Partnership Quality, Student’s Satisfaction, and Loyalty: A Study at Higher Education Legal Entities in Indonesia

2.1 Research Design

This research adopted a quantitative survey with a quantitative descriptive and regressive approach. The independent variable was PQ, while the dependent variables were SL and SS. This type of study tests hypotheses to explain the effect between variables. Data were collected on each student individually; hence, the unit of analysis is individual. Data were collected from students in the university with a legal entity, and one response was taken from each lecturer individually. The data were collected once and represent a portrait from a single point in time (Dami, Imron, Burhanuddin, & Supriyanto, 2022; Dami, Imron, Burhanuddin, Supriyanto, & Pellokila, 2023; Dami, Imron, Burhanuddin, & Supriyanto, 2024). Therefore, this study was represented as a cross-sectional study with a survey type.

2.2 Sample and Procedures

The study involved a population of students across 16 public universities with legal entities. The sampling was performed using purposive sampling, focusing on students with firsthand experience in using the services provided by the university in partnership with their partners. In total, this study involved 875 students from the university with a legal entity. The criteria for the number of research samples based on the structural equation modeling (SEM) analysis (Albright & Park, 2009) were determined by 5–10 times the number of manifest variables/indicators of the overall latent variables. This study used 3 latent variables with 14 manifest variables/indicators; thus, the minimum number of samples used in this study according to the criteria for the number of samples for SEM analysis was 14 × 10 = 140. Based on the formula, the study is feasible because it used 875 samples.

Data were garnered through the distribution of questionnaires facilitated, but the students actively participated in the college organization. Data collection was performed by distributing questionnaires among students with the approval of the universities through permission from the Dean. Specifically, the data were collected by distributing the Google Form link. In this study, when the questionnaires were distributed, 875 questionnaires were collected. So, the 875 respondents who participated answered all the questions.

2.3 Instrument

The PQ variable was measured through five factors consisting of trust, business understanding, benefit or risk share, conflict, and commitment (Aldridge & Rowley, 1998; Chen & Chen, 2014; Ee et al., 2013; Espino-Rodríguez & Ramírez-Fierro, 2018; Figueiredo et al., 2020; Heckman, King, Heckman, & King Katz, 1994; Khoo, Ha, & McGregor, 2017; Lahiri & Kedia, 2009; Lee & Kim, 1998; Park et al., 2017; Teeroovengadum et al., 2019; Yong-Sik & Yung Kyun, 2019). The SL variable was measured through three factors consisting of repeat purchasing or usage service, positive word of mouth, and demonstrating resistance to switch to the competitor (El-Kassar, Makki, & Gonzalez-Perez, 2019; Lai, Pham, Nguyen, Nguyen, & Le, 2019; Lie et al., 2019; McMullan & Gilmore, 2003; Otsetova et al., 2017; Waheed et al., 2021; Zaid & Patwayati, 2021). The SS variable was investigated using the factors of curriculum and learning, reliability, responsiveness, assurance, tangibility, and empathy (Abbas, 2020; Hew, Hu, Qiao, & Tang, 2020; Paposa & Paposa, 2023; Pedro, Leitão, & Alves, 2016). All survey items were evaluated using a 5-point Likert scale, ranging from (1) strongly disagree to (5) strongly agree.

2.4 Data Analysis

The garnered data were analyzed using a descriptive analysis through IBM SPSS Statistics 22. A statistical analysis was also performed using Structural Equation Modeling-Partial Least Squares (SEM-PLS) with SmartPLS 3 application. SmartPLS is a data processing application based on non-parametric statistics, thus eliminating the necessity for data with a normal distribution (Garson, 2016). Although using SmartPLS does not obligate the normal data distribution, it still requires a structural model test through examination of standardized root mean square residual, NonFit Index (NFI), R ², and Q ² (Garson, 2016; Hair et al., 2021; Hair Jr., Hult, Ringle, & Sarstedt, 2017a).

3.1 Demographic Profile

This study involved 875 students from 16 universities with legal entities as the participants. Their profile recorded in this research included their name, gender, duration of study, highest education level, and origin from a university with a legal entity. Their demographic data are presented in Table 1.

