Service Quality of and User Satisfaction with Non-State-Owned Academic Libraries in China: Integrating the Fuzzy Delphi Method with the Kano Approach

Abstract

Libraries are digitizing, and challenges are posed by digital technologies for institutions of higher education in China. This study aims to present the dimensionality of perceived service quality, its effect on customer satisfaction, and the case of a non-state-owned library’s academic service quality. A sample consisting of valid 453 respondents used online recruitment to retrieve answers to questionnaires. Ten experts were invited to review the questionnaire for content validity and question clarity. In this study, the Fuzzy Delphi method was used to establish questionnaire indices and the attributes of library academic service quality elements made available by the Kano model. Three dimensions, including emotional service, physical environment, and information control, which are correlated under the attribute classification of the Kano model, indicate support for the validity of using integrated models in measuring library service quality. The results, according to the improvements in the customer satisfaction matrix, provide nine elements to improve the quality of service and two major improvements to enhance the perception of service quality. In addition, users pay less attention to the use of academic resources and academic ethics, but these factors do not affect the quality of service.

1. Introduction

Digital technology is widely used in academic libraries. It is worth mentioning that the rapid development of advanced technologies, which caused new pressures and barriers in service quality assessment, led to the challenges of sustainability, equality, and access and has played an important role in libraries. Recently, service quality has become an important issue for higher education due to digital technology regulation, online restrictions, and internet control in China. Assessing service quality is essential to users in the current competitive educational environment. In other words, the assessment of a library’s success depends on the users’ judgment of quality [1,2].

The library has a positive impact on the teaching, learning, and research of academic staff, and satisfactory library services can be quite helpful in instilling confidence in library users [3]. An important indicator of the quality assessment of higher education is the assessment of the quality of service. The quality of services provided to library users (primarily students and researchers) is also a key factor affecting library performance. Hence, the development quality of university libraries is closely related to the sustainable and stable development of all higher education. Fu [4] proposed that “the library is the heart of a university” and highlighted university libraries’ important position on campuses. University libraries have an important position, mainly because they can support teaching, research, and extension services and promote the overall research and development capacities of universities. Providing good services for university teachers and students is the starting point and ultimate goal of university libraries. Therefore, how to continuously provide correct and efficient services to all teachers and students is always the direction that university libraries pursue and strive to achieve [5].

The key research question that this study aims to answer is how non-state-owned university libraries can provide better services, adjust existing services, and enhance users’ understanding of these services. However, library service quality assessments in the past stressed the relationships among resources, capabilities, utilizations, and effects rather than paying attention to users [6]. Parasuraman et al. [7] indicated that the SERVQUAL scale is highly effective when used for service quality assessments of the commercial field. Hence, some scholars are starting to use it for service quality assessments of libraries. Hebert [8] used it to investigate the service quality of interlibrary loans between large public libraries. Wisniewski [9] assessed the service quality of public libraries in the UK. Moreover, overlapping issues may appear in some aspects of quality [10]. Dennis et al. [11] updated an earlier 2010 longitudinal study of LibQUAL+™ qualitative and quantitative data from the University of Mississippi libraries.

Tajer [12] used the LibQUAL+™ model to assess the service quality of university libraries. Cabrerizo et al. [13] suggested that the LibQUAL+™ model has two major defects and used fuzzy language to correct them. Raza and Samim [14] showed that students have higher expectations for library services. Ramezani et al. [15] used the LibQUAL+™ model to assess library service quality at Iranian universities, and the results showed that users’ overall assessment of library quality is high. They used the LibQUAL+™ model to establish the assessment criteria system to determine the service quality of university libraries. Hunter and Perret [16] found that users of larger, better-funded libraries have higher expectations for information resource availability but not higher satisfaction scores. Additionally, there was no significant correlation between library usage statistics and user satisfaction. To assess the service quality of academic libraries, it is essential to determine key dimensions for evaluating service quality and satisfaction based on the available literature.

Authors such as Hu et al. [17] studied the factors influencing users’ perception of university digital libraries in China using the structural equation model. Most studies in Chinese on service quality assessment focus on the service quality of “211” and “985” key national university libraries in China, but fewer studies have been conducted on the academic service quality of non-state-owned universities. With investments in information technology, university libraries gradually develop towards being more intelligent and networked, which not only greatly improve the library management level and service quality but also make library functions more professional and diversified.

Moreover, in China’s non-state-owned colleges, there is a lack of electronic access to academic services, less literature discussion, and a lack of tools available to assess the quality of web-based library academic services—as determined by Hernon and Calvert [18]—making it difficult to evaluate the extent to which academic services provided by the library meet the needs and requirements of users. Few studies in the literature have examined the concept of academic service quality in a non-state-owned university environment, and none explore whether a comprehensive view of service quality can lead to the increased use of library academic network services.

To summarize, the importance of competition and service resources provided by information services is increasing; users may prefer to use other non-library internet services [19]. Different cultural values may also affect customer perception and service experience [20]. Digital skill divides still exist, which means that library services are difficult to evaluate under the influence of the information and communication technologies (ICTs) environment, which was determined by Ben Youssef et al. [21]. Additionally, few studies have examined the existence of academic services in the private school environment in China, and there is little discussion about whether a comprehensive view of service quality can lead to an increase in the quality of library services.

The research objectives are:

(1)

To identify the key attributes of the dimensions of perception-based library academic service quality.

(2)

To determine the relative importance of the perceived library service quality to the overall service quality.

