Page 1 of 15

287

Advances in Social Sciences Research Journal – Vol.7, No.8

Publication Date: August 25, 2020

DOI:10.14738/assrj.77.8854.

Effiyaldi, & Murniarti, E. (2020). Analysis Of Confirmatory Factors To Measure Public Trust In The Population Document Service Of

Population Department Of Jambi City. Advances in Social Sciences Research Journal, 7(8) 287-301.

Analysis Of Confirmatory Factors To Measure Public Trust In The

Population Document Service Of Population Department Of Jambi

City

Effiyaldi

STIKOM Dinamika Bangsa-Jambi

Erni Murniarti

Universitas Kristen Indonesia

ABSTRACT

This study aims to identify indicator variables that can measure

organizational transparency variables, work discipline and service

quality to public trust in administering residence documents of the

Department of Population of Jambi City. The research used

Confirmation Factor Analysis method. This study found that Chi-square,

RMSEA, GFI, NFI, CFI, IFI, RMR models have moderate fit. This means that

organizational transparency, work discipline and service quality affect

public confidence in the demographic document service of Jambi

residence. The results of this study can be used as a consideration and

evaluation of demographic document services for the community for

improvement and increased public confidence in the future.

Keywords: CFA, Chi-square, Factor, Public Trust.

INTRODUCTION

Jambi City as the center of Jambi provincial government is a gathering place for citizens with various

backgrounds in life. The city of Jambi, the majority of whom are migrants, mostly work in

government, private sector and trade. This causes the mobility of Jambi residents. High mobility will

result in interrelation between one part of the work with other work, between one institution and

another institution in order to expedite the objectives to be achieved from the work. Most of the

work involved is related to administrative activities, especially population administration. This

conditions makes the documents of population a necessity for the citizens of the city of Jambi. So

the availability of population documents will facilitate their activities. This has an impact on the

increasing number of Jambi residents who are taking care of obtaining population documents.

However, ironically, people feel reluctant to take care of the population documents.

There is an imbalance in population document services in the city of Jambi, including; people who

want to take care of population documents after feeling urged about their interests in connection

with the obligation to attach population documents as administrative requirements. People tend to

use the services of the officer to take care of population documents. There is an impression that the

cost of obtaining a population document is unclear and not transparent, there is no certainty when

the settlement of the population documents is administered, the work process is convoluted

Page 2 of 15

288

Advances in Social Sciences Research Journal (ASSRJ) Vol.7, Issue 8, August-2020

(Survey: 2011). Ironically, on the one hand, the community demands fast, transparent, and easy

services. Even the government has tried to improve services, but public service affairs are still felt

as unpleasant. Service users are often faced with so much uncertainty when dealing with

bureaucracy. It is uncertain when the output of a service can be obtained/completed.

Based on data from BAPPEDA [Development Planning Agency at Sub-National Level] Kota Jambi,

the data is processed from BPS, BKBKS, Religion Office (BPS Jambi City: 2010), that the total

population of the city of Jambi is 529,118 people spread over eight sub-districts, namely sub- districts; Kota Baru, Jambi South, Jelutung, Jambi Market, Telanaipura, Lake Teluk, Pelayangan, and

Jambi East. However, when more and more residents of the city of Jambi took care of obtaining

population documents, problems arose; starting from the attitude of officers who are indifferent,

less friendly, look down on the community, how to work is too mechanical, too long waiting times,

too strict on procedures and attitude of throwing responsibilities (Effiyaldi; 2015: 7).

This study aims to identify indicator variables that can measure the variables of organizational

transparency, work discipline, and service quality towards public trust in managing population

documents of the City of Jambi Population Office.

THEORITICAL REVIEW

Confirmatory factor analysis is a factor analysis technique which is a priori based on theories and

concepts that are known to be understood or predetermined (Hidayat: 2014), Confirmatory Factor

Analysis (CFA) is one of the multivariate analysis methods that can be used to confirm whether the

measurement model is built in accordance with what was hypothesized (Maiyanti; et all: 2008).

Factor analysis is used to reduce data, by finding relationships between variables that are mutually

independent (Stapleton, 1997 in Anonymous; 2008: 5), which are then collected in fewer variables

to determine the structure of latent dimensions (Garson; 2007 in Anonymous; 2008: 5), called the

factors. This factor is a new variable, also called a latent variable, a constructed variable, and has an

unobservable nature.