Among the 875 participants, 556 (64%) were female, while the remaining 319 (36%) were male. These participants have varied study periods, with 11.7, 0.34, 32.57, 0.23, 31.54, 0.69, 15.89, and 6.97% of them are in the second, third, fourth, fifth, sixth, seventh, and higher than eighth semesters. The data indicated that 87% of respondents were high school graduates pursuing their undergraduate degrees. The remaining of the participants held diploma 3 (1%), diploma 4 (2%), bachelor’s (9%), and master’s (1%) degrees. The last educational background of the participants was that of high school graduates.

3.2 Descriptive Analysis

The descriptive analysis was performed to provide fundamental information concerning the levels of PQ, SS, and SL from 16 state universities with legal entities in Indonesia. The data were garnered using a survey involving 875 respondents. The garnered descriptive data, including interval measures, frequencies, and percentage levels of PQ, SS, and SL, are presented in Table 2.

As shown in Table 2, the mean PQ value is 108.75, which is classified as a very high category. Particularly, these scores are distributed into 67.77%, classified as very high, while 16.19% as high and 14.28% as moderate. Meanwhile, the average SS score is 184.43, which is classified as a high category. In detail, 45.49% of the scores are categorized as high, while 36.34 and 16.34% are categorized as very high and moderate. The mean SL score is 48.19, classified in the high category, with 43.66% of the scores in the high category, while 35.77 and 18.06% in the very high and moderate category.

4.1 Measurement Model

The validity test on the variables’ indicators was performed using SmartPLS 3. This test was carried out to determine if the indicators measure the variable accurately and effectively. The obtained outer loading for each indicator is presented in Table 4 and Figure 2.

Figure 2

PLS-path model.

The obtained loading factor is presented in Table 3. This value was further used to identify the convergent validity. The data indicate that indicators PQ 4 (conflict) and SL1 (repeat purchasing or using services) were observed to be invalid, incapable of predicting, and cannot serve as excellent predictors. The invalid factors were gradually removed, particularly those with negative values (SL1), while those with greater than 0.4 were retained (Hair Jr., Matthews, Matthews, & Sarstedt, 2017b). As this study aims to identify facts based on research data, the validity and reliability of a factor are of concern (Esposito Vinzi, Chin, Henseler, & Wang, 2010). The obtained Cronbach’s alpha for the PQ variable is 0.898 > 0.6 (reliable). Further, the reliability of SS is 0.976 > 0.6 (reliable) and SL variable 0.444 < 0.6 (non-reliable). The obtained composite reliability is presented in Table 3, suggesting the composite reliability for PQ of 0.981 > 0.6 (reliable), while the reliability for SS of 0.929 < 0.6 (reliable) and for SL of 0.728 > 0.6 (reliable). Ideally, a variable is classified as valid if it obtains AVE > 0.5. As shown in Table 3, the AVE value for the PQ variable is 0.734 (>0.5), implying its validity. Conversely, the AVE values for the SL and SS variables are 0.642 (>0.5) and 0.893 (>0.5), indicating validity.

A discriminant validity test was performed through SmartPLS using cross-loading results, following particular criteria. The criteria specifically obligate each indicator to have a higher correlation value than other indicators in the same variable. Further, Table 5 presents that each indicator shows a higher correlation value than the other indicators in the same variable, after SL1 was removed because of that, SL1 had an outer loading <0.5 (−0.454) and a Cronbach alpha <0.7 (0.444). Therefore, the indicators are classified as valid.

4.2 Structural Model

Prior to the hypothesis test, the feasibility of the structural model was assessed through standardized root mean residual (SRMR) and NFI values garnered from the bootstrapping procedure in SmartPLS 3. A model is categorized as fit or satisfactory if its SRMR value is less than 0.100. Meanwhile, the investigation using NFI value was conducted using specific distinct criteria. The NFI value exceeding 0.80 indicates a good fit, while if the NFI is between 0.60 and 0.80, the model is categorized as a moderate fit. If the NFI is lower than 0.60, the model is classified as not fit (Hair Jr. et al., 2017b). The results of SRMR and NFI observation using SmartPLS 3 are shown in Table 6.

In addition, the obtained SRMR of 0.053 < 0.100 implies that the model meets the required criteria. The model attained a 0.924 score for the NFI, indicating its good fit criteria. Aside from the SRMR and NFI, the structural model was also assessed through the R ² and Q ² values.