(3)

To evaluate a library academic service quality model focusing on the expectation factor from the customer satisfaction matrix to improve the service quality of libraries.

The remainder of this paper is structured as follows: Section 2 provides a literature review, discusses the assessment of service quality, and presents the models for measuring users’ satisfaction with library services. Section 3 describes the samples and methods. Section 4 summarizes the findings and discusses the results. Section 5 concludes the paper.

2. Literature Review

2.1. Service Quality

The assessment of service quality and its content has been debated by the academic community. Producers believe that quality comes from producing the best products with the most economical means [22]. Additionally, such products will attract consumers’ positive attention and offer consistent specifications. In order to measure product quality better, enterprises formulate standard procedures and product specifications for their product quality assessment standard. Lehtinen and Lehtinen [23] indicated that service quality mainly comprises process quality and result quality. Cronin and Taylor [24] proposed that service quality is the difference between the desired services and the perceived services that are actually received. There is also the web-based quality of service, defined as services provided through information and communication technologies [25,26,27]. Lancaster and López [28] find satisfaction as the difference between service expectations and perceived performance. SERVQUAL and ES-QUAL [29] both measure the quality of electronic services. However, the number of dimensions to measure the quality of service and the interrelationships between these dimensions are problematic [29,30]. Malik and Malik [31] found that a disparity between the expectations and perceptions of respondents using a measure of SERVQUAL indicates that the library should enhance its services, especially in staff.

Most studies use quality of service tools from the commercial and market sectors, particularly SERVQUAL, SERVPERF, and e-SERVQUAL, which may not be applicable to non-profit library services in higher education. The most commonly used quality of service measurement tool in libraries and information services is LibQUAL+™. However, the quality of service is also an inconsistent measure of the level of match between service levels and customer expectations [29] based on the inconsistency theory. Some of the main objections relate to the predictive power of instruments, the validity of the five-dimension structure, and the length of the questionnaire [23,32]. In addition, one of the benefits of measuring the quality of library services is the standardization of a measure across libraries [33].

2.2. LibQUAL+™

LibQUAL+™ has become an international standard tool for measuring library satisfaction. LibQUAL+™ was developed to address the cross-pressure from the university to understand the benefits of investing in its library during the changing period and to demonstrate the value and effectiveness of the services provided [34]. The literature on LibQUAL+™ focuses on the instrument itself, the reliability and effectiveness of the instrument, and the methods used to analyze and interpret quantitative findings. Early approaches to LibQUAL+™ works of literature focus on the quantitative analysis of survey results, qualitative analysis, and case studies. Few studies have used expert interviews of administrating LibQUAL+™ or combined quantitative and qualitative methods. Only a few studies have studied LibQUAL+™ nationally or internationally, and most of their studies only discuss common patterns and experiences related to library assessment [35,36]. The goals of LibQUAL+™ are to help libraries better understand user perceptions of library service quality and enhance library staff members’ analytical skills for interpreting and acting on data. Recently, studies of digital libraries (DLs) have contributed evaluation criteria to accessibility, usage of interface user engagement, and collection quality [37].

Researchers’ expectations and requirements for better quality and functional library services are increasing, especially in academic digital libraries. Therefore, the main challenge is how to measure the performance of its digital library from the user’s perspective and the extent to which its digital library meets user needs [38,39,40]. Several digital library quality assessment methods based on user perception have been proposed. These methods provide users with a perceived performance of the services provided by the academic digital library but do not give any suggestions for improvement [41]. Focusing on the behavioral intent relationships with user behavior and attitudes, in Berlak et al.’s paper [42], for instance, the use of the Unified Theory of Technology Acceptance and Use (UTAUT) was adopted. In a recent paper, Zeebaree et al. [43] proposed an extended UTAUT model for linkages between citizens’ acceptance of e-government services and their expectations of their adoption and sustainability. Lněnička et al. [44] investigated the adoption of open government data among students and confirmed that six constructs facilitated significant relationships with behavioral intention. The rise of digital technology has had a major impact on the service of library quality. Masa’deh et al. [45] examined the impacts of e-textbooks on the academic achievement of university students in a bilingual environment in Jordan using structural equation modeling (SEM) analysis based on the theory of planned behavior (TPB). In this case, artificial intelligence (AI) was employed via five machine learning (ML) techniques.

To summarize, there is no unified conclusion about evaluating a system of library service quality, and thus the quality of the service of a library is difficult to measure and evaluate. These new technologies can make assessments of service quality easier in some ways; however, such modeling is unable to integrate these tools in a context of uncertainty and imprecision proposed by Lizarelli et al. [46]. The use of the fuzzy method in this study is an adaptable technique that deals with the problems of qualitative aspects, including expert opinions, of user satisfaction. Additionally, the fuzzy approach assists understanding by users when the context is fluid or the concepts are abstract, and it decreases the effects of noise [41].

3. Methodology

This study established the questionnaire by the Fuzzy Delphi Method (FDM) and used the Two-Dimensional Quality Method proposed by Kano et al. [47] to analyze the difference in the perception of the library academic service quality of non-state-owned universities, making it the basis for improving library service quality.

3.1. Fuzzy Delphi Method

In the traditional Delphi Method, many iterations of questionnaires are required to pursue the agreement of experts, but this consumes cost and time, and during the computational process, only intermediate data are taken as the range of expert opinions, leading to the neglect and distortion of expert opinions. Ishikawa et al. [48] imported the concept of fuzzy theory into the traditional Delphi Method. Its advantages are as follows.

