CC BY
2
UDC 331; DOI 10.18551/rjoas.2023-07.10
THE IMPACT OF OVERALL QUALITY AND SELF-EFFICACY ON CONATIVE USE WITH USER'S SATISFACTION AS A MEDIATION VARIABLE
Kusuma Gracias Dessy*, Rosidi, Iqbal Syaiful
Faculty of Economics and Business, University of Brawijaya, Malang, Indonesia *E-mail: dessykusuma8@student.ub.ac. id
ABSTRACT
The Ministry of Finance of the Republic of Indonesia built a web-based integrated system called the Agency Level Financial Application (SAKTI) to provide benefits: increased integration, efficiency, planning, implementation and accountability of the state budget. This study aims to determine and explain user satisfaction as a mediating variable for the causal relationship between overall quality and self-efficacy towards conative use of information systems. SAKTI users were a primary data that collected directly by distributing questionnaires. The questionnaire includes overall quality, self-efficacy, user satisfaction and conative use. This research focuses on the central work unit of the Ministry of Public Works and Public Housing. The population in this study were 552 people. The sampling technique used in this study was stratified random sampling consisting of 3 levels of users: operators, validators and approvers with a total of 247 respondents. Most respondents were male with a total of 125 people or 50.6%, while female were 49, 4%. In addition, most of the respondents have a bachelor's degree (131 people or 53%). Data analyzed using Structural Equation Modeling (SEM) - Partial Least Square (PLS). The findings show that overall quality and self-efficacy have a direct and independent effect on conative use of SAKTI. The results of this study also examine the relationship between overall quality, self-efficacy and conative use through user satisfaction.
KEY WORDS
Conative use, user satisfaction, self-efficacy, overall quality.
The e-government system is considered as part of a modern public administration system (Torres et al., 2005). The development of e-government in government is very diverse. There are three general types of e-government services, namely government to government (G2G), government to citizens (G2C) and government to businesses (G2B) (Wang & Liao, 2008). This study raises the issue of implementing an Information Technology (IT)-based Accounting Information System (AIS) which is included in G2G, namely the application of an information system owned by the central government for managing state finances.
The application of e-government in the financial sector is to adopt the Integrated Financial Management Information System (IFMIS). The application of IFMIS in a number of agencies can provide financial performance in the form of efficiency, effectiveness, accountability, transparency and security of comprehensive data management and reporting. The Ministry of Finance has implemented the State Treasury and Budget System (SPAN) which is part of the IFMIS. Within the scope of work units, the implementation of IFMIS is manifested in the form of improving state financial management business processes using integrated applications. The application is embodied in a system known as the Agency Level Financial Application System (SAKTI).
In using mandatory information systems, employees routinely use SAKTI only at the basic level and in accordance with the normal method determined by the regulations/modules for using SAKTI. This standard and routine use is surface level utilization (Agarwel, 2000) which makes it difficult to reach the full potential of the system. However, under mandatory circumstances, employees still retain the discretion to determine whether and to what extent, use SAKTI to support their duties (Silver, 1990). Meanwhile, the complexity of SAKTI allows users to utilize it at different levels. In contrast to previous
research, this study developed the Information System Success Model by Delone and Mclean (2003) by adding self-efficacy variables to explain the factors that influence conative use because this research uses an approach to the SAKTI background condition which has a mandatory nature.
A new perspective that can be used in measuring satisfaction using mandatory information systems is conative use. The conative use construct can be used when the information system used is mandatory (Kwahk et al., 2018). In contrast to the DeLone and McLean models which use actual use as one of the constructs, this study uses the conative use construct. Actual use is more interpreted in terms of "how long" to use the information system, so it is deemed inappropriate and will be biased if used in this study, especially in the implementation of SAKTI which is mandatory and will only be fully operational and web-based in early 2022. Conative terminology refers to willingness, namely the willingness of a person to do something proactively (Gerdes & Stromwall, 2008). SAKTI users can creatively and effectively continuously improve their competence and knowledge of APBN financial management in using SAKTI. Evaluation of conative use of SAKTI is considered important so that the decision to change the system becomes easier to increase organizational productivity.
From the piloting period to full national implementation in 2022, the implementation of SAKTI has encountered obstacles and challenges. These obstacles and challenges are an interesting thing in the application of a technology and information system (Eggers & Bellman, 2015). Problems with infrastructure/servers and an unstable internet network are still the main problems in application operation. Obstacles in sending OTP (One Time Password) are also a new problem where sometimes OTP is not sent. Coordination and communication between SAKTI users is also needed so that the budget execution business process can run smoothly. A specific problem at the Ministry of PUPR is that treasury officials who are SAKTI users still occupy structural positions and carry out other activities, so they cannot use the time for approval so that the OTP sent is no longer valid.
