The Impact of Digitalization on the Categories of Users of Library Services in the Period of Aging and Population Decline: A Case Study of Sapporo, Japan

Abstract

In relation to the restructuring of public facilities in the phase of aging and subsequent population decline, the digitalization of public services will reduce the cost of maintenance and management of public facilities and contribute to sustaining or even improving standards of services. This study compares attributes of users of physical and digitized electronic libraries, using the results of an original questionnaire survey conducted in Sapporo metropolitan area, Japan, analyzed via statistical methods. The results showed that electronic libraries provide their services to some different groups of citizens compared with physical libraries. Electronic libraries have a clear tendency to provide services to the working population, which means the service reaches more taxpayers who have contributed to the funding of public facilities but have not used them. On the other hand, the influence of age is not simple for physical and electronic library use, although there is certainly a trend toward higher use among younger age groups. The digital divide among the elderly is not clearly observed for the use of electronic libraries, which means that the digitalization of library services can be used by various categories of citizens, including the elderly, even in an aging society.

1. Introduction

In some countries and regions where the national population is aging, the sustainability of public services provided by existing public facilities is being questioned. Restructuring and reductions in public facilities and services to adapt to the shrinkage of demand and financial resources are under consideration both by the central government and local governments in Japan [1,2]. In doing so, the digitalization of public services is being discussed as an alternative, since it is not easy to simply reduce the number of facilities and services.

There are several benefits of digitization of public services. Bertot et al. [3] provided a general overview of innovations in public services due to digitalization and proposed the Digital Public Service Innovation Framework, which extends the “standard” provision of digital public services. In relation to the restructuring of public facilities in the phase of aging and subsequent population decline, the digitalization of public services will reduce the costs of maintenance and management of public facilities and contribute to sustaining or even improving standards of public services by allowing methods to provide public services much more efficiently, and even unmanned methods. In addition, digitalization could provide new and innovative public services.

In Japan, the Third National Spatial Strategy was formulated in July 2023 as a new national spatial plan for the whole national territory. It aims to create Regional Habitation Spheres (Chiiki-seikatu-ken), where various digital and physical public services are integrated [4]. In the process of enacting the National Spatial Plan, the Japanese government categorized public services into those that are easy to provide digitally and those that are difficult to provide digitally. While many public services, including supply and treatment services, such as water and waste, and welfare services, such as medical and nursing care, are required to be provided physically, the library service is considered to be a relatively viable option for public service through digitization.

The transition to digitalization is becoming a major issue in the discussion among library sectors around the world. Whereas the previous version of the Public Library Manifesto of IFLA (International Federation of Library Associations and Institutions) and UNESCO (United Nations Educational, Scientific, and Cultural Organization), published in 1994 [5], pointed to mere access to information and material, the new Manifesto in 2022 [6] emphasizes the importance of the provision of services both in-person and remotely through digital technologies. It is believed that digitization will enable comprehensive access to the more diverse tiers of library services that the public sector is responsible for.

Based on the above discussion, this study aims to compare the use of library services provided by digitized electronic libraries and those provided by existing physical public facilities by various attributes of users and discusses the potential of electronic libraries to contribute to the digitization of libraries. The author intends to explore the potential contribution of digitalization in the restructuring of public facilities in cities in Japan and other countries in a phase of population decline.

In this paper, electronic libraries are defined as a service that provides electronic books for users free of charge with limited licenses, based on the definition of Association for E-publishing Business Solution. The characteristic of electronic libraries that citizens can access them from outside the libraries, including their homes, by connecting to websites of electronic libraries, distinguishes them from services provided by traditional physical libraries and is also the focus of this study.

2. Existing Research on Library Digitization and Accessibility

2.1. Digitization of Libraries

Electronic library initiatives have been introduced and discussed in earnest since the 2010s, when electronic library systems began to advance. While there are several positive references to individual cases where services have been digitized, as described below, some literature [7,8,9,10] has emerged that expresses concern about the transformation of existing services.

The coronavirus pandemic, which began in 2020, has had a tremendous impact on library services and their digitalization, and has encouraged the use of electronic services. For example, the EBLIDA (European Bureau of Library Information and Documentation Association) considered library activities and trends that will become permanent in the post-COVID-19 age and identified five “new normals” for a European Library Agenda, including dedication to digital services and technologies [11]. However, there is still a paucity of literature discussing the evaluation of the implementation of electronic libraries. Matthews [12] insisted on the importance of the balance of physical and digital library services around governance models for library services based on the case of Scottish public libraries and the libraries’ responses to the coronavirus pandemic.

