Sage Journals: Discover world-class research

Abstract

This paper explores the exact relationship between the development level of American public libraries and the output of science and engineering publications. We use the sample data from 1996 to 2019 to build a time series model, and analyze the long-term equilibrium, short-term correlation and Granger causality relationships between them with stationarity test, cointegration test, Granger causality test, and impulse response function. The results suggest that there are a long-term equilibrium relationship and a bidirectional Granger causality between them. The two interact and promote each other. The impact of the development level of American public libraries on the output of science and engineering publications can be seen in fewer time periods, while the impact of the output of science and engineering publications on the development level of American public libraries take more time periods to show.

Keywords

Cointegration test Granger causality impulse response function public libraries science and engineering publications

Introduction

With the development of science and technology and the progress of society, public libraries are playing an increasingly important role. People use library resources to create new knowledge through learning, independent thinking and scientific experiments, etc., and produce scientific papers and patents and other products. These add new knowledge content for library resources. On the other hand, the output of these scientific researches also provides an impetus for the progress of society. Science and engineering publications are the carriers that reflect people’s scientific and technological activities, provide a bridge for scientific and technological workers to communicate, and are also the link for the transformation of science and technology into productive forces. The primary method of disseminating research findings is through publication of conference proceedings and peer-reviewed journal articles (i.e. publication output). Publication output reached 2.9 million articles in 2020, based on data from the Scopus database of S&E publications (White, 2021). Data on publication output indicate an increase in global research activity, a growth in the involvement and scientific capabilities of middle-income countries, and an internationally connected research ecosystem (White, 2021). Since 2000, 215 of the 216 Nobel Laureates in Science and Economics have published their research in Elsevier’s journals (Elsevier, 2021). Several examples are given below. Goodenough (2007, 2015), Goodenough and Kim (2011), (Goodenough and Huang, 2007), Whittingham (1979, 2000), Whittingham et al. (1995), Whittingham and Gamble (1975), and Yoshino (2014) were jointly awarded the 2019 Nobel Prize in Chemistry for “the development of lithium-ion batteries.” Lithium-ion batteries are found everywhere, from handheld devices to electric vehicles to solar power plants. James P. Allison and Tasuku Honjo shared the 2018 Nobel Prize in Physiology or Medicine “for their discovery of cancer therapy by inhibition of negative immune regulation” (Allison, 2015; Egen and Allison, 2002; Iwasato et al., 1990; Nishimura and Honjo, 2001; Okazaki and Honjo, 2005, 2006; Shimizu and Honjo, 1984). The therapy, now called “immune checkpoint therapy,” has fundamentally changed the prognosis and survival of patients with certain types of cancer. The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2019 was awarded jointly to Banerjee (2004), Banerjee et al. (2018), Chamon and Kremer (2009), Duflo (2004), and Duflo et al. (2015) “for their experimental approach to alleviating global poverty” . The research has a positive impact on policy-making to alleviate poverty. These scientific achievements have had a great impact on the world. It can be said that the improvement of people’s living standards is increasingly dependent on scientific research and technological innovation. It was estimated that 116,462, 216,657, and 764,553 papers were cited in government, news, and patent documents, respectively between 2005 and 2014 (Yin et al., 2021). This suggests that scientific publications are not merely an exercise of ivory tower academics but serve as a key linkage enabling public uses of scientific output. The output of science and engineering publications is the products of scientific and technological research activities in which people use resources of public libraries to create new knowledge. Naturally, it is believed that there is a certain connection between the development level of public libraries and the output of S&E publications. Their exact relationship is unclear and requires in-depth discussion and empirical study.

Literature review

Public libraries, as institutions meeting reading of users, play an important role in promoting the development at various levels of society such as community life (Lipkova et al., 2019), agricultural production (Ugwoke, 2013), innovation in science and technology (Cai, 2013), economic growth (Barron et al., 2005; Gao and Pan, 2013) and so on. Librarians and information workers offered agricultural information to farmers in Nigeria to promote boosting food and cash crops production (Ugwoke, 2013). Barron et al. (2005) conducted a study on the impact of public libraries in South Carolina on the local economy, and concluded that for every $1 spent by state and local government on SC public libraries, the return on investment was $2.86. Morris et al. (2006) studied economic sense of audiovisual materials in UK public libraries. A cost benefit of 1:1.34 using the PVB (present value benefits) based on maximum loan charges was found, meaning that the UK got £1.34 direct benefit from every £1.00 spent on the audiovisual service. Linhartová and Stejskal (2017) analyzed data from 37 public libraries in the Czech Republic and found that the public libraries’ benefit/cost ratio was 5.86%–6.17%, which meant that every $1.00 spent on the Czech public libraries, provided on average $6.00 benefit value to taxpayers. Analyzing data from 1979 to 2008 in China, Gao and Pan (2013) found that there was apparent two-way causality between the development of public libraries and economic growth. Cai (2013) verified that public libraries were correlative with scientific and technological innovation through correlation analysis and concluded that if the total expenditure of public libraries increased by 1%, then the growth rate of transaction volume in national technology market would be 1.34%.