Problems with the implementation of SAKTI were found in the downstream part of APBN management, namely financial reporting, which is still about features or menus on SAKTI which cannot be used to download all accrual account balances to process balance sheet, LO and LPE data per work unit. In addition, the financial reports produced also do not produce the actual conditions of the previous financial management business processes, such as there is still a minus ceiling even though the work unit has updated it and abnormal balances still appear on the trial balance even though improvements have been made.
Challenges and obstacles in the development of information systems will certainly make system developers always improve the quality and functionality of new systems and take advantage of their development (Wang & Lai, 2014). System quality, information quality, and service quality have been widely studied as the second construct (Aldholay et al., 2018). The overall quality of a system is certainly a reason for users to feel satisfied. Self-efficacy is defined as an individual's assessment of success in several computer domains and application-specific self-efficacy is defined as an individual's perception of success in using a particular application or system (Yi & Hwang, 2003). Confidence in self-efficacy in using the system plays a causal role in achievement and predicts effort and satisfaction in learning (McCormick & McPerson, 2003). This is also supported by research conducted by (Hong et al. 2016) which states that there is a positive and significant relationship between self-efficacy and satisfaction.
One of the most important things in implementing a new system is an acceptance process (Cooper & Zmud, 1990). There are several studies that have analyzed SAKTI's acceptability factors (Lucyanda, 2010; Nasrudin E & Widagdo Ari K, 2020; Prabowo, 2017). The dimension of user satisfaction is a dimension that is commonly used in the corporate sector. Ease of use of the system is important which has the potential to cause user rejection or reluctance to use SAKTI.
This research was conducted at the central Ministry of PUPR work units, totaling 69 spread across DKI Jakarta Province. The population scope was chosen because it is located in the capital city and can be used as a center for learning and knowledge about the state
treasury. The Work Unit of the Central Ministry of PUPR is a supervisor for work units in other regions.
Thus, the research questions are stated as follows:
• Does overall quality have positive impact towards user satisfaction?
• Does self-efficacy have positive impact towards user satisfaction?
• Does user satisfaction have positive impact towards conative use?
• Does user satisfaction mediate the impact of overall quality towards conative use?
• Does user satisfaction mediate the impact of self-efficacy towards conative use?
CONCEPTUAL FRAMEWORK AND HYPOTHESES DEVELOPMENT
The model starts with four variables and then is followed by the hypothesis development. This study applied DeLone and McLean model which is added by two other variables, Self-efficacy and conative use to discover the possible implementation of SAKTI nationally.
Figure 1 - Conceptual Framework
Hypotheses:
• H1: Overall quality have positive impact towards user satisfaction;
• H2: Self-efficacy have positive impact towards user satisfaction;
• H3: User satisfaction have positive impact towards conative use;
• H4: User satisfaction mediates the impact of overall quality towards conative use;
• H5: User satisfaction mediates the impact of self-efficacy towards conative use.
METHODS OF RESEARCH
This research uses a positivistic approach or commonly called quantitative research because this research uses a deductive approach which aims to test hypotheses . Based on the characteristics of the problem, this research is an associative research which is a research with problem characteristics in the form of a relationship between two or more variables (Sujarweni, 2022: 49). Based on the characteristics of the problem, this research is an associative research which is a research with problem characteristics in the form of a relationship between two or more variables (Sujarweni, 2022: 49).
The type of data used in this study is primary data that obtained from questionnaires distributed in the form of structured statements. A list of questions is given to employees who use the SAKTI application. The population in this study was all SAKTI users within the Central Ministry of Public Works and Public Housing (PUPR Ministry). There are three levels of SAKTI users who have different functions and tasks according to the user's authority and position, namely operator, validator and approver. There are at least 8 SAKTI users at every Work Unit of the Ministry of PUPR, consists of PPK, KPA and PPSPM, budgeting officers, expenditure treasurers, financial management staff, inventory officers/managers of State Property (BMN) and financial reporting officers. The number of Central Work Units of PUPR Ministry is 69, so the population in this study is 552 SAKTI users. Sampling in this study used the proportionate stratified random sampling method, namely taking samples from the population randomly with proportional strata. By using the slovin formula, the number of research samples is obtained as much as 232 SAKTI users in the Central Work Unit of the PUPR Ministry with the following details:
Table 1 - Total Research Sample
No User Level Total Population Total Sample
1 Operator 345 145
2 Validator 69 29
3 Approver 138 58
Total 552 232
Source of data obtained in this study comes from primary sources. Primary data (primary data) refers to information obtained directly related to the variable of interest for a particular purpose of a study (Sekaran & Bougie, 2022: 130). Questionnaires were distributed using the Google form because the samples were spread across DKI Jakarta Province and were collected within two weeks of receiving the research permit and the questionnaires in each work unit. For control, the researcher will monitor the questionnaires that have been collected so that they can fulfill the desired sample.