2.2. Library Service Standards and Accessibility

Various studies and editorials [13,14,15,16] have extensively discussed and evaluated the public nature of library services. Quantitative analysis of the level of general service in libraries began to emerge in the 2000s through several methods. For example, Hammond [17] assessed the relative efficiency of public library services of 99 public library systems in the UK by examining the relationship between library inputs and outputs using Data Envelopment Analysis (DEA) and insisted that scale inefficiency associated with over-subscription to serial publications can be seen in many library systems. There are some other studies in the UK, such as evaluating the downward trend in staff, spending and opening hours across children’s public library services [18], or estimating efficiency scores of the restructuring of local authority services [19].

Many studies on accessibility to libraries have emerged since the 2010s, when the Geographic Information System (GIS) and the associated data became widespread. In the US, Park [20] analyzed travel distances to libraries by tier. Adkins et al. [21] combined library openings and closings with census data to identify location bias. Donnelly [22] used spatial statistics, such as center of gravity, to show that the percentage of population within a library’s area varied by region and state. Donnelly [23] also showed that the average distance to the nearest library was 2.1 miles for the US as a whole but varied significantly by state.

Regarding other countries, Guo et al. [24] evaluated library services, including mobile libraries, via several methods for Hong Kong. In Canada, Allen [25] analyzed access to the nearest library by mode of transport and found that those who relied on public transport spent more time in the library. In Wales, Page et al. [26] found that the supply of library services had decreased in deprived areas as a result of fiscal austerity, creating spatial inequalities.

Based on a series of existing literature related to accessibility to public libraries, as shown above, Mandel et al. [27] searched for full-text library research that used GIS as a tool to evaluate library services and resources across two databases: Library and Information Science Source (LISS) and Library, Information Science, and Technology Abstracts (LISTA), and showed that GIS was primarily used to analyze service areas and manage facilities and collections.

2.3. Electronic Libraries and Accessibility

The discussion of the relationship between digitization and accessibility has centered on the issue of the digital divide, including the elderly and the poor, as exemplified in the papers by Kinney [28] and Comi et al. [29].

The COVID-19 pandemic was one of the major catalysts that led to the idea of connecting the issues of digitization and accessibility. Ćirić and Ćirić [30] compiled statistical data on access to digital content from public library websites before and after the Coronavirus pandemic in Serbia and reported that access was particularly high in March 2020, when the lockdown was most severe due to the COVID-19 pandemic, and this was reflected in reading time. However, they did not compare the results with physical library services.

Few studies of library digitization have targeted spatial accessibility, both before and after the COVID-19 pandemic. Higgs et al. [31] researched variations in accessibility to public library facilities in a local authority area in south Wales, using floating catchment area (FCA) approaches, and identified provision gaps of digital services. However, all digital services in the paper were considered to be services received by visiting a physical library, and they did not cover services that can be received electronically at home or any places other than libraries, such as digital library services.

3. Research Design

3.1. Acquirement of Data

This study used a statistical method of analyzing a questionnaire survey of citizens in the target area, as described below, about their use of libraries and electronic libraries.

As for the questionnaire survey, I mainly used response data on physical libraries and electronic libraries from an Internet survey conducted in October 2022 by Macromill, Inc., which is one of the major online survey firms in Japan and works with partner companies to utilize a large number of panels (respondents) to collect enough survey responses sought by subscribers without significant age or gender bias. I instructed this research firm on the specifics of the survey and its target respondents and asked the panel to respond using a web form provided by the company.

The survey consisted of a screening survey of 5357 residents in the target area (described below), asking about basic attributes and whether they use physical libraries, and a main survey of 1052 persons who reported using physical libraries in that survey, asking about detailed attributes, actual library use, and use of electronic libraries. The basic attributes (gender, age, occupation, etc.) that the survey firm recorded about the respondents were used as well.

The target area was Sapporo City, which was one of the first municipalities in Japan to introduce an electronic library in October 2014, and 11 neighboring municipalities that joined the scheme of inter-municipal cooperation, called a Coordinated Core Metropolitan Area (‘Renkei-chusu-toshiken’), where the day/night population ratio was set as one of the conditions (Figure 1).