The R ² coefficient serves as the instrument facilitating the investigation of the ability of the exogenous construct to describe the endogenous construct. Ideally, R ² falls between 0 and 1 (Sarstedt, Ringle, & Hair, 2017). It indicates the degree to which another can explain one variable’s variance. For the R ² coefficient evaluation, a strong model presents R ² > 0.75, while the moderately robust model has 0.50 < R ² < 0.75, while weak and very weak models have 0.25 < R ² < 0.5 and < 0.5 (Sarstedt et al., 2017). Further, the adjusted R ² represents the corrected R ² following the standard error. The adjusted R ² offers a more reliable illustration of an exogenous construct’s ability to explain the endogenous construct compared to the R ² (Table 7).

Predictive relevance (Q ²) was garnered through the blindfolding method (Table 7). This blindfolding method aims to investigate the predictive relationship between the constructs of the dependent variable in the model and the constructs of the independent variable (Henseler & Sarstedt, 2013). If ∑Q ² > 0 was obtained, the specific dependent constructs in the model are assessed with predictive relevance regarding the independent constructs. A model with predictive relevance has Q ² > 0, while Q ² < 0 indicates that a model has no predictive relevance. A value approaching 1 indicates a better model (Henseler & Sarstedt, 2013).

The obtained coefficient determination R ² presented in Table 7 suggests that PQ influences SS with an R ² of 0.824 and an adjusted R ² of 0.824. Therefore, the exogenous construct PQ influences SS by 0.824 or 82.4%. Meanwhile, the PQ and SS were observed affecting CL with an r ² value of 0.173 and an adjusted R ² of 0.172 or 17.2%. Accordingly, the exogenous construct PQ presents high effects on CL (>75%), while the effects of PQ and SS constructs on CL are exceedingly weak (25%).

Table 7 also presents Q ² of the PQ variable toward the SS variable, showing a score of 0.732 > 0. Therefore, PQ possesses significantly strong predicting relevance to SS as its score approaches 1. Meanwhile, the Q ² score of PQ and SS toward the SL variable was observed at 0.106 > 0, indicating weaker predictive relevance than the PQ toward SS.

In general, the obtained data signify a weak structural model depicting the influence of PQ and SS toward CL. Further, the direct influence of PQ on CL was observed as non-existent and non-significant. Meanwhile, its indirect influence is mediated by SS.

5.1 Direct Effect

Following the results of the analysis conducted through SmartPLS 3, the model of intervariable relationships shows a direct influence of PQ on SS, SS on CL, and PQ on SL. Therefore, with a 5% or 0.05 error tolerance level, the variable was concluded to have effects if t _count > 1.96, and its effect was deemed significant if P values <0.5. In contrast, if t _count < 1.96, then H ₀ was accepted. The obtained P values and t _count from the SmartPLS 3 calculation are presented in Table 8.

5.2 Partnership Quality on Student’s Satisfaction

The results presented in Table 7 suggest that the influence of PQ on SS has P-values of 0.000 and t _count of 128,866. As the obtained P-values < 0.5 and t _count > 1.96, signifying that H ₀ was rejected and there were direct and significant influences from PQ on SS.

5.3 Student’s Satisfaction on Student’s Loyalty

The analysis results using SmartPLS 3 indicate the effects of SS on SL with P-values of 0.000 and t _count of 3.704. Accordingly, as the P-values <0.5 and t _count > 1.96, H ₀ was rejected, and the SS variable presents significant direct effects on SL.

5.4 Partnership Quality on Student’s Loyalty

The analysis of the PQ variable’s effects on SL suggests P-values of 0.100 and t _count of 1.646. Thus, H ₀ was accepted and PQ variable was proven to have significant effects on SL, as the obtained P-values > 0.5 and t _count < 1.96.

5.5 Indirect Effects

For the indirect influence model focusing on the effects of PQ on SL, the analysis was conducted using SmartPLS 3. A variable is concluded to have an indirect influence if it presents a t-value >1.96, while its influence is deemed significant if P-values <0.5. The obtained t-value and P-values for the indirect influence are presented in Table 9.

From the calculation results, the t _count for PQ’s effects on SL mediated by SS is 3.702 with P-values of 0.000. Therefore, H ₀ was rejected, and there is an indirect significant influence of PQ on SL through SS.

6.1 Effects of Partnership Quality on Student’s Satisfaction

The research analysis suggests a significant influence of PQ on SS, as they show P-values of 0.000 < 0.5 and t _count of 128.866 > 1.96. This significant effect implies that a high-quality partnership of the university with legal entities indicates high SS and vice versa. This relationship is evident from the results of the analysis, suggesting that with excellently high PQ, SS falls within the “high” criteria.