KK1 KK2
SE
Figure 2 - Researh Model
The data analysis used is partial least squares (PLS) using the smart PLS application version 3.2.9. PLS is a multivariate analysis with several independent variables and dependent variables. PLS allows testing of measurements and structural models simultaneously (Abdillah & Hartono, 2015: 164). The SEM-PLS model was chosen based on several considerations. SEM-PLS does not require data assumptions when testing (Hair et al., 2014).
Overall Quality and Self-Efficacy are exogenous variables in this study. The 6 (six) indicators of the overall quality construct in this study were adapted from research (AlMamary et al., 2014; Djuitaningsih & Arifiyantoro, 2020). The self-efficacy construct uses 3 (three) indicators adapted from (Aldholay et al., 2018; Hong et al., 2016). User Satisfaction and Conative Use acts as an endogenous latent construct. The user satisfaction construct uses 4 (four) indicators adapted from (Amriani & Iskandar, 2019; Seddon & Kiew, 1996). Then the final measure of the study is the conative use construct using 3 (three) indicators adapted from research (Kwahk et al., 2018). The final step is to evaluate the model and test the hypothesis. Evaluation of the model in PLS is divided into evaluation of the outer model and the inner model. Evaluation of the outer model is carried out by measuring the value of the convergent, discriminant and reliability tests of validity. Evaluation of the inner model is done by looking at the value of the determinant coefficient.
RESULTS AND DISCUSSION
The measurement scale used in this study uses a Likert scale. A 5-point scale namely, Strongly Agree (SA): 5, Agree (A): 4, Undecided (U): 3, Disagree (D): 2, and Strongly Disagree (SD): 1. Respondents in this study were all financial managers who used the SAKTI application in the central PUPR Ministry work unit. A total of 274 incoming questionnaires. After re-examination, there were 27 because the respondents did not meet the criteria set by the researchers, namely being a work unit of the Central PUPR Ministry. So, the questionnaires that can be processed in this study are as many as 247 questionnaires. The following table summarizes the number of samples and the rate of return on the questionnaire:
Descriptive of Conative Usage Variable. SAKTI conative use shows a minimum score of 2 and a maximum score of 5. The standard deviation value of all indicators does not exceed the average value, so the data has a relatively low deviation. Respondents' assessment of the SAKTI conative usage variable with 3 question items was high. This can be seen from the average frequency distribution of 4.244 which means that respondents have a desire to use SAKTI proactively.
Table 2 - Descriptive Statistics of Conative Usage Variables
n/n Min Max Frequency of respondent's answer (F) and Percentage (%) Deviaton Standard
SS | S RR TS | STS Mean
F % F % F % F % F %
CU1 2 5 75 30% 154 62% 15 6% 3 1% 0 0% 4,219 0,604
CU2 2 5 78 32% 132 53% 33 13% 4 2% 0 0% 4,150 0.701
CU3 2 5 110 45% 118 48% 18 7% 1 0% 0 0% 4,364 0.634
Average Conative Use 4,244 -
Descriptive of Overall Quality Variable. Table 3 shows that in detail, the main criteria chosen by respondents in assessing the overall quality construct are SAKTI having good quality, which is indicated by a percentage of 64% on the KS3 indicator and reliable SAKTI which is indicated by a percentage of 61% on the KS6 indicator. Overall, the respondent's perception of the overall quality of the system was considered as very good.
Descriptive of Self-Efficacy Variables. Table 4 shows that in detail, the main criteria chosen by respondents in assessing the construct of self-efficacy is the ability to download and upload data on the SAKTI application which is indicated by a percentage of 55% on the ED3 indicator. Overall, the respondents' perceptions regarding the self-efficacy of SAKTI users were considered as very good.