Information on physical libraries within the target area was obtained from the Japan Library Association’s “Libraries in Japan 2019 Electronic Media Edition”. This dataset includes all physical libraries established by Article 2 of the Library Law in Japan. The survey added information on several other libraries not included in this dataset by providing an open-ended response field for using libraries that were not included in the older dataset. As a result, the number of physical libraries included in this study was 82.

On the other hand, in this metropolitan area, electronic libraries were provided in two cities at the time of the survey, including Eniwa City and Sapporo City, as well as in Hokkaido prefecture, a regional government. Citizens can use the physical and electronic libraries of the municipality in which they live. Moreover, commuters can also use the physical and electronic libraries of the municipalities to which they travel during the daytime. Registration should be performed once a year, at a physical library in person, along with personal identification. Therefore, this paper assumed that all electronic library users visited physical libraries at least once a year.

3.2. Attributes of Citizens

In this study, attributes of citizens that were most likely to be related to whether or not they used the system were established, as shown in

Table 1. Attributes of respondents and corresponding variables.

Attributes of Respondents Categories		Variables
(for physical and electronic libraries)
1. Basic Attributes	gender	dummy (male)
	age	age, dummies in 10-year increments
	occupation	dummies by several categories (employed, housewife, student, junior college/university/graduate student, high school student, technical college student)
	educational background	dummy (undergraduates and postgraduates)
	marriage	dummy (marriage)
	children	number of people, dummies by several categories (infants, toddlers, lower and upper elementary school students, middle school students, high school students)
2. Reading Habit Attributes	paper media	reading time per month (hours), dummies by three categories (books, newspaper and journals, comics)
	electronic media	reading time per month (hours), dummies by three categories (books, newspaper and journals, comics)
3. Accessibility Attributes	nearest library	distance (km), dummy (living in the same municipality as the nearest library)
	driving habits	driving time per month (hours), dummy (drive on their own/ride in the family car)
4. Corona Alert Attributes		5-point scale values by four categories (general alertness, visit to crowded facilities/places, proximity to strangers, conversation with others)
(for electronic libraries)
5. Accessibility Attributes to Electronic libraries		dummy (provision of electronic library service in the regidence municipality)
6. Electronic Device Use Attributes		dummies by several gadgets (computers, cell phones, smartphones, tablet devices, TVs, home game consoles, other devices)

, and collected from respondents through a questionnaire survey. In order to make the relationship between library usage and attributes objective and clear, questions posed to the panel were narrowed down and selected so that there were no potential confounding variables with large effects, and the questions were set simply to avoid inductive responses.

The attributes were broadly divided into three categories: general attributes, such as age, gender, and occupation, access attributes, represented by the distance between the residence and the library and whether or not a car was available, and reading habit attributes related to library use. These attributes could reveal the reality of the digital divide for the elderly discussed in the existing literature, as well as the advantages of digital service provision for citizens whose nearest public facilities are far away, and the state of service provision for working people who are less likely to receive public services.

The approximate locations of the respondents’ residences were ascertained by adding the zip code to the questionnaire. From the coordinates of all known library addresses and the addresses represented by the respondents’ zip codes, the nearest library for each respondent was extracted and its distance was calculated. All other attributes, including driving time, were also included in the data, all of which were obtained directly from the respondents through their responses (Table 1).

Also, since the survey was conducted in 2022, alertness against COVID-19 was added to the three attributes.

In addition to the above, the electronic library usage survey for physical library users included time spent in the physical library and the relationship with digitization-related attributes, such as the extent of electronic book usage, which were added for the categories of electronic library usage.

On the other hand, service levels such as library size and number of books were not taken into account here, since these factors are considered not to have a significant impact on the usage.

4. Results

4.1. Correlation between Physical Library Use and Citizen Attributes

First, from the screening survey described above, I have described how citizen demographics are related to physical library use. The dummy variables in the above questionnaire results were analyzed by the χ-square test, measuring whether the variables presented on the nominal scale among the attributes listed in Table 1 made a significant difference in the use of physical and electronic libraries, while the interval and proportionality scales were analyzed by the correlation coefficient in relation to utilization. The results are shown on the left side of

Table 2. χ-square tests between variables of attribute categories and use of physical and electronic libraries.