In the analysis of conformational factors, a researcher has a concept in advance of a hypothesis

based on the concept of structural factors. Then made a number of factors that will be formed, as

well as what variables are included in each factor that is formed and it is definitely the goal. The

formation of a confirmatory factor (CFA) intentionally based on theory and concepts, in an effort to

obtain new variables or factors that represent several items or sub-variables, which is observable

variables (Hidayat: 2014).

Factor Analysis of a latent variable is measured based on several indicators that can be measured

directly. The difference between CFA First Order and Second-Order CFA is that the Second Order

CFA latent variable is not measured directly through the assessment indicators but through other

latent variables (Fernanda, J.H: 2009 in Sari and Trijoyo: 2011; 1). Generally, there are 3 categories

of identification in a simultaneous equation that is Unidentified where the estimated number of

parameters (t) is greater than the amount of known data (s / 2), the data is the variance and

covariance of the observed variables. Identification Just identified by the criteria t = s / 2. And Over

Identified identification is with the criteria t ≤ s / 2 (Fernanda, J.H: 2009 in Sari and Trijoyo: 2011;

2). So the purpose of confirmatory factor analysis is to statistically confirm the model that the

researcher has built (Education Statistics: 2009). Confirmatory factor analysis uses invariant to

Page 3 of 15

URL: http://dx.doi.org/10.14738/assrj.78.8854 289

Effiyaldi, & Murniarti, E. (2020). Analysis Of Confirmatory Factors To Measure Public Trust In The Population Document Service Of Population Department Of

Jambi City. Advances in Social Sciences Research Journal, 7(8) 287-301.

scale and correlation or covariance matrix in estimating the structural factors. However, in theory,

the factor estimation uses the maximum likelihood procedure (Widhiarso: 2004). Therefore,

planning analysis is driven by the theory of the relationship between observed variables and

unobserved variables (Schreiber, et al: 2006). Furthermore, Schreiber, et al (2006) said, that when

Confirmatory Factor Analysis (CFA) was carried out, researchers used a hypothetical model to

estimate population covariance matrices compared with observed covariance matrices.

Technically, the researcher wants to minimize the difference between the estimated matrix and the

observed matrix.

In the confirmatory factor analysis, there is no need to convert the data into standardized data. The

data preparation is only when setting variables, some data are observable/manifest variables, some

are latent variables. The purpose of this factor analysis is to explain and illustrate by reducing the

number of parameters available. For the variable reduction stage to a further stage, in Confirmatory

factor analysis is known as second-order factor analysis. This factor analysis not only reduces

observations to latent extracts but also reduces the resulting latent extracts to other latent extracts

(Widhiarso: 2004). In the Confirmatory Factor Analysis, latent variables are considered as causal

variables (independent variables) that underlie indicator variables (Ghozali, 2003).

The very basic objectives of confirmatory factor analysis are: first to identify the relationship

between variables by conducting a correlation test. The second objective is to test the validity and

reliability of the instrument. In testing the validity and reliability of instruments or questionnaires

to obtain valid and reliable research data with confirmatory factor analysis. So the purpose of

confirmatory factor analysis is to statistically confirm the model that the researcher has built

(Education Statistics: 2009). In general, the steps to do a factor analysis are; 1. Model specifications.

2. Identification of the model. 3. Estimated model. 4. Testing the model. 5. Modification of the model

(Lewis: 2017)

Several measures of model suitability are often used to assess the feasibility of a model (Bollen,

1989 in Maiyanti; et al: 2008); Test χ2; the model is good if the χ2 test is not real at any particular

level. GFI (Goodness of Fit Index); that based on common practice, the feasibility of a model is that

the GFI value is greater than 0.90 and the maximum value is 1 (Sharma, 1996). AGFI (Adjusted

Goodness of Fit Index); that a model can be said to be good if the AGFI value is greater than 0.80 and

the maximum value is 1 (Sharma, 1996). RMSEA (Root Mean Square of Error Approval); if RMSEA

≤ 0.08, in general, the model is already representing the actual data (Sharma, 1996).

Some previous studies include; Efendi and Purnomo's research (2013: 106), confirmatory factor

analysis is used to find out the indicators that contribute greatly to the traffic awareness survey

with the parameter estimation method is the maximum likelihood estimation (MLE) method.