(1)

The expert opinions can be expressed completely.

(2)

The number of surveys can be reduced, and time and expense can decrease.

(3)

The fuzzy theory is more rational and desirable for expert cognition.

(4)

The individual attributes proposed by experts can also be considered.

Through deduction of the fuzzy theory, the consensus is determined as a concept of an average number. Jeng [49] partially corrected this method and used two triangular fuzzy numbers to obtain Gⁱ (see Figure 1), which is the importance value of expert consensus. It is more objective and reasonable than the average obtained by using one triangular fuzzy number and reduces the number of repeated questionnaire surveys. The steps are below.

Step 1: Design a fuzzy expert questionnaire and form an expert group. Each expert is individually related to each assessment item and gives a numerical interval. In the numerical interval, the minimum value represents the expert’s most conservative cognitive value for the quantitative fraction of the assessment item, and the maximum value represents the expert’s most optimistic cognitive value for the quantitative fraction of the assessment item.

Step 2: In the following order, gather statistics on the most conservative cognitive value and the most optimistic cognitive value given by all the experts for each assessment item I; delete the extreme values outside twice the standard deviation; and then calculate the minimum value $C_L^i$, the geometric average $C_M^i$, and the maximum value $C_U^i$ of the most conservative cognitive value as well as the minimum value $O_L^i$, the geometric average $O_M^i$, and the maximum value $O_U^i$ of the most optimistic cognitive value.

Step 3: Obtain the most conservative cognitive triangular fuzzy number $C^i=(C_L^i,C_M^i,C_U^i)$ and the most optimistic triangular fuzzy number $O^i=(O_L^i,O_M^i,O_U^i)$ of each assessment item i built in Step 2.

Step 4: Verify whether the expert opinions reach a consensus; this can be judged in the following ways:

(1) It may be that two triangular fuzzy numbers are not overlapped ($C_U^i\leqq O_L^i$); this represents that the interval values of all expert opinions have a consensus segment and that the opinions tend to fall within the range of this consensus segment. In this way, the consensus importance value $G^i$ of this assessment item i equals the arithmetic average of $C_M^i$ and $O_M^i$ (see Equation (1)).

$$G^i=\frac{C_M^i+O_M^i}{2}$$

(1)

(2) If two triangular fuzzy numbers are overlapped ($C_U^i$>$O_L^i$), and $Z^i=C_U^i-O_L^i$, the gray zone of fuzzy relation is less than $M^i=O_M^i-C_M^i$, the interval range of the optimistic cognitive geometric average, and the conservative cognitive geometric average of the expert assessment items. This represents that the interval values of all expert opinions have no consensus segment, but opinions are not divergent due to the big difference between the two experts’ opinions with extreme values (the minimum value in optimistic cognition and the maximum value in conservative cognition) and other experts’ opinions. In this way, the consensus importance value $G^i$ of this assessment item i equals the fuzzy set obtained from the intersection (mix) operation of the fuzzy relation of two triangular fuzzy numbers so as to determine the quantitative fraction of this fuzzy set with the maximum membership degree (see Equation (2)).

$$G^i=\frac{\left[\left(C_U^i\times O_M^i\right)-\left(O_L^i\times C_M^i\right)\right]}{\left[\left(C_U^i-C_M^i\right)+\left(O_M^i-O_L^i\right)\right]}$$

(2)

(3) If two triangular fuzzy numbers are overlapped ($C_U^i$ > $O_L^i$), and $Z^i=C_U^i-O_L^i$, the gray zone of fuzzy relation is greater than $M^i=O_M^i-C_M^i$, which is the interval range of the optimistic cognitive geometric average and the conservative cognitive geometric average of the expert assessment items. This represents that the interval values of all expert opinions have no consensus segment, and opinions are divergent due to the big difference between the two experts’ opinions with extreme values (the minimum value in optimistic cognition and the maximum value in conservative cognition) and other experts’ opinions. These assessment items of the divergent opinions (the minimum conservative cognitive value $O_L^i$, the geometric average and the maximum value) and (the minimum optimistic cognitive value, the geometric average and the maximum value) are provided to experts as reference. Furthermore, Steps 1 to 4 are to be repeated for the next questionnaire survey until all the assessment items are convergent and the consensus importance value $G^i$ is obtained.

3.2. Kano Two-Dimensional Quality Method

The concept of Two-Dimensional Quality comes from the Motivator–Hygiene Theory proposed by Kano et al. [47], Herzberg [51], and other scholars from Japan who employed the two-factor theory used for motivating employees to study service quality and proposed the Two-Dimensional Quality model. Their studies considered that customer satisfaction or dissatisfaction is not caused by the same aspect but by different aspects. In a separate discussion on customer satisfaction and dissatisfaction, Kano systematically organized customer demands in the first place to convert customer demands into products to improve enterprise competitiveness. Service quality is divided into five major categories (see Figure 2). First is the “Attractive quality element (A)”: when this quality attribute element is inadequate, customers do not feel dissatisfied. However, with this quality element, customers would be satisfied. Second is the “One-dimensional quality element (O)”: when this quality attribute element is more than adequate, customers feel more satisfied. In the case of lower adequacy, customers are less satisfied. Third is the “Must-be quality element (M)”: when this quality attribute element is adequate, customers take it for granted, and their satisfaction is not increased with adequacy. In the case of inadequacy, customers feel dissatisfied. Fourth is the “Indifferent quality element (I)”: regardless of whether the quality attribute element is adequate, customers do not care about it much and do not feel satisfied or dissatisfied. Fifth is the “Reverse quality element I”: when the quality attribute element is more than adequate, on the contrary, customers feel less satisfied.