Attributes Categories	Sub-Categories	Physical Library			χ2 Test and p Value			Electronic Library			χ2 Test and p Value
		Resp.	Lib. Users	%				Resp.	Lib. Users	%
All		5357	1282	23.9	-			1052	216	20.5	-
1. Basic Attributes
age	10s	81	24	29.6	0.226			19	3	15.8	0.606
	20s	391	100	25.6	0.429			70	23	32.9	0.008	**	+
	30s	815	232	28.5	<0.001	***	+	185	54	29.2	0.001	**	+
	40s	1201	302	25.1	0.263			257	57	22.2	0.452
	50s	1419	279	19.7	<0.001	***	−	225	32	14.2	0.008	**	−
	60s	1008	222	22.0	0.115			188	33	17.6	0.264
	70s	397	112	28.2	0.038	*	+	98	12	12.2	0.033	*	−
	80s	45	11	24.4	0.935			10	2	20.0	0.967
	over 65	967	242	25.0	0.378			210	28	13.3	0.004	**	−
gender	male	2622	579	22.1	0.002	**	−	470	100	21.3	0.591
marriage	married	3261	828	25.4	0.002	**	+	683	137	20.1	0.605
occupation	employed	2152	491	22.8	0.117			380	109	28.8	<0.001	***	+
	housewife	797	214	26.9	0.036	*	+	189	27	14.3	0.019	*	−
	student	142	45	31.7	0.028	*	+	36	7	19.4	0.869
	junior college/university/graduate student	86	30	34.9	0.016	*	+	23	6	26.1	0.505
	high school/technical college student	56	15	26.8	0.615			13	1	7.7	0.249
educational background: university and post graduate		1976	625	31.6	<0.001	***	+	500	113	22.6	0.114
(having) children	infants	244	87	35.7	<0.001	***	+	70	15	21.4	0.848
	toddlers	307	127	41.4	<0.001	***	+	100	36	36.0	<0.001	***	+
	infants and toddlers	459	175	38.1	<0.001	***	+	138	41	29.7	0.004	**	+
	lower grades of elementary school	328	133	40.5	<0.001	***	+	110	28	25.5	0.177
	higher grades of elementary school	315	122	38.7	<0.001	***	+	99	30	30.3	0.011	*	+
	infants to elementar school students	840	316	37.6	<0.001	***	+	255	64	25.1	0.038	*	+
	middle school students	343	113	32.9	<0.001	***	+	92	24	26.1	0.167
	high school students	393	97	24.7	0.717			75	16	21.3	0.859
	middle and high school students	619	173	27.9	0.013	*	+	137	30	21.9	0.671
2. Reading Habit Attributes
paper media	books	2439	1017	41.7	<0.001	***	+	842	177	21.0	0.432
	newspaper and journals	2384	808	33.9	<0.001	***	+	677	155	22.9	0.011	*	+
	comics			36.7	<0.001	***	+	434	114	26.3	<0.001	***	+
electronic media	books	1120	416	37.1	<0.001	***	+	337	132	39.2	<0.001	***	+
	newspaper and journals	1502	534	35.6	<0.001	***	+	452	137	30.3	<0.001	***	+
	comics	1406	473	33.6	<0.001	***	+	387	130	33.6	<0.001	***	+
3. Accessibility Attributes
living in the same municipality as the nearest library		5263	1264	24.0	0.273			1034	211	20.4	0.443
accessibility	drive on their own	3131	834	26.6	<0.001	***	+	680	157	23.1	0.006	**	+
by automobiles	ride in the family car	2414	704	29.2	<0.001	***	+	579	141	24.4	<0.001	***	+
4. Corona Alert Attributes		(no dummy variables are generated)
5. Accessibility Attributes to Electronic libraries
living in the municipality providing electronic library								835	194	23.2	<0.001	***	+
6. Electronic Device Use Attributes
	computers							857	186	21.7	0.049	*	+
	cell phones							207	75	36.2	<0.001	***	+
	smartphones							932	188	20.2	0.42
	tablet devices							359	102	28.4	<0.001	***	+
	TVs							280	84	30.0	<0.001	***	+
	home game consoles							175	68	38.9	<0.001	***	+
	other devices							5	5	100.0	<0.001	***	+

Note: “Resp.”: No of Respondents, “Lib. Users”: No of Library Users, ***: 0.1%, **: 1%, *: 5% significant; +: “positive”, −: “negative”.

and

Table 3. Correlation analysis between variables of attribute categories and use of physical and electronic libraries.