Ersalora Research (2013) with the title Confirmatory Factor Analysis on the Tourism Attraction of

Muara Jaya Curug in Majalengka Regency. Based on the analysis of the Muara Jaya waterfall tourist

attraction divided into two criteria, there are good and moderate. Natural factors, religious factors,

recreational facilities, and health facilities are good criteria. While the criteria are being included

factors from socio-culture, history, shopping facilities, infrastructure, and food and accommodation

facilities. Research Rachmawati, et al (2014: 74) which examines Confirmatory Factor Analysis of

the Indonesian Intermediate Collective Intelligence Test (TIKI-M), that each TIKI-M sub-test

measures aspects that should be measured according to the construct when viewed from

Page 4 of 15

290

Advances in Social Sciences Research Journal (ASSRJ) Vol.7, Issue 8, August-2020

standardized loading and t-value of each sub-test for the intelligence factor measured. Research

Seok, et al (2016), a study that tested employees' trust in their supervisors. In this study using

confirmatory factor analysis to examine the dimensions of Employee Confidence. This analysis is

carried out with Structural Equation Modeling to assess the suitability of the model. Besides, the

model's reliability and validity was also measured by involving 514 randomly selected employees

from the public and private sector organizations in Kota Kinabalu, Sabah, Malaysia. The findings of

this study can help improve productivity in an organization by increasing trust and building

relationships between employees and employers.

Research by Naveed, et al (2017) who identified and used nine dimensions to measure

organizational change, and 380 bank managers. To adjust the dimensions and their contribution to

the main construction of the first order and the second confirmatory factor analysis uses. The

results show that processes, strategies, attitudes, structures, culture and technology are the main

predictors of organizational change.

METHOD

This research uses descriptive research method. Data in this study were analyzed using

Confirmatory Factor Analysis (CFA) using LISREL 8.80. The general models used in confirmatory

factor analysis are as follows 1; (Bollen, 1989):

x = ΛXξ + δ (1) (1)

With:

x = is a vector for q x 1 indicator variables

ΛX = is a matrix for the loading factor (λ)

coefficient which shows the relationship of X with ξ size q x n

ξ (ksi) = is a vector for latent variables of size n x 1

δ = vector for measurement error measuring q x 1

If the data are ordinal scale, the polychoric correlation matrix is more suitable for estimating model

parameters. To get the polychoric correlation there are adjustments to the linear variables (Wirda:

2002). For example, the C and D categories are considered to be related to the continuous variables

that are X and Y, by: C = ci if γi-1 ≤ X <γi, i = 1, 2, 3, L, r D = dj if τj-1 ≤ Y <τj, j = 1, 2, 3, L, s Where γi

and τj are threshold parameters with γo = τo = - ∞ and γi = τj = ∞. The threshold parameters and

ordinal variable values are taken monotonously increasing γ1 <γ2 <L <γr-1 and c1 <c2 <L <cr. With

the same analogue it applies to τj and dj. To calculate the polychoric correlation matrix, you can use

the Data Prelis 2.30 program. Whereas the Data Prelis 2.30 program is in LISREL version 8.30 or

version 8.50. LISREL (Linear Structural Relationship) is a computer software package used to

operate structural equation modeling methods (Jöreskog and Sär bom, 1996).

In the Confirmatory Factor Analysis with the maximum likelihood method, in the process of

estimating the parameters of the model using a variety structure, which basically removes the

charter matrix Σ (matriks) with the sample matrix S or polyphoric correlation matrix (Σ). Suppose

the fitting function is stated with F (S, Σ), which is a function that depends on S and Σ. If the

parameter parameter θ is substituted in Σ, then Σ (θ) is obtained. The value of the fitting function

on θˆ is F (S, Σ (θˆ)) (Bollen; 1989 in Maiyanti; et al: 2008).

Page 5 of 15

URL: http://dx.doi.org/10.14738/assrj.78.8854 291

Effiyaldi, & Murniarti, E. (2020). Analysis Of Confirmatory Factors To Measure Public Trust In The Population Document Service Of Population Department Of

Jambi City. Advances in Social Sciences Research Journal, 7(8) 287-301.

After testing the validity, the first step in interpreting the resulting Confirmatory Factor Analysis

model is to conduct a feasibility assessment of the model itself, whether the model is feasible or not.

In assessing a model so far there is no single measure to assess the feasibility of a model (Maiyanti;

et al: 2008). At least three authors were found who suggested using at least three model feasibility

tests (Kline: 1998), (Sharma: 1996 in Maiyanti; et al: 2008).