Quality attributes are classified by the Kano classification method, which Matzler and Hinterhuber [52] corrected (see

Table 1. Kano’s evaluation model of teaching quality attributes questionnaire.

Library Quality		Insufficiency
Student’s Answer		5	4	3	2	1
Sufficiency	Satisfied (5)	Q	A	A	A	O
	Certainly (4)	R	I	I	I	M
	Irrelevant (3)	R	I	I	I	M
	Reluctantly (2)	R	I	I	I	M
	Dissatisfied (1)	R	R	R	R	Q

Notes: A: attractive; O: one-dimensional; M: must-be; I: indifference; R: reversal; Q: questionable.

) to divide quality elements into attractive quality, one-dimensional quality, indifferent quality, must-be quality, reverse quality, and invalid quality.

After correcting the Kano model and proposing the classification table of two-dimensional quality element correction, Matzler and Hinterhuber [52] proposed the customer satisfaction coefficient to determine how to improve some quality attribute elements and measured the increased Satisfaction Increment Index (SII) and Dissatisfaction Decrement Index (DDI). The results can be used as references to determine key service quality elements and improvement priority (Equations (3) and (4)):

SII = (A + O)/(A + O + M + I)

(3)

DDI = (O + M)/(A + O + M + I) * (−1)

(4)

(A: attractive, O: one-dimensional, M: must-be, I: indifference.)

After service element items are calculated and the SII and DDI of the questions of all dimensions are averaged, the obtained result is the general average of SII and DDI of all dimensions, and the key dimension is the highest one.

3.3. Research Subjects

Private higher education in China has gone through many changes since the 1990s. From the open government data in the past ten years (see

Table 2. Non-state-owned universities in China.

Panel A: Development Situation of Non-State-Owned Universities
Year	University	Non-State-Owned	Percent	Student	Non-State-Owned	Percent
2007	1908	297	15.57%	18,848,954	1,630,661	8.65%
2008	2263	640	28.28%	20,210,249	4,013,010	19.86%
2009	2305	658	28.55%	21,446,570	4,461,395	20.80%
2010	2358	676	28.67%	22,317,929	4,766,845	21.36%
2011	2409	698	28.97%	23,085,078	5,050,687	21.88%
2012	2442	707	28.95%	23,913,155	5,331,770	22.30%
2013	2491	718	28.82%	24,680,726	5,575,218	22.59%
2014	2529	728	28.79%	25,476,999	5,871,547	23.05%
2015	2560	734	28.67%	26,252,968	6,109,013	23.27%
2016	2596	741	28.54%	26,958,433	6,162,035	22.86%
2017	2631	747	28.39%	27,535,869	6,284,554	22.82%
Panel B: 15 non-state-owned universities in Guangzhou City
Number of students in school			16,706
Paper collections			1,535,778
Electronic collections			1,154,950

Source: National Bureau of Statistics of China (http://www.stats.gov.cn) (accessed on 1 July 2022).

), the proportion of non-state-owned Chinese universities among all national universities has grown sharply from 2007 (15.57%) to 2017 (28.39%). The proportion of enrolled students at non-state-owned universities to enrolled students at all national universities has continuously increased from 2007 (8.65%) to 2017 (22.82%) as well, indicating that non-state-owned universities recruit a large number of students each year and play an important role in meeting the needs of the current society for higher education. The research object is the non-state-owned university in Guangzhou. There were 17 non-state-owned universities in 2019. Excluding 2 incomplete data sets, there are 15 college survey data sets. Panel B shows an average of about 16,000 students at school and houses over 1.5 million volumes of book collections and 1.5 million electronic collections. These data sources come from the National Bureau of Statistics of China (http://www.stats.gov.cn) (accessed on 1 July 2022).

Table 3. Data description.

Background Variables	Items	Sample	Percentage
Gender	Male	235	51.9%
	Female	218	48.1%
Grade	Freshman	90	19.9%
	Sophomore	101	22.3%
	Junior	105	23.2%
	Senior	157	34.7%
Number of visits	Once a week	92	20.3%
	More than twice a week	168	37.1%
	Once a month	43	9.5%
	Two to three times a month	100	22.1%
	No more than 6 times a year	50	11.0%
Purpose of the visit	To borrow or browse leisure books, audio-visual materials and magazines	83	18.3%
	To borrow or browse professional books, audio–visual materials, and magazines	129	28.5%
	To find the information required by the teacher	59	13.0%
	To find information about further studies, exams, or employment	18	4.0%
	For academic activities	13	2.9%
	Self-study	128	28.3%
	Using the database	10	2.2%

Source: questionnaire results.

shows the sample data of respondents, including gender, grade, number of visits, and purpose in order to understand the distribution of library users. According to the survey results, the gender ratio of respondents was 49.1% for women and 50.9% for men. The proportion of freshmen was 19.9%, the proportion of sophomores was 22.3%, the proportion of juniors was 23.2%, and the proportion of seniors was 34.7%. The proportion of library users was 20.3% once per week, 37.1% twice a week, and 9.5% once per month. The proportion of library visits for borrowing or browsing leisure books and audio–visual materials and magazines was 18.3%; the proportion of professional books, audio–visual materials, and magazines was 28.5%; the self-study ratio was 28.3%; and the database usage rate was only 2.2%.

