, 1981;Sholihin & Ratmono, 2013, p.73), which is the score of the average variance extracted (AVE) of each latent sub-variable.Latent sub-variables could explain at the minimum 50% of the indicator variance; the percentage is obtained from the square of 0.7, which is 0.49, rounded up to 0.50.

Discriminant Validity
The greater the difference between the AVE score and the score of correlation among the latent sub-variables in the same column, the firmer the discriminant validity of the construct being tested.The cross-loading scores lower than the loading factor scores and cross-loading scores not greater than 0.50 indicate that the discriminant validity is fulfilled (at the minimum 49% correlation with constructs).There is no vertical and lateral collinearity matter when the scores for full collinearity variance inflation factor (VIF) are below 2.5 as maximum criterion (Kock, 2013).
The construct of Honesty has an AVE root score in the column of Principle of Adherence and Universality which is greater than the score of the correlation between latent sub-variables in the same column.There is no vertical and lateral collinearity matter, either, of the construct of Honesty because the full collinearity VIF scores are already below 2.5 as the maximum criterion.In this case, the full collinearity VIF scores are 2.054 in the limited try out, 2.254 in the main testing, and 1.275 in the extended testing.
The construct of Discipline has an AVE root score in the column of Obedience and Integrity which is greater than the score of the correlation between the latent subvariables in the same column.There is no vertical and lateral collinearity matter, either, of the construct of Discipline because the full collinearity VIF scores are already below 2.5 as the maximum criterion.In this case, the full collinearity VIF scores are 1.404 in the limited try out, 1.508 in the main testing, and 1.474 in the extended testing.
The construct of Responsibility has an AVE root score in the column of Obligation and Endurance which is greater than the score of the correlation between the latent sub-variables in the same column.There is no vertical and lateral collinearity matter, either, of the construct of Responsibility because the full collinearity VIF scores are already below 2.5 as the maximum criterion.In this case, the full collinearity VIF scores are 2.338 in the limited try out, 1.774 in the main testing, and 1.823 in the extended testing.
The construct of Decency has no discriminant validity because it has only one latent sub-variable, namely, Politeness, so it cannot be correlated.Therefore, there is no available information concerning discriminant validity data of the construct of Decency.
The construct of Care has an AVE root score in the column of Attentiveness and Appreciativeness which is greater than the score of thecorrelation between the latent sub-variables in the same column.There is no vertical and lateral collinearity matter, either, of the construct of Care because the full collinearity VIF scores are already below 2.5 as the maximum criterion.In this case, the full collinearity VIF scores are 1.817 in the limited try out, 1.363 in the main testing, and 1.388 in the extended testing.
The construct of Self-Confidence has an AVE root score in the column of Assertiveness and Tenacity which is greater than the score of the correlation between latent sub-variables in the same column.There is no vertical and lateral collinearity matter, either, of the construct of Self-Confidence because the full collinearity VIF scores are already under 2.5 as the maximum criterion.In this case, the full collinearity VIF scores are 2.451 in the limited try out, 1.978 in main testing, and 1.843 in extended testing.