Table 3 - Descriptive Statistics of Overall Quality Variables
n/n Min Max Frequency of respondent's answer (F) and Percentage (%) Deviaton Standard
SS S RR TS STS Mean
F % F % F % F % F %
SQ1 1 5 92 37% 136 55% 17 7% 1 0% 1 0% 4,283 0,643
SQ2 2 5 67 27% 139 56% 39 16% 2 1% 0 0% 4,097 0,672
SQ3 1 5 65 26% 157 64% 22 9% 2 1% 1 0% 4,146 0,632
SQ4 1 5 53 21% 125 51% 61 25% 7 3% 1 0% 3,899 0,775
SQ5 2 5 67 27% 139 56% 40 16% 1 0% 0 0% 4,101 0,663
SQ6 2 5 79 32% 151 61% 16 6% 1 0% 0 0% 4,247 0,583
IQ1 2 5 118 48% 114 46% 10 4% 5 2% 0 0% 4,397 0,665
IQ2 2 5 112 45% 119 48% 13 5% 3 1% 0 0% 4,377 0,643
IQ3 2 5 60 24% 123 50% 62 25% 2 1% 0 0% 3,976 0,725
IQ4 3 5 114 46% 122 49% 11 4% 0 0% 0 0% 4,417 0,576
IQ5 2 5 122 49% 107 43% 13 5% 5 2% 0 0% 4,401 0,683
SeQ1 1 5 101 41% 128 52% 14 6% 3 1% 1 0% 4,316 0,671
SeQ2 1 5 73 30% 134 54% 35 14% 4 2% 1 0% 4,109 0,726
Average Overall Quality 4,213 -
Table 4 - Descriptive Statistics of Self-Efficacy Variables
n/n Min Frequency of respondent's answer (F) and Percentage (%) Deviaton Standard
Max SS S RR TS STS Mean
F % F % F % F % F %
SE1 2 5 113 46% 122 49% 11 4% 1 0% 0 0% 4,405 0,595
SE2 2 5 97 39% 132 53% 16 6% 2 1% 0 0% 4,312 0.627
SE3 2 5 137 55% 100 40% 8 3% 2 1% 0 0% 4,506 0.603
Average Self-Efficacy 4,408 -
Table 5 - Descriptive Statistics of User Satisfaction Variables
n/n Min Max Frequency of respondent's answer (F) and Percentage (%) Deviaton Standard
SS S RR TS STS Mean
F % F % F % F % F %
US1 1 5 95 38% 136 55% 13 5% 2 1% 1 0% 4,304 0.643
US2 2 5 122 49% 108 44% 13 5% 4 2% 0 0% 4,409 0.666
US3 2 5 123 50% 106 43% 14 6% 4 2% 0 0% 4,409 0.672
US4 2 5 112 45% 117 47% 15 6% 3 1% 0 0% 4,368 0.653
Average User Satisfaction 4,373 -
Table 6 - Outer Loadings, AVE, and Communality Value
Variable Indicator Factor Loadings AVE Communality
Information Quality (IQ) IQ1 0,904 0,717 0,717
IQ2 0,916
IQ3 0,625
IQ4 0,855
IQ5 0,900
System Quality (SQ) SQ1 0,845 0,638 0,638
SQ2 0,762
SQ3 0,821
SQ4 0,736
SQ5 0,760
SQ6 0,858
Service Quality (SeQ) SeQ1 0,921 0.843 0.843
SeQ2 0,915
Overlall Quality (OQ) IQ 0,928 0.579 0.579
SQ 0,919
SeQ 0,850
Self-efficacy (SE) SE1 0,923 0,830 0,830
SE2 0,901
SE3 0,909
User Satisfaction (US) US1 0,877 0.841 0.841
US2 0,915
US3 0,933
US4 0,942
Conative Use (CU) CU1 0,868 0.684 0.684
CU2 0,827
CU3 0,784
Table 5 shows that in detail, the main criterion chosen by respondents in assessing the user satisfaction construct is that SAKTI can help work efficiently as indicated by a percentage of 50% on the KP3 indicator. Overall, the respondent's perception of system user satisfaction was considered as very good.
Model evaluation is done by testing the outer model and inner model. Testing of the outer model is carried out to determine the validity and reliability of the model. While testing the inner model is used as a predictor between latent variables.
The parameters used in the convergent validity test are factor loading values of more than 0.7, Average Variance Extracted (AVE) values of more than 0.5, and communality values of more than 0.5 (Hair et al. 2006); Abdillah & Hartono, 2015: 196). Indicators that have these conditions can be said to be valid and can be used in this study. The following are the results of the convergent validity test:
Table 7 - Outer Loadings, AVE, and Communality Value
Variable Indicator Factor Loadings AVE Communality
Information Quality (IQ) IQ1 0,904 0,717 0,717
IQ2 0,916
IQ3 0,625
IQ4 0,855
IQ5 0,900
System Quality (SQ) SQ1 0,845 0,638 0,638
SQ2 0,762
SQ3 0,821
SQ4 0,736
SQ5 0,760
SQ6 0,858
Service Quality (SeQ) SeQ1 0,921 0.843 0.843
SeQ2 0,915
Overlall Quality (OQ) IQ 0,928 0.579 0.579
SQ 0,919
SeQ 0,850
Self-efficacy (SE) SE1 0,923 0,830 0,830
SE2 0,901
SE3 0,909
User Satisfaction (US) US1 0,877 0.841 0.841
US2 0,915
US3 0,933
US4 0,942
Conative Use (CU) CU1 0,868 0.684 0.684
CU2 0,827
CU3 0,784
Through table 7, it can be seen that there is one indicator that has a factor loading value below 0.7, namely KI3. However, in this study, from the development of a measurement scale, a loading value of 0.5 to 0.6 is considered sufficient (Abdillah & Hartono, 2015: 206). In addition, it can be seen that the AVE value and communality value have fulfilled the rule of thumb stated by Hair et al. (2006) and Abdillah & Hartono (2015: 196) with a value of more than 0.5. Through this explanation, it can be shown that all indicators for each variable have fulfilled the convergent validity test.