Attributes of Respondents Categories	Sub-Categories	Variables	Physical Library			Electronic Library
			Coll. Coef.	p Value		Coll. Coef.	p Value
1. Basic Attributes
age		age	−0.033	0.016	*	−0.132	0.000	***
children	infants	number	0.055	0.000	***	0.055	0.075
	toddlers	number	0.094	0.000	***	0.146	0.000	***
	infants and toddlers	number	0.092	0.000	***	0.125	0.000	***
	lower grades of elementary school	number	0.093	0.000	***	0.069	0.026	*
	higher grades of elementary school	number	0.078	0.000	***	0.110	0.000	***
	infants to elementar school students	number	0.116	0.000	***	0.136	0.000	***
	middle school students	number	0.061	0.000	***	0.077	0.012	*
	high school students	number	0.011	0.431		0.042	0.178
	middle and high school students	number	0.041	0.002	**	0.070	0.024	*
2. Reading Habit Attributes
paper media	books	hour/month	0.362	0.000	***	0.088	0.004	**
	newspaper and journals	hour/month	0.142	0.000	***	0.109	0.000	***
	comics	hour/month	0.164	0.000	***	0.166	0.000	***
electronic media	books	hour/month	0.145	0.000	***	0.278	0.000	***
	newspaper and journals	hour/month	0.122	0.000	***	0.233	0.000	***
	comics	hour/month	0.094	0.000	***	0.186	0.000	***
3. Accessibility Attributes
Accessibility to libraries	distance to the nearest library	meter	−0.040	0.004	**	−0.019	0.548
Accessibility by automobiles	drive on their own	hour/month	0.021	0.119		0.043	0.163
	ride in the family car	hour/month	0.073	0.000	***	0.158	0.000	***
4. Corona Alert Attributes	general alertness	5-point scale	−0.059	0.000	***	0.030	0.330
	visit to crowded facilities	5-point scale	−0.066	0.000	***	0.026	0.404
	places, proximity to strangers	5-point scale	−0.032	0.018	*	−0.012	0.707
	conversation with others	5-point scale	−0.024	0.082		−0.021	0.497

Note: “Coll. Coef.”: Colleration Coefficient, ***: 0.1%, **: 1%, *: 5% significant.

, respectively.

For age/age, the correlation between simple age and use was significant but not very high. When analyzed by age dummies, the first significant increase in use was in the 30s, which is strongly related to the fact that they are in the child-rearing generation. Physical library use was significantly higher among the respondent group that actually had children of all ages, from infants to middle school students. The marriage dummy was positive, the male dummy was negative and significant, and the housewife dummy was positively significant, which was assumed to be due to the strong relationship with households raising children. On the other hand, it was significantly lower among those in their 50s and did not correlate with those aged 65 or older, the general threshold for the elderly in Japan.

The most significant of the other basic attributes was educational background. People who had obtained a university or post-graduate degree were significantly more likely to use physical libraries. The attribute of reading habits, which was related to library use, had a strong and significant effect on both paper media and electronic media, clearly showing a high correlation, as expected.

The distance from the residence to the nearest library was also strongly correlated with accessibility to libraries, indicating that proximity to libraries is important. On the other hand, it did not matter whether they lived in the same municipality as the nearest library, as a certain number of users may use libraries in other nearby municipalities. Another important factor related to accessibility attributes was the availability of automobile transportation. The correlation with driving by oneself was high, but the correlation with driving time was not significant, suggesting that the correlation was not with whether one usually drives, but with whether one can drive.

Of the four types of alertness against COVID-19, all but against conversation were significant. Conversation is thought to be influenced by the fact that most library users do not anticipate conversation when using the library.

4.2. Correlation between Electronic Library Use and Citizen Attributes

Next, the survey revealed how citizen demographics are related to the use of electronic libraries. The results are shown on the right side of Table 2 and Table 3, respectively, in the same manner as the use of physical libraries.

As for age, the negative correlation with the use of electronic libraries was stronger. Specifically, more respondents in their 20s and 30s used electronic libraries, while fewer respondents in their 50s and 70s used them. Among the attributes, the correlation with employed respondents was positively strong, but this also seemed to have a strong relationship with the age of the respondents. Whether or not they were students was generally not significant, which suggested that the electronic libraries were highly used exclusively by those who were employed. The use of electronic libraries among child caregivers was not as significant as the use of physical libraries, other than caregivers of toddlers. This may reflect the fact that many parents still use physical books, although many picture books and other children’s books for reading with their children have been published in Japan.