Instrument Reliability
The instrument reliability of the construct is indicated by two types of score, namely, the scores of composite reliability (CR) and Cronbach's alpha (CA) coefficient of reliability.Both these scores should be above 0.70 as the requirement for reliability of the construct (Nunnaly, 1978;Sholihin & Ratmono, 2013).CR and CA of each construct are further discussed as follows.
The results of the analysis show that the construct of Honesty has two CR scores and two CA scores in more than one instance.They are CR scores of 0.909 and 0.930 in the limited try out, 0.847 and 0.924 in the main testing, and 0.862 and 0.896 in the extended testing as well as CA scores of 0.866 and 0.910 in the limited try out, 0.729 and 0.901 in the main testing, and 0.679 and 0.860 in the extended testing for the latent sub-variables Principle of Adherence and Universality.All those CR and CA scores show the levels above 0.70.
The results of the analysis show that the construct of Discipline also has two CR scores and two CA scores in more than one Model of affective assessment of primary school... -39 Amir Syamsudin, Budiyono, & Sutrisno instance.They are CR scores of 0.906 and 0.968 in the limited try out, 0.871 and 0.952 in the main testing, and 0.854 and 0.942 in the extended testing as well as CA scores of 0.870 and 0.956 in the limited try out, 0.802 and 0.933 in the main testing, and 0.772 and 0.918 in the extended testing for the latent subvariables of Obedience and Integrity.All those CR and CA scores also show the levels above 0.70.
The results of the analysis show that the construct of Responsibility likewise has two CR scores and two CA scores in more than one instance.They are CR scores of 0.904 and 0.927 in the limited try out, 0.897 and 0.909 in the main testing, and 0.874 and 0.892 in the extended testing as well as CA scores of 0.840 and 0.894 in the limited try out, 0.827 and 0.866 in the main testing, and 0.783 and 0.838 in the extended testing for the latent subvariables Obligation and Endurance.All those CR and CA scores likewise show the levels above 0.70.
The results of the analysis show, however, that the Construct of Decency has only one CR score and only one CA score in more than one instance.They are CR scores of 0.968 in the limited try out, 0.966 in the main testing, and 0.960 in the extended testing as well as CA scores of 0.956 in the limited try out, 0.953 in the main testing, and 0.945 in the extended testing for the latent sub-variable of Politeness.All those CR and CA scores, however, still show the levels above 0.70.
The results of the analysis show that the construct of Care has two CR scores and two CA scores in more than one instance.They are CR scores of 0.880 and 0.922 in the limited try out, 0.887 and 0.904 in the main testing, and 0.860 and 0.895 in the extended testing as well as CA scores of 0.818 and 0.887 in the limited try out, 0.808 and 0.857 in the main testing, and 0.754 and 0.844 in the extended testing for the latent sub-variables Attentiveness and Appreciativeness.All those CR and CA scores again show the levels above 0.70.
The results of the analysis show that the construct of Self-Confidence has two CR scores and two CA scores in more than one instance.They are CR scores of 1.000 and 0.954 in the limited try out, 1.000 and 0.939 in the main testing, and 1.000 and 0.911 in the extended testing as well as CA scores of 1.000 and 0.935 in the limited try out, 1.000 and 0.902 in the main testing, and 1.000 and 0.853 in the extended testing for the latent subvariables of Assertiveness and Tenacity.All those CR and CA scores again show the levels above 0.70.

Conclusion
The result of this research is the instrument of affective assessment of primary school students and a manual book for users.The convergent validity of the constructs of Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence meets the criterion of the loading factor scores of their indicators being greater than or equal to 0.70 and there being no cross-loading score above 0.50.
The discriminant validity of the constructs of Honesty, Discipline, Decency, Responsibility, Care, and Self-Confidence is acceptable because the difference between the AVE score and the score of the correlation between the latent sub-variables in the same column is sufficiently great for each construct.The reliability of the constructs of Honesty, Discipline, Responsibility, Decency, Care, and Self-Confidence meets the criterion of their scores of composite reliability and Cronbach's alpha coefficient of reliability in the limited try out, the main testing, and the extended testing being greater than or at the minimum equaling 0.70.The results of the tests on the model of affective evaluation of the social competence of primary school students indicate that it meets the criteria for convergent validity, determinant validity, reliability, and fitness.

Suggestion
Indicators for research variables had better be more than five in number per latent sub-variable to avoid the running out of indicators because of mortality in the limited try out, the main testing, and the extended testing.The choice of the program for data analysis had better fit the type of data obtained from the field and the data could not be forced to meet the requirements for analysis which are demanded by a program of analysis.The program of analysis named SEM PLS Warp 3.0 could analyze data which are not normal like those the researcher has, so that it has made possible the completion of this research report.
The elicitation of data from the objects of research has not been as easy as it has been thought because of the behavioral dynamics of the human individuals involved as research subjects.Emotional closeness between the researcher and the research subjects had better be built up before the research is in progress.

Figure 1 .
Figure 1.Steps of model development

Table 1 .
Data of the construct of Honesty

Table 2 .
Convergent validity of the construct of Honesty

Table 3 .
Instrument reliability of the construct of Honesty

Table 4 .
Data of the construct of Discipline

Table 5 .
Convergent validity of the construct of Discipline

Table 6 .
Instrument reliability of the construct of Discipline Model of affective assessment of primary school... -31 Amir Syamsudin, Budiyono, & Sutrisno

Table 7 .
Data of the construct of Responsibility

Table 8 .
Convergent validity of the construct of Responsibility

Table 9 .
Instrument reliability of the construct of Responsibility

Table 10 .
Data of the construct of Decency No out-ayer data, but SEM PLS Warp PLS 3.0 analysis program reduces from 887 to 589

Table 11 .
Convergent validity of the construct of Decency

Table 12 .
Instrument reliability of the construct of Decency Model of affective assessment of primary school... -33 Amir Syamsudin, Budiyono, & Sutrisno

Table 13 .
Data of the construct of Care

Table 14 .
Convergent validity of the construct of Care

Table 15 .
Instrument reliability of the construct of Care

Table 16 .
Data of the construct of Self-Confidence

Table 17 .
Convergent validity of the construct of Self-Confidence

Table 18 .
Instrument reliability of the construct of Self-Confidence Model of affective assessment of primary school... -35

Table 19 .
Convergent validity of Social Competence