The resulting model must be tested through structural model equation testing using Lisrel 8.30

software (Joreskog and Sorbom, 1989). The model analysis method used is the maximum likelihood

extraction method with Oblimin rotation to confirm the dimensions of the originating instruments

(Nimako: 2012). In the first analysis, the model obtained did not meet the model accuracy index. By

using the modification indices recommendations, it is found that to get a fit model, the researcher

must link the measurement error that is realized in the error covariance measurement between the

openess factor and the extrovert.

After modification, the model that meets the accuracy is found. The results are presented with a

factor loading value which is considered strong (Garson 2007; Kline 2005 in Nordin; 2012). GFI

(goodness fit index) is an index of the accuracy of the model in explaining the model being compiled.

CFA analysis is done by summing the scores of each as observed variables. But it can be found to

have negative values that are not acceptable. According to Kline (2005) in Nordin (2012),

estimation of negative variance or unacceptable values, the revised model, and a suitable index

show better results.

In this study, the population is all households that are domiciled in the city of Jambi, in this case,

each household is represented by a household member (husband/wife/adult child). The number of

households/heads of households in the city of Jambi based on the results of the 2010 population

census is 126,829 households/heads of households spread across eight districts. Distribution of

Households / Households can be seen in the following table;

Table 1. Average Members of District Households in Jambi City

Sub- distric

the number of family

heads

The number of

population

Average

Household

Member

Kota Baru 33.245 137.856 4,15

Jambi Selatan 29.678 123.201 4,15

Jelutung 14.578 60.141 4,13

PasarJambi 3.286 12.988 3,95

Telanaipura 22.823 92.603 4,06

DanauTeluk 2.310 11.803 5,11

Pelayangan 2.483 12.895 5,19

Jambi Timur 18.426 77.631 4,21

Total 126.829 529.118 4,18

Source: 2010 City of Jambi Population Census, Central Statistics Agency of Jambi

City.

Determination of the size of the sample used refers to the determination of sampling by Joreskog,

K.G (1999), that the size of the sample used is at least five times the number of indicators. In this

Page 6 of 15

292

Advances in Social Sciences Research Journal (ASSRJ) Vol.7, Issue 8, August-2020

study the number of indicators used was 64 indicators (18 + 16 + 14 + 16), then the sample size

used was 64 people. This number is very adequate because a minimum sample of 30 is considered

a large sample size for statistical analysis (Cooper and Schindler, 2006). Weedaman & Thompson

(2003) argue that the RMSEA value is relatively independent of the sample size. Fan & Sivo (2007)

also stated that NFI, GFI, and AGFI fit values have high sensitivity to sample size. Similarly, Marsh

(1988) in their study found that RMR, GFI, and AGFI values were positively influenced by sample

size. As a result, questions in a single-factor structure are accepted (Evrekl, et al: 2010).

In this study, the questionnaire was developed using a Likert scale. This scale is used to measure

the level of agreement or disagreement of respondents to a series of statements that measure an

object (Istijanto: 2010: 87). Before the questionnaire is used to collect data, the questionnaire is

first tested for validity and reliability. In the Confirmatory Factor Analysis the hypothesized model

must be valid which refers to the ability of an indicator to measure what is actually wanted to be

measured (Supranto, J: 2004). Validation is a process carried out by the composer or user of the

instrument to collect data empirically to support the conclusions generated by the instrument score.

Validity is the ability of a measuring instrument to measure its measurement goals (Ahiri; 2009).

The validity of indicators in measuring latent variables is assessed by testing whether all loading

(λi) is real by using t-test for a certain level of confidence α. For this reason, further confirmation is

needed, namely checking its validity and reliability. This can be done with Factor Analysis, so it is

called the Confirmatory Factor Analysis. So in principle we will only confirm based on existing

theories or concepts on the accuracy (valid and reliable) of the instruments made (Arisanti: 2010).

While reliability is the consistency of an instrument measuring something to be measured reliability

indicates the extent to which the results of measurements with the tool can be trusted. Or reliability

is the proportion of the diversity of test scores caused by systematic diversity in the test taker

population (Ahiri: 2009: 17). The reliability test of the instrument aims to find out and guarantee

that an instrument/questionnaire is indeed reliable to be used in collecting data. Therefore,

reliability is an index that shows the extent to which a measuring tool can be trusted or reliable

(Wiersma: 1986 in Margono: 2013)

In this study, the total population was 25,366 people/family, while the number of respondents was

320 people spread across eight sub-districts within the city of Jambi. The following is general

information about the characteristics of the respondents;

Page 7 of 15

URL: http://dx.doi.org/10.14738/assrj.78.8854 293

Effiyaldi, & Murniarti, E. (2020). Analysis Of Confirmatory Factors To Measure Public Trust In The Population Document Service Of Population Department Of

Jambi City. Advances in Social Sciences Research Journal, 7(8) 287-301.