3.4. Private Expert Questionnaire

This study proposed a framework for library service quality assessment based on the literature, used the Fuzzy Delphi Method to design an expert questionnaire, and issued and collected questionnaires. At this stage, experts in relevant fields were employed to evaluate the importance of all the questions and to verify the expert validity of the questionnaire contents by e-mail. The initial questionnaire was then confirmed after correction based on an analysis of the expert questionnaire results to determine the applicability and validity of this study’s questionnaire.

In this study, ten experts were invited to conduct expert validity verification on the content of the questionnaire, and the applicability of the library service index items was modified into a prediction questionnaire based on the results of the expert questionnaire to confirm the applicability and validity (see

Table 4. Expert decision-making for each interval value.

Most Conservative Cognitive Value	(1)	(2)	Most Optimistic Cognitive Value	(1)	(2)
Expert_1	6	7	Expert_1	10	10
Expert_2	6	6	Expert_2	10	10
Expert_3	4	7	Expert_3	7	10
Expert_4	6	6	Expert_4	10	10
Expert_5	7	7	Expert_5	10	10
Expert_6	3	2	Expert_6	10	9
Expert_7	4	4	Expert_7	8	9
Expert_8	5	5	Expert_8	10	10
Expert_9	5	7	Expert_9	10	10
Expert_10	5	6	Expert_10	10	10
Mean	5.1	5.7	Mean	9.5	9.8
S.D.	1.2	1.64	S.D.	1.08	0.42
S.D. (−2 times)	2.41	2.43	S.D. (−2 times)	7.34	8.96
S.D. (+2 times)	7.49	8.97	S.D. (+2 times)	11.67	10.64
$\mathrm{Min} \left(C_L\right)$	3	4	$\mathrm{Min} \left(C_L\right)$	8	9
$\mathrm{Mean} \left(C_M\right)$	4.96	5.39	$\mathrm{Mean} \left(C_M\right)$	9.44	9.79
$\mathrm{Max} \left(C_U\right)$	7	7	$\mathrm{Max} \left(C_U\right)$	10	10

Source: Expert questionnaire results.

). Research experts who were chosen to participate had a Ph.D. and often used the library’s academic resources services or were experts in professional knowledge and practical experience in library work. The distribution and collection of the Fuzzy Delphi Method expert questionnaire were carried out by e-mail, and the recovery rate was 100%.

The assessment method of this questionnaire mainly focuses on the individual’s cognition of the scale of the score and gives a fuzzy interval value of 0 to 10 points. The higher the score, the more important the measurement item is. We can check the fuzzy relationship between the two fuzzy numbers to check whether the experts reached a consensus on the evaluation project. As shown in

Table 5. Expert opinions for consensus criteria.

Filter	(1)	(2)
Consensus value_G	7.59	8.34
Verification value_M-Z	5.47	6.40
Convergence	Yes	Yes
Suggested deletion	0	0

Source: Expert questionnaire results.

, if the verification value is greater than 0, the expert opinion reached a consensus. For the measurement projects that the experts suggested adding to or did not reach a consensus on, the next expert questionnaire was required, and the average opinion range of the experts was given to the experts for reference until all the measurement projects reached a consensus. Generally, in the range of 0–10, the threshold value is 6–7. The screening principle of this study is 80% of the number of experts who agree to be included in the measurement project, and the expert consensus importance value of the indicator is the threshold value of 7. When the two conditions are established at the same time, the evaluation framework of the service quality of the library can be included.

4. Results

According to the measurement index framework determined by the expert questionnaire, in total, 120 questionnaires were issued, and 100 valid questionnaires were collected. The effective rate of collected questionnaires was 83%. Cronbach’s α coefficient shows that the Cronbach’s α reliability of the forecast questionnaire was 0.968, and all dimensions were greater than 0.80 (see

Table 6. Reliability analysis summary sheet.

Dimension	Item	Item Number	Cronbach’s α after Items Are Deleted
			Kano Model
			Adequate Elements	Inadequate Elements
Emotional Service	1. Librarians are friendly, warm, and polite	A1	0.871	0.890
	2. Librarians are well-groomed and civilized	A2	0.862	0.885
	3. Librarians’ work behaviors are standardized and orderly	A3	0.891	0.869
	4. Librarians are willing to assist users in all the problems they encounter in the library	A4	0.846	0.869
	5. Librarians can answer questions from users at any time or online	A5	0.844	0.862
	6. Librarians have enough knowledge and the use of academic resources for users	A6	0.867	0.868
	7. Librarians reliably deal with users’ problems	A7	0.854	0.864
	8. Librarians pay attention to the opinions of users	A8	0.862	0.881
	Cronbach’s α		0.878	0.888
Physical Environment	9. The library is comfortable and attractive academic environment	B1	0.915	0.875
	10. The library is quiet and users can concentrate on reading	B2	0.905	0.887
	11. The environment in the library encourages users to learn and seek knowledge	B3	0.921	0.866
	12. The library provides spaces for personal or a team to help with study or research	B4	0.923	0.874
	13. Sufficient academic resources for access	B5	0.927	0.890
	Cronbach’s α		0.934	0.901
Information Control	14. Paper and electronic journals stored in the library can meet demands	C1	0.936	0.939
	15. Paper data can meet demands	C2	0.939	0.938
	16. Electronic data can meet demands	C3	0.942	0.938
	17. Library collections are effective in time and novel	C4	0.937	0.938
	18. Guides and marks in the library are clear, easy to understand, and convenient for users to obtain required resources	C5	0.937	0.936
	19. Computers and other devices can make it easier for users to search for the required information	C6	0.937	0.935
	20. The library website can enable users to find the required information	C7	0.941	0.935
	21. The reservation and renewal process of the library is convenient	C8	0.937	0.936
	22. Academic resources can be obtained through campus cooperation	C9	0.941	0.936
	23. Electronic resources are available outside the library	C10	0.944	0.937
	24. The library has an academic database teaching and learning camp	C11	0.940	0.938
	25. Develop education and academic ethics for library use	C12	0.940	0.941
	Cronbach’s α		0.944	0.942

Source: Questionnaire results.