Discriminant validity indicates that the constructs are different and must not have a high correlation (Abdillah & Hartono, 2015: 195). Rule of thumbs that must be met by looking at the AVE root value and cross loading. The AVE root value is required to be more than the latent variable correlation and the cross loading value is > 0.7. The following is the AVE root value and the correlation of other latent variables. Based on table 8, it can be seen that the AVE root value is more than the correlation values of other latent variables.
The next parameter is to see the cross loading value. Based on table 9, it can be seen that there is still a cross loading value <0.7. However, in this study, the development of a loading value measurement scale of 0.5 to 0.6 was considered sufficient (Abdillah & Hartono, 2015: 206). This shows that the cross loading value has fulfilled the rule of thumb for the discriminant validity testing section. Overall, by looking at the table above, it can be concluded that the indicators used have fulfilled the discriminant validity test.
Table 8 - AVE Root Value and Latent Variable Correlation
n/n SE IQ OQ SeQ US SQ CU
SE 0.911
IQ 0.686 0.847
OQ 0.702 0.928 0.761
SeQ 0.628 0.751 0.850 0.918
US 0.718 0.782 0.806 0.730 0.917
SQ 0.596 0.734 0.919 0.694 0.687 0.799
CU 0.605 0.594 0.651 0.561 0.623 0.603 0.827
Table 9 - Cross Loading Value
n/n SE IQ OQ SeQ US SQ CU
SE1 0,923 0,658 0,674 0,616 0,67 0,569 0,546
SE2 0,901 0,588 0,612 0,537 0,624 0,533 0,55
SE3 0,909 0,628 0,631 0,563 0,667 0,526 0,557
IQ1 0,653 0,904 0,851 0,707 0,729 0,676 0,533
IQ2 0,623 0,916 0,849 0,698 0,719 0,663 0,503
IQ3 0,345 0,625 0,58 0,374 0,425 0,508 0,308
IQ4 0,593 0,855 0,769 0,612 0,653 0,587 0,558
IQ5 0,639 0,900 0,847 0,73 0,735 0,66 0,577
OQ1 0,559 0,659 0,797 0,604 0,646 0,845 0,544
OQ2 0,412 0,511 0,671 0,495 0,46 0,762 0,486
OQ3 0,482 0,638 0,771 0,577 0,571 0,821 0,464
OQ4 0,375 0,524 0,676 0,543 0,507 0,736 0,415
OQ5 0,449 0,49 0,659 0,486 0,478 0,76 0,434
OQ6 0,554 0,669 0,808 0,607 0,608 0,858 0,538
OQ7 0,623 0,916 0,849 0,698 0,719 0,663 0,503
OQ8 0,345 0,625 0,58 0,374 0,425 0,508 0,308
OQ9 0,593 0,855 0,769 0,612 0,653 0,587 0,558
OQ10 0,639 0,9 0,847 0,73 0,735 0,66 0,577
OQ11 0,653 0,904 0,851 0,707 0,729 0,676 0,533
OQ12 0,628 0,734 0,793 0,921 0,706 0,619 0,568
OQ13 0,524 0,643 0,767 0,915 0,633 0,656 0,46
SeQ1 0,628 0,734 0,793 0,921 0,706 0,619 0,568
SeQ2 0,524 0,643 0,767 0,915 0,633 0,656 0,46
US1 0,657 0,741 0,758 0,66 0,877 0,652 0,586
US2 0,643 0,683 0,703 0,653 0,915 0,59 0,568
US3 0,658 0,722 0,741 0,663 0,933 0,632 0,56
US4 0,674 0,72 0,753 0,699 0,942 0,644 0,571
SQ1 0,559 0,659 0,797 0,604 0,646 0,845 0,544
SQ2 0,412 0,511 0,671 0,495 0,46 0,762 0,486
SQ3 0,482 0,638 0,771 0,577 0,571 0,821 0,464
SQ4 0,375 0,524 0,676 0,543 0,507 0,736 0,415
SQ5 0,449 0,49 0,659 0,486 0,478 0,76 0,434
SQ6 0,554 0,669 0,808 0,607 0,608 0,858 0,538
CU1 0,524 0,532 0,599 0,514 0,552 0,57 0,868
CU2 0,498 0,489 0,559 0,463 0,506 0,549 0,827
CU3 0,477 0,448 0,449 0,411 0,485 0,369 0,784
Reliability Testing. The reliability test is used to find out how consistent the measuring instrument is used (Abdillah & Hartono, 2015: 196). The rule of thumb used in testing reliability is to look at the composite reliability value and Cronbach's alpha value. A construct is said to meet the reliability test if the composite reliability value and Cronbach's alpha value are > 0.7, or equal to 0.6 because it is still acceptable ((Hair et al., 2006). Based on table 10, it can be seen that Cronbach's alpha and composite reliability values fulfill the rule of thumbs. Both of these parameters have values above 0.7, so that the measuring instrument used in this study is reliable.