The relationship with reading habits was strongly significant with the group that used electronic media, but the correlation with the group that used paper media was limited, judging from the significance of the dummy variable. This result suggested that as the use of electronic media becomes more widespread in Japan in the future, digital libraries will be used more and more.

The correlation of the use of electronic libraries with the distance from the residence to the nearest library was not significant, though the correlation of the use of physical libraries was significant, as described above. As noted above, registration at the physical library is required to use the electronic library, but proximity to the library is not related to use of the electronic library. This may have implications for physical library placement and library services in remote areas. On the other hand, both driving on one’s own and riding in the family car were significant for car use, almost as significant as for the physical library. It was inferred that this result was related to the need to go to the physical library to register once a year.

Alertness against COVID-19 was no longer significant, including general alertness and against visiting crowded facilities, which were strongly significant for physical libraries. The use of electronic libraries has proven to be an effective service as an alternative to physical libraries, as it involves little or no contact.

For the questions related to electronic libraries, which asked only about physical library users, whether or not they lived in a local municipality that offers electronic libraries had a strong and significant effect on their use of electronic libraries. Electronic libraries are also offered in Hokkaido prefecture, as mentioned above, and all respondents to this survey were given the opportunity to use electronic libraries, more or less. Commuters to and from the basic municipality that provides electronic library services can also use that municipality’s electronic library. Actually, however, the result was that those who lived in basic municipalities that offer digital libraries were more likely to use them. The reason for this may be that there are more opportunities to use the electronic libraries of both regional and basic municipalities.

In relation to habits of using electronic gadgets, each gadget was generally positively and significantly related to electronic library use, excluding smartphones. Smartphone users may use a lot of electronic media provided by non-public electronic libraries, such as private services.

4.3. Verification of the Effect of Variables by Logistic Regression Analysis

Next, using the survey described above, simple logistic regression models assuming linearity and estimating parameters by the maximum likelihood method were constructed for the usage of both physical and electronic libraries, using various attributes of the respondents described above as the explanatory variables to determine the relationship between each attribute and library use.

The model of electronic library use was constructed, based on library users (N = 1052) instead of the total sample (N = 5357), because those who used electronic libraries visited physical libraries at least once a year for registration.

SPSS version 27.0.1.0 was used as the software to construct the model. For both logistic regression analyses, the forward selection method with the Wald test was applied, and if a variable was selected that was not significant at the 5% level, a model with only the excluded variable was again constructed using the forced-entry method. For all sub-categories that had both dummy and numeric variables, dummy variables were used in the models. The VIF (Variance Inflation Factor) was used to check multicollinearity, and those with a VIF higher than 5 were removed. Indeed, multicollinearity was confirmed for several indicators, and variables were removed.

The results of the logistic regression analysis constructed for physical library use are shown in

Table 4. Logistic regression analysis for physical library use.

Attributes		Odds Ratio	p Value	95% Conf. Interval		VIF
Categories	Sub-Categories (Variables)			Lower	Higher
1. Basic Attributes
gender	male dummy	0.669	0.000	0.564	0.794	1.419
age	50s dummy	0.785	0.008	0.656	0.938	1.197
	60s dummy	0.783	0.018	0.640	0.959	1.275
occupation	employed dummy	0.846	0.034	0.725	0.988	1.149
	university and post graduate dummy	1.644	0.000	1.410	1.917	1.168
children	toddlers dummy	1.659	0.001	1.238	2.223	1.154
	lower grades of elementary school dummy	1.532	0.004	1.144	2.051	1.212
	higher grades of elementary school dummy	1.425	0.017	1.065	1.906	1.127
2. Reading Habit Attributes
paper media	books	5.486	0.000	4.631	6.499	1.358
	newspaper and journals	1.305	0.002	1.106	1.541	1.451
electronic media	newspaper and journals	1.368	0.000	1.164	1.608	1.259
	comics	1.321	0.001	1.126	1.550	1.142
3. Accessibility Attributes
nearest library	distance to the nearest library	0.900	0.009	0.831	0.974	1.016
driving habits	drive on their own	1.305	0.001	1.113	1.530	1.225
	ride in the family car	1.274	0.002	1.096	1.480	1.170
4. Corona Alert Attributes
	visit to crowded facilities	0.845	0.000	0.794	0.899	1.020
Constant		0.126	0.000

Note: “Conf. Interval”: Confidential Interval.