) and were adopted. The questionnaires were formally issued, and 505 questionnaires were collected, including 453 valid questionnaires, and the effective rate was 89.7%.

4.1. Classification of the Kano Model Elements

The service quality items of the case library were classified according to Table 1, in which 25 quality items were, respectively, classified as attractive quality, one-dimensional quality, must-be quality, indifferent quality, and reverse quality. However, in the case of the same category of different two-dimensional quality attributes, the criterion of two-dimensional quality attribute classification is M > O > A > I [53].

4.2. Classification of Quality Elements

From the analysis of statistical results, library service quality items are classified as attractive quality, one-dimensional quality, and indifferent quality, excluding must-be quality and reverse quality. Furthermore, except for the information control dimension being classified as a one-dimensional quality and indifferent quality, emotional service and physical environment dimensions are of one-dimensional quality (see

Table 7. Category of teaching quality attributes in Kano’s model.

Dim.	Item	%					Class.	SII	DDI	Dim. Class.	SII Dim.	DDI Dim.
		A	O	M	I	R
Emotional Service	A1	24.7	40.8	12.8	20.9	0.9	O	0.659	−0.541	O	0.668	−0.444
	A2	27.6	29.8	10.8	30.7	0	I	0.579	−0.410
	A3	28.9	36.4	10.8	23.6	0	O	0.655	−0.473
	A4	21.6	49.7	9.3	19.1	0	O	0.715	−0.591
	A5	40.4	26.0	10.6	21.7	0	A	0.672	−0.371
	A6	43.6	24.9	6.4	22.8	0.2	A	0.705	−0.318
	A7	36.4	32.2	7.9	22.6	0	A	0.694	−0.404
	A8	30.4	34.7	9.4	24.3	0.2	O	0.656	−0.444
Physical Environment	B1	14.4	62.0	9.5	13.4	0.2	O	0.769	−0.720	O	0.736	−0.594
	B2	15.7	60.9	8.4	14.3	0.2	O	0.771	−0.698
	B3	29.2	45.5	8.0	16.7	0	O	0.751	−0.538
	B4	26.4	48.8	8.0	16.2	0.2	O	0.756	−0.570
	B5	21.9	48.8	12.2	16.4	0	O	0.711	−0.613
Information Control	C1	31.9	33.6	9.3	24.2	0	O	0.661	−0.432	O/I	0.632	−0.431
	C2	30.8	29.6	7.9	31.3	0	A	0.608	−0.375
	C3	32.2	32.0	8.1	26.9	0.2	A	0.647	−0.404
	C4	34.7	22.3	6.2	36.3	0	I	0.572	−0.286
	C5	21.7	54.3	10.2	13.1	0.2	O	0.764	−0.649
	C6	25.5	49.7	8.8	15.5	0	O	0.756	−0.589
	C7	29.6	40.0	7.9	21.9	0	O	0.700	−0.480
	C8	28.7	37.3	8.3	24.8	0.4	O	0.664	−0.460
	C9	31.1	28.7	6.1	33.2	0.4	I	0.601	−0.353
	C10	29.1	22.7	5.7	40.0	1.0	I	0.532	−0.292
	C11	22.3	33.6	8.1	34.8	0.4	I	0.565	−0.422
	C12	24.2	18.1	6.9	49.2	0.8	I	0.430	−0.253

Note: M: Must-be quality; O: One-dimensional quality; A: Attractive quality; I: Indifferent quality.

).

The results are as follows:

• Attractive quality: Respondents considered five service quality elements, A5, A6, A7, C2, and C3. Respondents suggested that they would be satisfied if the library offered the above service elements, and they would feel no difference or not mind in the case of inadequate supply. Consequently, at present, such services in the library are between the ordinary level and satisfactory level.
• One-dimensional quality: There are 14 service quality elements, including A1, A3, A4, A8, B1, B2, B3, B4, B5, C1, C5, C6, C7, and C8. The actual operations of the case library are mostly the basic services that are the service items prioritized to be provided. The more adequate the quality attribute elements are, the more satisfied users are. Comparatively, the failure to provide such services immediately causes dissatisfaction, and the more inadequate the one-dimensional quality is, the less satisfied users are.
• With the coming of the information age, users are improving their capabilities to use modern facilities to obtain information. Hence, the attribute of the three services, including C6, C7, and C8, is of one-dimensional quality. In this case, improvements are made by adding self-service borrowing and returning equipment and re-planning the library website.
• Indifferent quality: Respondents considered six items, including A2, C4, C9, C10, C11, and C12. In this case, for Item A2, the service staff are basically student volunteers and have good images. Most respondents believe that this service quality element is adequate and has no effect on the satisfaction with the library service. Item C9 is about academic resources that can be obtained through campus cooperation. Inter-library cooperation service is a cooperative way to achieve resource sharing and make up for the lack of collections.