Table 10 - Cronbach's Alpha and Composite Reliability Value
n/n Cronbach's Alpha rho A Composite Reliability
SE 0.898 0.899 0.936
IQ 0.897 0.915 0.926
OQ 0.938 0.944 0.947
SeQ 0.813 0.814 0.915
US 0.937 0.937 0.955
SQ 0.886 0.891 0.913
CU 0.768 0.773 0.866
The final conclusion from the overall testing of the outer model, namely the constructs and indicators are valid and reliable. The results of convergent validity, discriminant validity and reliability tests have fulfilled the rule of thumbs.
Structural Model Testing (Inner Model). Testing of the structural model (inner model) is carried out with the aim of knowing the degree of variation in the independent variable changes to the dependent variable. The parameter used is R2 (r-square). The higher the R2 value generated, the better the prediction model in a study. In addition, testing the inner model also aims to determine the level of significance in testing the hypothesis. The parameter used is the path coefficient value. The following are the results of testing the structural model (inner model) in this study.
Table 11 - R-Square
Construct R Square
KP 0.695
PK 0.388
In addition to testing the validity and reliability in evaluating the model, an evaluation of the inner model is also carried out by looking at the R-square value. The R-square value for user satisfaction is 0.695, meaning that 69.5 percent of the variation in user satisfaction is explained by the characteristics of the overall quality of the system (information quality, service quality, and system quality) and the self-efficacy of SAKTI users, the remaining 30.5 percent cannot be explained independent variables or can be explained by variables outside the research model. Furthermore, the R-square value for conative use is 0.388, meaning that 38.8 percent of the variation in conative use is explained by user satisfaction, the remaining 61.2 percent is explained by other variables not present in this research model.
Table 12 - Path Coefficient Value
n/n Original Sample (O) T Statistics (|O/STDEV|) P Values Decision
KK -> KP 0.596 10.126 0,000 Accepted
ED -> KP 0.300 5.055 0,000 Accepted
KP -> PK 0.623 13.300 0,000 Accepted
Based on table 13 above, it can be concluded that the results of hypothesis testing are as follows:
Hypothesis 1 states that the overall system quality construct has a positive effect on SAKTI user satisfaction. Based on the test results which can be seen in Table 12, the t-statistic value shows the number 10.126 or more than 1.64. In addition, the value of beta (P) also shows a positive number, namely 0.596. Based on these results, it can be concluded that the overall quality of the system has a positive effect on individual satisfaction in using SAKTI. Based on this, it can be said that hypothesis 1 is accepted.
The results of this study are in accordance with the first hypothesis of the study, namely the overall quality has a positive effect on SAKTI user satisfaction. Similar to the research conducted by Aldholay et al. (2018) who tested the overall quality of online learning on user satisfaction, namely students at nine state universities in Yemen. The results of his research show that overall quality has a positive and significant influence on online learning user satisfaction among students at nine state universities in Yemen. From the description above, it can be concluded that SAKTI users believe that the higher the quality of SAKTI as a whole, the more satisfying and comfortable it will grow when using the application. This can be interpreted, if you use SAKTI, the work will be completed effectively and efficiently.
Hypothesis 2 states that self-efficacy has a positive effect on SAKTI user satisfaction. Based on the test results that can be seen in Table 12, the t-statistic value shows the number 5.055 or more than 1.64. In addition, the value of beta (P) also shows a positive number, namely 0.300. Through these results, it can be concluded that self-efficacy has a positive effect on individual satisfaction in using SAKTI. Based on this, it can be said that hypothesis 2 is accepted. This implies that the more confident users are with their ability to search for information related to SAKTI, download and upload data on the SAKTI application and are
able to complete tasks, the more they understand that SAKTI can fulfill their expectations, its existence is important for system users.
The results of the study are in line with research conducted by (Hong et al., 2016) which states that there is a positive and significant relationship between self-efficacy and satisfaction. Students who feel confident learning musical instruments through social media will increase learning satisfaction through social media.
Hypothesis 3 states that the construct of user satisfaction has a positive effect on conative use of SAKTI. Based on the test results which can be seen in Table 12, the t-statistic value shows the number 13.300 or greater than 1.64. In addition, the value of beta (P) also shows a positive number, namely 0.623. Based on these results, it can be concluded that user satisfaction has a positive effect on conative use of SAKTI. Based on this, it can be said that hypothesis 3 is accepted. These results are in line with research conducted by (Kwahk et al., 2018) which states that user satisfaction has a positive influence on the conative use of mandatory information systems.