The high-quality partnership is attributed to several factors, including trust, business understanding, benefit or risk sharing, and commitment. These factors present outer loading >0.7, implying their robust capacity as the predictor of PQ. As reported by Ee et al. (2013), Espino-Rodríguez and Ramírez-Fierro (2018), Lahiri and Kedia (2009), Lee and Kim (1998), and Park et al. (2017), trust, business understanding, benefit or risk share, and commitment are the excellent predictor of PQ. However, the factor of conflict is identified as having 0.478, suggesting its possibly weaker predicting capacity in the PQ domain. Ideally, an excellent partnership is characterized by minimum conflicts as it streamlines the partnership activities (Osei-Kyei, Chan, Yu, Chen, & Dansoh, 2019). However, in a long-term relationship, a conflict is nearly inevitable because of distinct viewpoints from numerous stakeholders (Osei-Kyei et al., 2019). This feeble ability to predict conflict may also be attributed to the low conflicts between the university, legal entities, and their partners.

Similar to our findings, the study conducted by Thung (2019) on the Go-Jek partners and Istiqomawati (2017) on the customers of Payung Argodana Pudak Bank, Semarang, also reported that excellent partnership induces greater customer satisfaction.

6.2 Effects of Student’s Satisfaction on Student’s Loyalty

SS has also recorded significant influence on SL, as indicated by the garnered P-values of 0.000 < 0.5 and t _count of 3.704 > 1.96. The significant and positive correlation between SS and SL signifies that greater SS corresponds to higher levels of SL and vice versa. As recorded in this study, high SS corresponds to high loyalty. Additionally, high SS may also be attributed to the curriculum and learning, reliability, responsiveness, assurance, tangibility, and empathy (Aldridge & Rowley, 1998; Chen & Chen, 2014; Figueiredo et al., 2020; Heckman et al., 1994; Khoo et al., 2017; Teeroovengadum et al., 2019; Yong-Sik & Yung Kyun, 2019). All these factors have been observed to present an outer loading value of >0.7. Linearly, Azizan and Yusr (2019), Brunner et al. (2008), Dam and Dam (2021), Gong and Yi (2018), Idris (2022), Latif et al. (2021), Olaleye et al. (2021), Waheed et al. (2021), and Yang and Peterson (2004) reported positive and significant effects from customer satisfaction on their loyalty.

6.3 Effects of Partnership Quality on Student’s Loyalty

From the analysis results, PQ variable influence on SL presents P-values of 0.100 and t _count of 1.646; thereby, PQ has no direct influence on SL. In this study, students’ loyalty was measured through three factors, namely repeat purchasing or service usage, positive word of mouth, and demonstrating resistance to switch to the competitor (El-Kassar et al., 2019; Lai et al., 2019; Lie et al., 2019; McMullan & Gilmore, 2003; Otsetova et al., 2017; Waheed et al., 2021; Zaid & Patwayati, 2021). Among those factors, the positive word of mouth and resistance to switching to the competitor are classified as great predictors with outer loading of 0.669 and 0.780. In contrast, the repeat purchasing or using service predictor attained an outer loading of −0.454.

The absence of impacts from PQ on SL can be attributed to numerous factors, such as students’ low interest in pursuing higher education, leading to a diminished repeat service usage factor in the loyalty construct. Further, this low inclination to pursue higher education has been widely reported across Indonesia, proving the rarety of higher education compared to secondary education. In 2017, approximately 16% of Indonesian youth have completed higher education, far below the average of G20 countries (38%) (OECD, 2019). Further, in 2022, 42% of young adults leave school without a high school qualification, 14% higher than the OECD country average (OECD, 2022). These data significantly impacts students’ loyalty, specifically the service usage predictor, since the limited pursuit of higher studies results in a lack of continuation for students to use university services repeatedly.

People’s low interest in pursuing higher education is also affected by the unexamined variables, such as the role of parents (Hariyadi, Cahyadi, & Slamet, 2022) and economic issues (Sudradjat, Muktiadji, Gendalasari, & Hendrian, 2015; Zaini, Mansor, Yusof, & Sarkawi, 2019). The issues of the economy also cause low repeat purchasing factors in the PQ dimension. Although the PQ presents no direct effects on students’ loyalty, it has indirect effects mediated by the students’ satisfaction.