Tabel 2. CHarateristics and Distribution of Respondents

No Category Total No Category Total

1

Level of Education

3.

Agama

Senior high school 107 Moslem 235

Diploma 90 Catholic/Protestant 49

S1 97 Hindu 28

S2 26 Budha 18

Total 320 Total 320

2

Profession

4

Gender

Civil Servant 57 Male 185

Private/ 66 Female 135

Teacher/ Lecturer 59 Amount 320

Farmer 44

5

Age

House wife 43 18 – 25 years 65

College student 51 26 – 40 165

> 40 90

Total 320 Total 320

Source; Data Processed.

The following is the operationalization of research variables in this study;

Table 3. Description of Public Trust Variables in this study

Indikator Item

Internal factor (personal) Z.1.1, Z.1.2

Eksternal factor (institution characteristik) Z.2.1, Z.2.2, Z.2.3, Z.2.4

Integrity Z.3.1, Z.3.2

Competence Z.4.1, Z.4.2, Z.4.3, Z.4.4

Consistensy Z.5.1, Z.5.2, Z.5.3, Z.5.4

Loyalty Z.6.1

Openness Z.7.1

Table 4. Description of Variable Data on Public Service Quality (Y)

Indikator Item

Tangibles (penampilan) Y.1.1, Y.1.2, Y.1.3, Y.1.4, Y.1.5

Empathy (kemauan memberi layanan) Y.2.1, Y.2.2

Reliability (Kehandalan) Y.3.1, Y.3.1, Y.3.1

Responsiveness (kesediaan membantu) Y.4.1, Y.4.2

Assurance (jaminan) Y.5.1, Y.5.2, Y.5.3, Y.5.4

Page 8 of 15

294

Advances in Social Sciences Research Journal (ASSRJ) Vol.7, Issue 8, August-2020

Table 5. Description of Employee Disciplinary Variable Data (X2)

Indicator Item

Preventive X2.1.1, X2.1.2, X2.1.3

Corrective X2.2.1, X2.2.2, X2.2.3

Job goals and job abilities X2.3.1, X2.3.2

As an example X2.4.1, X2.4.2

Without ulterior motive X2.5.1

Justice X2.6.1

Firmness X2.7.1

Human relations X2.8.1

Table 6. Description of Data on Organizational Transparency Variables (X1)

Indicator Item

Mechanism X1.1.1, X1.1.2, X1.1.3, X1.1.4

Information access channel X1.2.1, X1.2.2, X1.2.3

Media/Tools/Material/Complaint X1.3.1, X1.3.2, X1.3.3, X1.3.4

Public Right of Information X1.4.1, X1.4.2, X1.4.3, X1.4.4, X1.4.5

RESULT

The goodness of fit test of the model for the confirmatory factor analysis of the public trust variable

is obtained as follows;

Table 7. GOF Model 1 Testing Results Public Trust

Goodness of fit

(GOF) Indeks Cut off value Hasil output Description

Chi-square P ≥ 0.05 372.44 (p=0.0) not fit

RMSEA ≤0,08 0.074 fit

GFI ≥0,9 0.89 not fit

NFI ≥0,9 0.89 not fit

CFI ≥0,9 0.93 good fit

IFI ≥0,9 0.93 good fit

RMR ≤ 0.05 1.13 not fit

From the table above, the chi-square value and its probability (p) <0.05. This shows the model is

not good. Next, an analysis of each of the indicators that constitute public trust is provided in the

standardized estimate value graph and the model's t-value as follows:

Page 12 of 15

298

Advances in Social Sciences Research Journal (ASSRJ) Vol.7, Issue 8, August-2020

Figure 7 and Figure 8. Standardized Value of Organizational Transparency Analysis and T-Count

Value Confirmatory Analysis of Organizational Transparency

From this figure, it can be seen the results of the estimated parameter relationship between latent

variables and indicator variables.