Construct Testing with Mediation Effects and its Discussion. At this stage, PLS-SEM analysis with mediating effects will be discussed, namely the relationship between exogenous and endogenous constructs through connecting variables or intermediate variables. In other words, the influence of exogenous variables on endogenous variables can be direct but also through intermediate or mediating variables. There are 3 conditions in testing mediation effects (Baron and Kenny, 1986). First, examining the effect of exogenous variables on endogenous variables must be significant at a t-statistic > 1.64 and form a c path. Second, testing the effect of exogenous variables on the mediating variable must be significant at a t-statistic value > 1.64 forming a path. Third, testing simultaneously the effect of exogenous and mediating variables on endogenous variables. The effect of the mediating variable on the endogenous variable must be significant and form path b (Baron and Kenny, 1986).
The stages of mediation testing can be seen in Figure 4 below:
With mediator
Figure 4 - Mediation Testing Stage
After these three conditions are met, conclusions are drawn regarding mediation. The following are three conclusions when the three conditions are met:
• If path c' is significant but the beta coefficient is the same as path c, then the mediation hypothesis is not supported;
• If path c" and path c remain significant, but the value of the beta coefficient in path c" < beta coefficient of path c, there is a partial mediating relationship;
• If path c" is not significant even though path c is significant and the value of the beta coefficient in path c" < beta coefficient of path c, there is a full mediating relationship.
The first step in testing the mediating effect is to look at the effect of the exogenous construct on the endogenous variables and it must be significant at a t-statistic > 1.64.
Table 14 shows if the overall quality of the system has a positive and significant effect on the conative use of SAKTI with a coefficient value of 0.458 and a t-statistic value of 7.477 or greater than 1.64. This shows that the first condition for carrying out the mediation test on
the fourth hypothesis is met. The fourth hypothesis testing can then be carried out because the first condition for mediation testing is fulfilled.
Table 13 - Mediation Hypothesis Path Coefficient Value
n/n Original Sample (O) T Statistics (|O/STDEV|) P Value
KK -> PK 0,458 7,447 0.458
ED -> PK 0,283 4,512 0.283
KK -> KP 0,602 10,678 0.000
ED -> KP 0,293 5,234 0.000
KP -> PK 0,172 1,829 0.068
Subsequent tests for the first condition show that self-efficacy has a positive and significant effect on conative use of SAKTI with a coefficient value of 0.283 and a t-statistic value of 4.512 or more than 1.64. This shows that the first condition for carrying out the mediation test on the fifth hypothesis is fulfilled. The fifth hypothesis testing can then be carried out because the first condition for mediation testing is fulfilled.
The second step of testing the mediation hypothesis is carried out by testing the exogenous construct and the mediator construct. Table 14 shows that the constructs of overall system quality and self-efficacy on user satisfaction constructs have respective coefficient values of 0.602 and 0.293 and respective t-statistic values of 10.678 and 5.234 or greater than 1.64. The results of this study indicate that there is a significant influence between the exogenous construct and the mediator construct, which means that the second condition for carrying out the mediation test has been fulfilled.
The third stage is to test the mediator construct with endogenous constructs. Table 14 shows that the construct of user satisfaction with the SAKTI conative use construct has a coefficient value of 0.172 and a t-statistic value of 1.829 or greater than 1.64. The third condition is fulfilled if the mediator construct significantly influences the endogenous construct and the fourth stage of testing can be carried out.
After the three test conditions have been met, the next step is to conclude mediation.
Table 14 - Mediation Conclusions
n/n Original Sample (O) T Statistics (|O/STDEV|) Mediation Conclusions
C c' c c'
KK -> PK 0,458 0,296 7,447 5,164 Partial Mediation
ED -> PK 0,283 0,235 4,512 3,266 Partial Mediation
To find out how far the variable of user satisfaction can mediate the relationship between the quality of the whole system and self-efficacy on conative use of SAKTI can be seen in the specific indirect effects table.
Table 15 - Specific Indirect Effects
n/n Original Sample Sample Mean Standard Deviation T Statistics P Values
OQ>US>CU 0.103 0.097 0.057 1.820 0.069
SE>US>CU 0.051 0.049 0.030 1.673 0.095
From the results of the PLS analysis above, it shows that the overall quality has a significant influence (p = 0.103) on the conative use of SAKTI with a t-statistic value of 1.820 or greater than 1.64. The results of the study show that user satisfaction will partially mediate the relationship between the overall quality of the system and conative use of SAKTI. Based on this, it can be said that the fourth hypothesis is accepted.
The fifth hypothesis states that user satisfaction mediates the relationship between self-efficacy and conative use of SAKTI. The results of the analysis show that self-efficacy has a positive and significant effect on conative use of SAKTI with a coefficient value of 0.051 and a t-statistic value of 1,673 or greater than 1.64. The results showed that user satisfaction partially mediates the relationship between self-efficacy and conative use of SAKTI. Based on this, it can be said that the fifth hypothesis is accepted.
CONCLUSION
The success of the implementation of mandatory SAKTI can be seen from the behavior of users who really want to involve and use the system or what is commonly referred to as conative use. The results of this study indicate that conative use is influenced by system user satisfaction, so that users will proactively provide positive feedback and increase self-capability both for using the system and knowledge about APBN financial management. To increase the conative use of SAKTI, this can be done by increasing the satisfaction and comfort of system users.