6.4 Effects of Partnership Quality on Student’s Loyalty through Student’s Satisfaction

Mediated by SS, PQ influences students’ loyalty, as evidenced by t _count of 3.702 and P-values of 0.000. The obtained data suggest significant indirect effects of PQ on SL through the SS. Student or user satisfaction has been commonly reported as the mediating variable in various research models (Istiqomawati, 2017; Thung, 2019). Linearly, Istiqomawati (2017) and Thung (2019) also reported that excellent PQ generates user satisfaction. Meanwhile, Prasantyo and Kartika (2023), Sari, Widayanti, and Soedarto (2021), and Trieanto, Ismiasih, and Manumono (2022) uncovered positive effects of user satisfaction on their loyalty. Customer satisfaction is one of the intervening variables influencing the partnership’s impact on customer loyalty (Istiqomawati, 2017).

In educational services, cultivating SS and loyalty is essential since it translates into financial support for the university. In the long-term strategy, SL can materialize competitive advantage. Essentially, loyalty is the ongoing support for a product over a period of time (Twum, Yalley, Agyapong, & Ofori, 2021; Yang & Peterson, 2004). The alumnae can continuously support universities through great loyalty through financial contributions and work opportunities for fresh graduates (Rojas-Méndez, Vasquez-Parraga, Kara, & Cerda-Urrutia, 2009). Thus, alumni loyalty is a significant value for higher education institutions as it can be presented in time and financial resources (Snijders et al., 2019).

6.5 Strength of Study

This study is relevant amidst the transition trend of Indonesian state higher education institutions transitioning into universities with legal entities. This transformation aims to streamline the operational aspects of teaching, research, and community engagement in the organization for greater effectiveness and efficiency (Swasti, Sonhadji, Bafadal, & Supriyanto, 2018). This study also aligns with the research focusing on quality improvement and innovation of higher education during the Industrial Revolution 4.0 (Supriyanto, Djum, & Benty, 2019). Quality is inherently subject to change and the evolution of times (Supriyanto et al., 2019). Accordingly, in the 4.0 era, universities must continuously enhance their quality.

In addition, sustainable development has been declared as a global program (Leal Filho et al., 2022). Particularly, in the recent digital economy era, higher education plays a pivotal role in economic advancement (Kholiavko, Popelo, Melnychenko, Derhaliuk, & Grynevych, 2022). However, the higher education institutions in Indonesia exhibit relatively minimum contribution to the nation’s progression due to their insufficient quality. This low quality is induced by numerous factors, such as governance mistakes in the institutional stage (Rosser, 2023).

In facing these quality and management issues in higher education institutions, partnerships with other foundations appear as the alternative (Leal Filho et al., 2022). The partnership has enhanced sustainable development, particularly in critical resilience and sustainable finance (Akomea-Frimpong et al., 2023). Partnership is established based on the needs of an organization. For instance, a school can partner with a training center to improve students’ competencies (Yuanita, Supriyanto, & Mustiningsih, 2020).

Following the results of the analysis, several conclusions are made. First, PQ directly impacts the level of SS. Second, SS has a direct effect on SL. Third, the quality of the partnership is observed to have a direct influence on student quality. Fourth, the study revealed an indirect influence of PQ on SL, mediated by SS. In this study, students’ loyalty is established through their satisfaction, while the satisfaction is constructed through partnership. Students’ great loyalty and satisfaction characterize a successful higher education institution (García-Rodríguez & Gutiérrez-Taño, 2021; Latif et al., 2021; Rojas-Méndez et al., 2009). For this loyalty and satisfaction, the university must form a partnership with a legal entity. The analysis’s finding emphasizes that partnerships are influential in the sustainable development of universities with legal entities. Borodiyenko, Zlenko, Malykhina, Kim, and Diachkova (2022) described the need to support educational institutions’ strategic development, particularly through bureaucracy, business orientation, communication, and partnerships.

The research uniqueness lies in detecting repeat purchasing or service usage factors that fail to predict students’ loyalty variables, as evidenced by the obtained outer loading value of −0.454 < 07. This finding implies that conflict cannot predict excellent PQ. The failure of the conflict factor in predicting the PQ is attributed to the absence of conflict between the university and the legal entity and its partnership. Ideally, Excellent partnership materializes into a harmonious and secure environment, enhancing a company’s productivity (Herman et al., 2023). Further, the data analysis also suggests a high category of PQ. Interestingly, we also observed that the repeat purchasing or service usage factor on SL has a low predicting ability with an outer loading of −0.454. The failure of the repeat purchasing or repeat services usage factor is possibly caused by students’ low interest in reusing services or their lack of interest in pursuing further education (OECD, 2019, 2022), coupled with issues related to insufficient parental support and economic factors (Sudradjat et al., 2015; Sudradjat, Muktiadji, Gendalasari, & Hendrian, 2018).