DISCUSSION

Based on table 7 about the GOF test results of the public trust variable, it is known that the chi- square value and its probability (p) <0.05. This shows the model is not good. However, please note

that the chi-square value is very sensitive to the number of samples so that another fit test is needed.

From several other Goodness of fit index criteria, the model shows that the model is eligible, but

there are also some Goodness of fit index criteria, which shows the model does not meet the

requirements. Based on this, it can be argued that the model has a moderate fit.

Based on Figure 1 and Figure 2. Standardized Value of Confirmatory Analysis of Service Quality and

T-Count Value of Confirmatory Analysis of Service Quality that is above, the chi-square value, and

its probability (p) <0.05. This shows the model is not good. T-count value shows that all indicators

have a t-value greater than t-table of 1.96 (α = 5%). which shows that the indicators jointly present

a unidimensional variable for public trust.

Based on table 8. GOF Model 1 Testing Results of Public Service Quality, from the table above, the

value of chi-square and its probability (p) <0.05. This shows the model is not good. However, please

note that the chi-square value is very sensitive to the number of samples so that another fit test is

needed. From several other Goodness of fit index criteria, the model shows that the model is eligible,

but there are also some Goodness of fit index criteria, which shows the model does not meet the

requirements. Based on this, it can be argued that the model has a moderate fit.

Based on Figure 3 and Figure 4. Standardized Value of Confirmatory Analysis of Service Quality and

T-Calculate Value of Confirmatory Analysis of Service Quality. Based on the t-count value shows that

all indicators (16 indicators) have a t-value greater than the t-table of 1.96 (α = 5%). which shows

that the indicators jointly present a unidimensional variable for public service quality (Y).

Based on table 9. The results of the Gof Model 1 employee work discipline test found that from the

above table, the value of chi-square and its probability (p) <0.05. This shows the model is not good.

However, the chi-square value is very sensitive to the number of samples so it needs another fit test.

From several other Goodness of fit index criteria, the model shows that the model is eligible, but

Page 15 of 15

URL: http://dx.doi.org/10.14738/assrj.78.8854 301

Effiyaldi, & Murniarti, E. (2020). Analysis Of Confirmatory Factors To Measure Public Trust In The Population Document Service Of Population Department Of

Jambi City. Advances in Social Sciences Research Journal, 7(8) 287-301.

Rachmawati, et al (2014). Confirmatory Factor AnalysisTes Inteligensi Kolektip Indonesia Tingkat Menengah(TIKI- M).Jurnal Psikologi Pendidikan dan Perkembangan Vol. 3, No. 1, April 2014. Fakultas Psikologi Universitas Airlangga.

Sari dan Trijoyo (2011). Confirmatory Factor Analysis untuk Mengukur Unidimensional

Indikator Performa Pengelolaan Lingkungan Hidup dalam Survei Publik. Institut Teknologi Sepuluh Nopember (ITS).

Seok, et al (2016). Confirmatory Factor Analysis of Employee Trust Scale. IRA-International Journal of Management &

Social Sciences ISSN 2455-2267; Vol.04, Issue 03 Pg. No. 536-547 Institute of Research Advances.

Schreiber, et al (2006). Reporting Structural Equation Modeling And Confirmatory Factor Analysis Results: A Review”.

The Journal Of Educational Research. Department Of Foundations & Leadership, Duquesne University, 600 Forbes

Avenue, Pittsburgh.

Sharma, S. (1996). Applied Multivariate Techniques.John Wiley & Sons, Inc. New York.

Statistik Pendidikan (2009). Analisis Faktor Konfirmatori. Forum Penelitian Behavioral

https://statistikpendidikan.wordpress.com/2009/01/12/analisis-faktor-konfirmatori/.

Supranto, J, (2004). Analisis Multivariat, Arti dan Interpretasi. Edisi 1. Rineka Cipta. Jakarta.

Widaman, K.F. & Thompson, J.S (2003).On Specifying The Null Model For Incremental Fit Indices In Structural

Equation Modeling.Psychological Methods, 8, 16-37.

Widhiarso, Wahyu. 2014. “Evaluasi Faktor Dalam Big Five : Pendekatan Analisis Faktor Konfrmatori”. Tesis Program

Pascasarjana Universitas Gadjah Mada Yogyakarta.

Wirda,Y(2002).“Suatu Kajian tentang Persamaan Struktural Linier dengan Variabel Laten dan Penerapannya pada

Data Pendidikan”. Tesis.Program pasca sarjanaIPB. Bogor.