. The overall percentage of correct answers for the model was 78.2%, but for users it was 27.2%.

The first attribute that resulted most strongly was the attribute of reading books in paper media, among the reading habits attribute. Both the significance and coefficient were high, simply indicating that the reading habit of paper books had a strong influence on the use of physical libraries. The reading habits of other media were also recognized as having some influence on physical library use.

Having children also had a significant impact on utilization. The 30s age group, which was significant in the χ-square test and correlation analysis, was not represented here, but this group was regarded in the model to be the child-rearing generation that raises infants to elementary school students. The attribute employed, which was not significant in the χ-square test and correlation analysis, was negative and significant in this model. This may correspond to the fact that housewives and students were significant in the χ-square test and correlation analysis. Men and those in their 50s and 60s were negative and significant. This was consistent with the results of the χ-square test and correlation analysis described above.

For accessibility, distance to the nearest library was negatively significant, while car use was positively significant. In this model, an increase of one kilometer in distance to the nearest library resulted in a 10% decrease in utilization. Apart from basic attributes, accessibility was found to affect physical library use.

For alertness against COVID-19, the multicollinearity considerations resulted in a decrease to “visit to crowded facilities”, and it had a certain impact. While it was highly significant, its coefficient was not as large as the coefficients of the other variables. The results showed that the impact of alertness against COVID-19 on physical library use was not as great as that of other attributes.

Next,

Table 5. Logistic regression analysis for electronic library use.

Attributes		Odds Ratio	p Value	95% Conf. Interval		VIF
Categories	Sub-Categories (Variables)			Lower	Higher
1. Basic Attributes
age	50s dummy	0.577	0.016	0.369	0.902	1.048
occupation	employed dummy	1.685	0.003	1.201	2.363	1.057
children	toddlers dummy	2.134	0.003	1.293	3.520	1.028
2. Reading Habit Attributes
	electronic media: books	3.714	0.000	2.632	5.240	1.137
3. Accessibility Attributes
	living in the same municipality as the nearest library	0.288	0.027	0.095	0.871	1.018
4. Corona Alert Attributes
	(no significant variable selected)
5. Accessibility Attributes to Electronic libraries
	living in the municipality providing electronic library	2.979	0.000	1.794	4.946	1.022
6. Electronic Device Use Attributes
	cell phones	1.692	0.008	1.146	2.499	1.124
	home game consoles	1.650	0.017	1.094	2.489	1.153
Constant		0.132	0.000

Note: “Conf. Interval”: Confidential Interval.

shows the results of the logistic regression analysis conducted for the use of the electronic libraries. As noted above, this is a result among only physical library users. The overall percentage of correct answers for the model was 82.2%, but for users, the percentage was 29.2%.

The results showed that those who had the habit of reading electronic books used electronic libraries more often, as expected. A strong tendency for those who lived in municipalities that offer electronic libraries to use them could also be seen.

Regarding the proximity to the nearest libraries, those who lived in the places whose nearest libraries were in the same municipalities had a significant and highly negative coefficient effect on electronic library use. The distance to the nearest library, which was significant in the model of physical library use, was, however, no longer significant in the model. These facts were interpreted to mean that, unlike physical libraries, the physical proximity did not affect the degree of use of electronic libraries

Regarding electronic devices, habits of using cell phones and home game consoles were significantly positive, which followed the results of Table 2 (the table for the χ-square test). Though the electronic media was assumed to be read primarily by tablet devices, smartphones, or computers, they were not extracted as significant variables in the model of electronic libraries. The relationship with the habits of using electronic gadgets was not simple, since there are a great number of media read by electronic gadgets, including not only media provided by electronic libraries but also electronic books and websites provided by various private sectors, in general. Thus, the relationship between the use of electronic libraries and habits of using electronic gadgets was not simple.

In the basic attributes, caregivers with toddlers, who were more likely to use physical libraries, also used electronic libraries more. On the other hand, those in their 50s who showed a significantly lower use of physical libraries also demonstrated a significantly lower use of electronic libraries. Employed respondents showed a significantly lower use of physical libraries, but a significantly higher use of electronic libraries. Together with the results of the χ-square test, this means that the physical and electronic libraries encouraged the use of completely different hierarchies.