7.1 Implication, Limitations, and Future Research

Theoretically, this research confirms that the factors of PQ, including trust, business understanding, benefit or risk share, and commitment, serve as excellent predictors with an outer loading >0.7, as reported by Ee et al. (2013), Espino-Rodríguez and Ramírez-Fierro (2018), Lahiri and Kedia, (2009), Lee and Kim (1998), and Park et al. (2017). However, the conflict factor presents low predicting power with an outer loading value of 0.478 < 0.5.

In addition, the student’s satisfaction factor consists of curriculum and learning, reliability, responsiveness, assurance, tangibility, and empathy. Each of these factors serves as a good predictor with an outer loading value >0.7, similar to the report from Aldridge and Rowley (1998), Chen and Chen (2014), Figueiredo et al. (2020), Heckman et al. (1994), Khoo et al. (2017), Teeroovengadum et al. (2019), and Yong-Sik and Yung Kyun (2019). Furthermore, the factors of SL variable, specifically positive word of mouth and demonstrating resistance to switching to a competitor, emerge as reasonably good predictors, as previously reported by El-Kassar et al. (2019), Lai et al. (2019), Lie et al. (2019), McMullan and Gilmore (2003), Otsetova et al. (2017), Waheed et al. (2021), and Zaid and Patwayati (2021). However, the repeat purchasing or service usage factor has an outer loading value of −0.454, indicating its incapacity to predict the students’ loyalty. Further, the regression analysis results suggested (1) a direct influence of the PQ variable on SS; (2) a direct influence of the student’s satisfaction variable on SL, evidenced by P values of 0.000 < 0.5 and a t _count of 3.704 > 1.96; (3) there is no direct influence of the PQ variable on student’s quality, with P values of 0.100 > 0.5 and a t _count of 1.646 < 1.96; and (4) there is an indirect influence of the PQ variable on student’s loyalty through student’s satisfaction, indicated by a t _count of 3.702 > 1.96 and P values of 0.000 < 0.5.

Practically, the research results of this research imply the need for Indonesian universities with legal entities to support sustainable development in Indonesia by focusing on the quality of sustainable partnerships (Akomea-Frimpong et al., 2023). With the complex issues encountered by higher education institutions with a legal entity, collaboration with partners is essential (MacDonald et al., 2022). This partnership is particularly essential to fulfill their needs and resolve the shortcomings of the university with a legal entity, such as in the aspects of human resources, finances, facilities, and services. The increasing quality of partnership results in greater SS and loyalty (Dwijayanti, Ridwan, & Rohani, 2021; Hestiningtyas et al., 2022; Pastika, Suparta, & Dewi, 2016; Sari et al., 2021; Trieanto et al., 2022). Alumni loyalty provides essential support to the university, specifically through verbal endorsements, along with financial contributions in the form of job opportunities for fresh graduates (Hassan & Shamsudin, 2019) and solidarity among alums. Their positive word of mouth, financial contribution, and job offerings for the fresh graduates are crucial for the university (Rojas-Méndez et al., 2009). Therefore, their loyalty is essential for the higher education institution (Snijders et al., 2019).

This research is relevant to studying quality improvement and innovation in higher education during the 4.0 era (Supriyanto et al., 2019), leading to transformed expected quality (Supriyanto et al., 2019). In this 4.0 era, the university has to continuously enhance its quality, including the quality of its partnership. The results of this study corroborate previous research suggesting the influence of PQ on satisfaction (Dwijayanti et al., 2021; Sari et al., 2021; Trieanto et al., 2022), satisfaction on loyalty (Hestiningtyas et al., 2022; Pastika et al., 2016), as well as PQ to loyalty mediated by satisfaction (Hestiningtyas et al., 2022; Pastika et al., 2016).

This research was only conducted in 16 public universities in Indonesia. Meanwhile, there are still many other private universities that have not been researched. Therefore, future research can make a comparison between public universities and private universities. This research only focuses on PQ and its influence on SS and loyalty. Therefore, further research is still open to examining aspects of policy, management, leadership, academic service quality, total quality management, productivity, decision-making, and quality culture to improve the high quality of future organizations. It is hoped that further research can provide the best service quality model for students or customers in higher education.