The results of this study indicate that the overall quality of the system consisting of system quality, information quality and service quality as well as self-efficacy has a positive effect on user satisfaction. First, to increase the satisfaction and comfort of system users, do this by improving the quality of the system, including increasing ease of use, increasing the speed of access to servers and increasing system security. Second, to increase the satisfaction and comfort of system users, it can be used by improving the quality of information, namely providing information that is complete, relevant, accurate, up to date and has a format that is clear and easy to understand. The third way to increase the satisfaction and comfort of system users is by improving service quality, namely by preparing support staff who have good skills, speed and responsiveness in dealing with system problems.
Furthermore, the results of this study indicate that self-efficacy influences user satisfaction. Confidence in the ability of a user to use an information system will give positive feelings regarding the use of SAKTI and feel the benefits of the system in completing work.
Overall, the results of this study prove that SAKTI users are willing to proactively involve themselves in using SAKTI even in a mandatory environment if the system is of good quality overall and users have high self-confidence in their ability to operate the system so that users will have a sense of comfortable and satisfied with the use of the system.
REFERENCES
1. Abdillah, W., & Hartono, J. (2015). Partial Last Square (PLS) - Alternatif Structural Equation Modelling (SEM) dalam Penelitian Bisnis (D. Prabowo, Ed.). CV. Andi Offset.
2. Afwande, M., Mbugua, S., & Kilwake, J. (2021). Framework for Evaluation of Effectiveness of Integrated Financial Management Information Systems Architecture Implemented in Public Hospitals in Kenya. International Journal of Scientific Research in Science, Engineering and Technology, 38-41. https://doi.org/10.32628/ijsrset207473.
3. Aldholay, A., Isaac, O., Abdullah, Z., Abdulsalam, R., & Al-Shibami, A. H. (2018). An extension of Delone and McLean IS success model with self-efficacy: Online learning usage in Yemen. The International Journal of Information and Learning Technology, 35(4), 285-304.
4. Al-Mamary, Y. H., Shamsuddin, A., & Aziati, N. (2014). The Relationship between System Quality, Information Quality, and Organizational Performance The Integrated Framework of Cloud Computing Implementation in Higher Education Institution: A Review of Literature View project Development of integrated model for Successful implementation of MIS View project The Relationship between System Quality, Information Quality, and Organizational Performance. International Journal of Knowledge and Research in Management & E-Commerce, 4. https://www.researchgate.net/publication/269463861.
5. Amriani, T. N., & Iskandar, A. (2019). Analisis Kesuksesan Implementasi Sistem Aplikasi Keuangan Tingkat Instansi (SAKTI) pada Satuan Kerja di Lingkungan Badan Pendidikan and Pelatihan Keuangan (BPPK). Kajian Ekonomi and Keuangan, 3(1), 54-74.
6. Djuitaningsih, T., & Arifiyantoro, D. (2020). Individual And Organizational Impacts: Information And System Quality Influence On Attitude Towards Use And User Satisfaction Of Agency-Level Financial Application System. Acta Informatica Malaysia, 4(1), 10-18. https://doi.org/10.26480/aim.01.2020.10.18.
7. Hong, J. C., Hwang, M. Y., Szeto, E., Tsai, C. R., Kuo, Y. C., & Hsu, W. Y. (2016). Internet cognitive failure relevant to self-efficacy, learning interest, and satisfaction with social media learning. Computers in Human Behavior, 55, 214-222. https://doi.org/10.1016/j.chb.2015.09.010.
8. Kwahk, K. Y., Ahn, H., & Ryu, Y. U. (2018). Understanding mandatory IS use behavior: How outcome expectations affect conative IS use. International Journal of Information Management, 38(1), 64-76. https://doi.org/10.1016/j.ijinfomgt.2017.07.001.
9. Prabowo, N. T. (2017). Analisis Sistem Aplikasi Keuangan Tingkat Instansi (Sakti) Dengan Pendekatan Technology Acceptance Model (TAM). http://www.wikiapbn.org/
10. Seddon, P. B., & Kiew, M.-Y. (1996). A Partial Test And Development Of Delone And Mclean's Model Of Is Success 3.
11. Sujarweni, V. W. (2022). Metodologi Penelitian Bisnis and Ekonomi. Yogyakarta: Pustakabarupress.
12. Torres, L., Pina, V., & Acerete, B. (2005). E-government developments on delivering public services among EU cities. Government Information Quarterly, 22(2), 217-238. https://doi.org/10.1016/j.giq.2005.02.004.
13. Wang, Yi Shun., & Liao, Yi Wen. (2008). Assessing eGovernment systems success: A validation of the DeLone and McLean model of information systems success. Government Information Quarterly, 25(4), 717-733. https://doi.org/10.1016/j.giq.2007.06.002.