5. Discussion

5.1. Findings

This paper analyzed the attributes of citizens that influence the use of library services and digitized electronic libraries by simple statistical approaches. In particular, I compared physical and electronic libraries, with a view to the possibility of providing access to digitalized library services to different groups of the population than has been provided by physical public facilities in the traditional way.

As a whole, electronic libraries provide their services to some different groups than physical libraries. The use of electronic libraries was somewhat less correlated with having children than the use of physical libraries. In addition, electronic libraries provide services toward the working population, which means the service reaches more taxpayers who have contributed to the funding of public facilities but have not used them.

The influence of age was not simple for physical and electronic library use, although there was certainly a trend toward higher use among younger age groups. As for the electronic libraries, the digital divide among the elderly was not clearly observed. It was inferred that the digitalization of library services can be used by many citizens, including the elderly, based on the results of the analysis, although further research is needed to determine the impact of the digital divide.

Regarding spatial accessibility, while the distance to the nearest library was inversely related to the percentage of users for physical libraries, this relationship was not observed for electronic libraries. Thus, even if some libraries were to close and the number of physical library locations was reduced, in accordance with the population decline, the wide spread of electronic libraries may help mitigate the decline in users. In the reorganization of public facilities in a municipality, some of the services of physical libraries could be considered for replacement by electronic libraries. As for the use of library services by the elderly, whose physical accessibility is generally limited, substitution through digitization of library services is likely to be effective if the impact of the digital divide is small.

As attributes analyzed only for the use of electronic libraries, living in a municipality that offers electronic libraries encouraged their use, so the introduction of electronic libraries in each municipality will have the effect of increasing the number of users. Citizens’ use of various electronic devices was also generally correlated with whether they used the electronic libraries. As the use of electronic devices becomes even more common in the future, the use of electronic libraries will further increase. These findings revealed that library services can expand their use to a more diverse range of users through the digitization of library services. Since the digital divide due to age and spatial accessibility was not significant in digital libraries, it is expected to improve accessibility for the elderly and residents on the peripheral areas. As the restructuring of public facilities continues along with the decline in the population, the provision of public services through digitization, including electronic libraries, has shown the potential to be an alternative.

As noted above, it is difficult to compare the results of this study with those of existing studies because there are no other similar studies, but when contrasted with the increase in access to digital content during the period of the coronavirus pandemic shown by Ćirić and Ćirić [30], this study found that COVID-19 alert attributes and the use of electronic libraries had no correlation. However, this issue needs to be scrutinized in the future, since these studies used different variables.

5.2. Limitation

This study is unique in that a questionnaire survey was conducted with a relatively large sample, details about the attributes of the users were collected, and these were linked to their usage status. However, the study did not take into account the attributes of each physical or electronic library and their differences. In this survey, even the most heavily used libraries had about 200 samples of users, and many libraries had only 10 samples or less. Thus, it was difficult to create a model that took library quality into account.

In reality, for example, it is assumed that the tendency of usage differs to some extent between the central libraries and branch libraries. There may also be differences between libraries and electronic libraries, as well as between individual libraries and electronic libraries, in terms of the quantity and quality of their collections. Other factors, such as ease of use of electronic books operated by different software and user satisfaction with the service, will also have a certain impact, and these too will be affected differently depending on the attributes. Furthermore, since this study used an Internet survey, the population that does not use the Internet was not included in the survey.

In this study, it was not possible to obtain a dataset large enough to include these factors and characteristics in the model. The methodology of this study seemed to make further elaboration difficult due to the method of collecting responses, considering the general public, the number of samples that could be collected through that method, and considerations of personal information. Thus, more sophisticated analysis should be conducted by also combining other data in the future.

6. Conclusions

This study analyzed the differences in use by citizens by various attributes between the traditional method of providing public services through public facilities and the new digitized method, using libraries as an example. As indicated in the previous section, considering the differences in the attributes of the users of the two methods, providing public services through digitization in combination with traditional public services would increase the benefits to a wider range of citizens. This is likely to have a strong bearing on the financial resources that cover the cost of public services and on the decision-making process to finance such services amid population decline and fiscal pressures.