Inglesi-Lotz et al. (2014) investigated the relationship between knowledge (research output) and economic growth in US over 1981–2011. Using a two-state regime switching vector smooth autoregressive model, the authors found that there was a unidirectional Granger causality from research output to GDP in the sample. Using annual data for the number of nation publications in SCI and SSCI indexed journals and annual GDP data for the period 1981–2007, Lee et al. (2011) examined the correlation between GDP and research output for 24 countries and a region. The empirical findings suggested that there was a mutual causality between research output and economic growth in Asia, whereas in Western countries the causality was much less clear. Solarin and Yen (2016) examined the effect of research output on economic growth in 169 countries for the period, 1996–2013. The results indicated that research output had positive impact on economic growth, irrespective of whether the sample was for developing or developed countries. Using the extended Cobb-Douglas model with capital per worker and the quantity of scientific and technical journal articles (research publications) per worker, Kumar et al. (2016) explored the link between research articles and economic growth in China and US over the sample period, 1981–2012. The authors found that there was a bidirectional causality between research publications per worker and output per worker for China, and there was a unidirectional causality from output per worker to research publications per worker for USA.

In the existing literature, it has not been found to study the relationship between the development level of American public libraries and the output of S&E publications with empirical analysis. This paper builds a time series model to investigate the relationship between them. Introducing stationarity test, cointegration test, Granger causality test, and impulse response function examines the relationship between them. Compared to Inglesi-Lotz et al. (2014), Gao and Pan (2013), Lee et al. (2011), and Kumar et al. (2016), there are the following differences. First, the research subjects are different. Second, modeling is more difficult due to the complexity of a study object in this paper. Since there are many indexes for evaluating the development level of public libraries, and there are cross-coupling between these indexes, how to effectively extract a comprehensive index that truly reflects their development level is a difficult problem for us, whereas they are not like that, and their objects themselves come with an index, respectively. Third, unlike them, we also use impulse response function to analyze the relationship between variables.

Model construction and data analysis

Construction of evaluation index system and data sources

A time series model is divided into an explanatory variable and an explained variable. Suppose that the output of S&E publications is an explained variable, and the development level of public libraries in the United States an explanatory variable. The data of evaluation indexes of the output of S&E publications are derived from “Publications Output: U.S. Trends and International Comparisons” (White, 2021) published by the National Center for Science and Engineering Statistics (NCSES). The time span of the indexes is 1996–2019. Under the direction of the National Science Council, the data are compiled by NCSES, which is a principal federal statistical agency subordinate to the National Science Foundation (NSF) and Bureau of Social, Behavioral, and Economic Sciences. Index reports were extensively reviewed by internal and external subject matter experts, federal agencies, members of the National Science Council, and NCSES statistical reviewers to ensure data accuracy. A more scientific and accurate score counting method is adopted, that is, the publication counts are allocated according to the proportion of co-authors for each country or institution mentioned in the articles.

For the construction of the evaluation index system of the development level of American public libraries, considering that the factors of the development of American public libraries are relatively complex, it is difficult to completely reflect their comprehensive development by only selecting a single index. According to relevant theories, the index system is divided into four dimensions: infrastructure, literature resources, user services, operating revenues and expenditures. Then the four dimensions are decomposed into 14 secondary indexes, as shown in Table 1, in an effort to completely reflect the development level of public libraries in the United States. The data of evaluation indexes of the development level of American public libraries come from the Public Libraries Survey (PLS) published annually by the Institute of Museum and Library Services (IMLS, 2021). The data are collected from approximately 9000 public libraries with approximately 17,000 individual public library outlets in the 50 states, the District of Columbia, and outlying territories. To ensure the consistency of the data time span, the time span was also chosen as 1996–2019. In addition, in order to eliminate the impact of price changes, the GDP deflator index with 1996 as the base period was used to deflate the above-mentioned indexes of the design amount. The deflator index comes from the data released by the Bureau of Economic Analysis of the US Department of Commerce.

Table 1.

The comprehensive evaluation index system of public libraries development level in the United States.

Development Level of American Public Libraries	Infrastructure	×₁	Number of public libraries
		×₂	Number of stationary outlets
		×₃	Bookmobiles
	Literature resources	×₄	Print materials (in thousands)
		×₅	Current print serial subscriptions (in thousands)
		×₆	Audio (in thousands)
		×₇	Video (in thousands)
	User services	×₈	Library visits (in thousands)
		×₉	Total circulation (electronic & physical, in thousands)
		×₁₀	Interlibrary loans provided to (in thousands)
		×₁₁	Interlibrary loans received from (in thousands)
		×₁₂	Paid FTE staff
	Operating revenues and expenditures	×₁₃	Total operating revenues (in thousands)
	Operating revenues and expenditures	×₁₄	Total operating expenditures (in thousands)

Analysis of development level indexes of public libraries in the United States

Since there are many indexes to measure the development level of public libraries, if all are copied without any treatment, it is not suitable for subsequent analysis and testing. In addition, there may be multicollinearity problems between variables, which cause the information reflected in the index data to overlap each other, and obscure the true characteristics of the things to be reflected. Therefore, using validity analysis and factor analysis, the dimensionality of the indexes can be reduced as much as possible without losing information, that is, 14 indexes representing the development level of public libraries are reflected through a few indexes.