For “C11. The library has an academic database teaching and learning camp”, whether this service element is provided adequately does not cause satisfaction or dissatisfaction. The library’s database mainly provides tools and books, which can be used to query, browse, print, and download the required data and papers online, and has new functions such as diary notification, search notification, and citation notification. The library provides learning and learning database teaching camps, which help researchers to improve their work efficiency. However, due to the small number of users, statistics show that users are not paying much attention to this service.

For “C10. Electronic resources are available outside the library”, less attention is paid to this service element because students can easily purchase electronic resources at inexpensive costs, which can be used immediately after payment.

For “C12. Develop education and academic ethics for library use”, regardless of whether this service quality element is provided adequately, users’ assessment of the overall service quality of the library does not increase or reduce. Academic ethics education aims to cultivate good academic ethics in teachers and students of higher education and ensure the appropriateness and legitimacy of academic activities. In this study, teachers mentioned how they used library resources and emphasized the importance of academic ethics when teaching, but most users pay less attention to library conduct education.

4.3. Customer Satisfaction Coefficient

User satisfaction is crucial to the growth of enterprises [54]. The biggest enlightenment of the Kano model is the non-linear and asymmetric relationship between quality performance and user cognition. Traditionally, it is considered that a linear and symmetric relationship between both is just a special case, in fact, and the indices of quality improvement performance should be simultaneously considered from two perspectives, which are user satisfaction increment and user dissatisfaction decrement. In order to determine users’ expectations for all library service quality elements, the user satisfaction coefficient is calculated to confirm the user satisfaction increment and user dissatisfaction decrement that can occur simultaneously when a quality element is improved (see Table 7). The closer the satisfaction increment index is to 1, the greater the effects of this element on user satisfaction are; the closer the absolute value of the dissatisfaction decrement index is to 1, the greater the effects of this element on user dissatisfaction decrement are.

After the user satisfaction coefficient of the service element items is obtained by calculation, the satisfaction increment coefficient and the dissatisfaction decrement coefficient are sorted (see

Table 8. User satisfaction index sort.

SII sort		DDI sort
Item Number	SII	Item Number	DDI
B2	0.771	B1	0.720
B1	0.769	B2	0.698
C5	0.764	C5	0.649
B4	0.756	B5	0.613
C6	0.756	A4	0.591
B3	0.751	C6	0.589
A4	0.715	B4	0.570
B5	0.711	A1	0.541
A6	0.705	B3	0.538
C7	0.700	C7	0.480
A7	0.694	A3	0.473
A5	0.672	C8	0.460
C8	0.664	A8	0.444
C1	0.661	C1	0.432
A1	0.659	C11	0.422
A8	0.656	A2	0.410
A3	0.655	C3	0.404
C3	0.647	A7	0.404
C2	0.608	C2	0.375
C9	0.601	A5	0.371
A2	0.579	C9	0.353
C4	0.572	A6	0.318
C11	0.565	C10	0.292
C10	0.532	C4	0.286
C12	0.430	C12	0.253
Mean value	0.664	Mean value	0.467

Note: Gray bottom greater than the mean value.

). In the satisfaction increment coefficient, there are 10 service element items that need to be improved, which are “B2, B1, C5, B4, C6, B3, A4, B5, A6, and C7”. These are the items that have a satisfaction increment coefficient and dissatisfaction decrement coefficient simultaneously with a value higher than the average number. Therefore, regardless of the aspects, just by improving these items, an efficiently upgraded service quality can be obtained.

4.4. User Satisfaction Matrix

In order to show the significance of the user satisfaction coefficient more clearly, when measuring service quality and strengthening the primary key attributes, this study established the user satisfaction matrix and took the satisfaction increment coefficient as the horizontal axis, dissatisfaction decrement coefficient as the vertical axis, and the individual general average of satisfaction increment coefficient and dissatisfaction decrement coefficient as the cross-point (see Table 8) in order to divide the coefficient distribution chart of all the questions into four quadrants and draw the locations of individual elements. Figure 3 shows the preferential improvement items of library quality elements.

The elements in the first quadrant indicate that they highly increase user satisfaction and decrease user dissatisfaction, so more resources should be invested to improve these service elements. The elements in the second quadrant slightly increase user satisfaction, eliminate user dissatisfaction, and the essential elements can be maintained at a certain level. The elements in the third quadrant are not very helpful for increasing satisfaction value and decreasing the dissatisfaction value. Even if a lot of manpower and materials are devoted to improving the quality of these services, the overall efficiency is not enhanced. Therefore, it is not necessary to put much effort into these service elements. The elements in the fourth quadrant can increase user satisfaction but cannot eliminate dissatisfaction.

In Figure 3, the priority to improve service quality elements can be determined according to the features of the quadrant, including 25 quality elements. The first quadrant shows that it highly increases satisfaction and decreases dissatisfaction, and nine items are expectation factors: A4, B1, B2, B3, B4, B5, C5, C6, and C7. The quality attributes in this zone both increase motivation factors of user satisfaction and decrease hygiene factors of user dissatisfaction. Hence, the library should invest more resources into improving these quality elements, and administrators should prioritize improving the quality elements in this zone.

The quality elements in the second quadrant are attractive factors, including two items, A1 and A3. With the effects of hygiene factors, the two items can decrease the dissatisfaction coefficient and slightly increase user satisfaction but can also eliminate user dissatisfaction. These service elements should be maintained at a certain level. The administrators should first invest resources to eliminate the causes of user dissatisfaction.

The quality elements in the third quadrant are indifferent factors, including 10 items, A2, A8, C1, C2, C3, C4, C9, C10, C11, and C12. They slightly decrease the user dissatisfaction coefficient and increase the user satisfaction coefficient, and this zone is the quality attributes that users pay the least attention to. After the attributes in other zones are improved, if any, the extra resources can be invested into improving quality attributes.

The quality elements in the fourth quadrant are must-be factors, including three items, A5, A6, and A7. The effects of motivation factors can increase user satisfaction but have little effect on eliminating user dissatisfaction. Improving these service quality elements can increase user satisfaction but cannot reduce user dissatisfaction. The administrators can maintain a certain level of service and expect to improve the elements in this zone after the improvement of expectation factors and attractive factors.

4.5. Discussion

In this study, the three dimensions of the case library exhibit strictly speaking one-dimensional quality (see Table 7), indicating that all service items are, respectively, one-dimensional quality and attractive quality. Generally, in the traditional concept, the physical environment services of a library are regarded as basic services—that is, a library provides physical environment services as a matter of course, and adequate supply does not increase customer satisfaction; otherwise, customers would be dissatisfied. However, in this study, the classification of service quality attributes of the physical environment broke the existing concept that it is generally regarded as the essential quality. The gracious attitude and politeness of librarians are also signs of respect for users. Real-time response and correct feedback through online services is also a positive, friendly attitude. Similarly, librarians can solve problems correctly and are important factors for users to be respected.

As seen in Figure 3, there are nine elements in the first quadrant, covering, respectively, “A4. Librarians are willing to assist users in all the problems they encounter in the library”; “B1. The library is comfortable and attractive academic environment”; “B2. The library is quiet and users can concentrate on reading”; “B3. The environment in the library encourages users to learn and seek knowledge”; “B4. The library provides spaces for personal or a team to help with study or research”; “B5. Sufficient academic resources for access”; “C5. Guides and marks in the library are clear, easy to understand, and convenient for users to obtain required resources”; “C6. Computers and other devices can make it easier for users to search the required information”; and “C7. The library website can enable users to find the required information”. The administrators should prioritize improving the quality elements in this zone, and user satisfaction with service quality can be significantly increased. Similarly, Bussell et al. [55] concluded that students’ preference to learn on demand depends upon available resources that support this type of learning style.

The service elements in the improvement zone with the highest priority are “B4. The library provides spaces for personal or a team to help with study or research” and “B5. Sufficient academic resources for access”. Therefore, with limited resources, administrators should improve these two elements as a priority to improve user satisfaction with the library service. With adequate resources, “A4. Librarians are willing to assist users in all the problems they encounter in the library”, “B1. The library is comfortable and attractive academic environment”, “B2. The library is quiet and users can concentrate on reading”, “B3. The environment in the library encourages users to learn and seek knowledge”, “B4. The library provides spaces for personal or a team to help with study or research”, “C6. Computers and other devices can make it easier for users to search the required information”, and “C7. The library website can enable users to find the required information” should be improved next. Regrettably, limited access to the internet, such as the Google web browser, results in a bias in the measurement of academic satisfaction. However, Google and Google Scholar are heavily relied upon for accessing information by graduate students, as proposed by Nicholas et al. [56].

5. Conclusions and Suggestions

The purpose of this study is to investigate the differences in the perception and quality of academic services in non-state-owned universities and to explore strategies for improving the quality of library services. The respondents submitted 453 valid questionnaires for the study based on the Kano model, the service quality attributes were classified, and the fuzzy analysis method was used to incorporate expert opinions and analyze the user satisfaction matrix.

The results suggest that friendly librarian service can convey confidence in the professional knowledge, strengthen the user’s impression, and improve the service quality. The results indicate there is a disparity between the respondent’s opinions and expectations. Additionally, the convenience of data acquisition, the effectiveness of problem-solving, and the comfort of the venue environment can significantly improve user satisfaction with service quality. In the case of limited resources, administrators should prioritize the improvement of research spaces and sufficient academic resources, such as e-textbooks, as proposed by Masa’deh et al. [45]. In addition, the use of academic resources and the provision of academic ethics education are relatively less used—which may be limited by the composition of students—resulting in the lessened concern of students and the quality of service not being affected in this survey. Hence, the order of library service improvement should give priority to one-dimensional quality service items, followed by attractive quality and must-be quality service items.

The results also provide nine elements to improve the quality of service and two major improvements to enhance the perception and difference of service quality. With the development of network technology, library management staff can overcome the library’s system and service inadequacies, but they still need to be innovative and find new ways to promote their resources and services and strengthen their expertise. Although the library has successfully provided them with a good atmosphere, a large academic database, and other facilities, there is still room for improvement in staff training.

This study has some limitations. While this paper focused on undergraduates with faculty from non-state-owned universities, it might be worthwhile to examine teachers as users of other groups, user attributes in terms of frequency of usage, and training experiences with online instruction. Future research could examine the changing perceptions of other academic services as library services become more digital and impersonal under the current digitization and the COVID-19 pandemic environment. Guiding by these models, this study suggests that the role of digital capability for library administrators needs to be developed. Later studies can also reconsider using an extended UTAUT model proposed by Zeebaree et al. [43] to apply the concept of a suitability assessment for feedback according to expert opinions in different cases.