Standardization

By sorting out the data of indexes, it is found that values between them vary greatly. If the original data are directly used for analysis, the role of the index with higher value in the comprehensive analysis will be highlighted, the effect of the index with lower value will be relatively weakened. For this, it is necessary to standardize the original data. In this paper, $Z$ -score standardized method is used to process all data.

The formula for $Z$ -score standardization is

Z_{i} = \frac{X_{i} - \bar{X}}{σ}, σ = \sqrt{\frac{\sum {(Χ - \bar{Χ})}^{2}}{n}}

(1)

where $Z_{i}$ is the standardized value of $X_{i}$ and $σ$ is the standard deviation.

Validity analysis

Validity analysis is used to examine whether the design of data of indexes is reasonable by KMO and Bartlett sphericity test. After testing, the KMO value is 0.758 > 0.7, and the observation value of statistic of Bartlett sphericity test is 956.184, and the corresponding probability ρ value is 0 < 0.01. The test results are shown in Table 2. It indicates that the 14 index variables are not completely independent, have a strong correlation, and are suitable for factor analysis.

Table 2.

KMO and Bartlett’s test.

Kaiser-Meyer-Olkin		0.758
Bartlett’s sphericity test	Approximate chi-square	956.184
	df	91
	Sig.	0

Factor analysis

The validity analysis implies that the design of data of indexes is reasonable, and now it is verified by factor analysis. Table 3 shows the results of factor contribution rates. The eigenvalues of the first two factors are greater than 1, and the explained cumulative variance contribution rate is 91.218%, which can reflect the overall level of the development of public libraries in the United States. Hence, it is appropriate to extract the first two components as common factors.

Table 3.

Total variance explained by each index.

Component	Eigenvalue	Variance contribution rate %	Cumulative variance contribution rate %
1	10.364	74.03	74.03
2	2.406	17.189	91.218
3	0.883	6.309	97.527
4	0.192	1.373	98.9
5	0.07	0.498	99.398
6	0.049	0.348	99.747
7	0.013	0.095	99.842
8	0.01	0.07	99.912
9	0.005	0.037	99.95
10	0.003	0.024	99.973
11	0.002	0.012	99.986
12	0.001	0.009	99.995
13	0.001	0.005	100
14	3.766E−5	0.000	100

In order to better explain the extracted common factors, the maximum variance method is used to perform orthogonal rotation for the factors. The variance contribution rates of the two common factors after the rotation are 68.148% and 23.07%, respectively. It can be found from the component matrix after the rotation that the common factor F₁ has a strong correlation with literature resources, user services, operating revenues and expenditures. It can be interpreted as a subjective factor in the development of public libraries. The common factor F₂ has a strong correlation with infrastructure, and can be interpreted as an objective factor in the development of public libraries. The contribution rates show that the current development of public libraries in the United States is mainly pushed by the subjective factor, and the objective factor cannot be a stable factor to promote the development of public libraries due to its large fluctuation.

Comprehensive evaluation score for the development of public libraries in the United States

On the basis of extracting the two common factors, the component coefficient matrix is obtained, and the estimated values of the common factors are calculated. The calculation formula is:

F_{i} = \sum_{j = 1}^{n} a_{i j} * X_{j}

(2)

where, $a_{i j}$ is the score coefficient, $i$ = 1, 2, $n$ = 1, 2, . . ., 24.

After the values of the two common factors $F_{1}$ and $F_{2}$ are obtained, they are weighted with the variance contribution rate of each factor after rotation as the weight to obtain the comprehensive score F of the development level of public libraries in the United States, as shown in Figure 1.

Figure 1.

Comprehensive scores of the development level of public libraries in the United States.

It can be seen from Figure 1 that for the period 1996–2009, the comprehensive evaluation scores of the development of public libraries in the United States showed an increasing trend year by year. In 2010, there was a relatively obvious decline. After that, although there were occasional fluctuations, the whole trend was relatively stable. In the last few years, there was a slow upward trend again. By consulting relevant reports and data, it is found that in 2010, compared with 2009, operating revenues and expenditures, service hours and staffing decreased. Operating revenues and expenditures fell for the first time since 1996. Meanwhile, the number of visits to libraries declined due to the increase in online services. In addition, budget cuts led to a reduction in literature resources. Affected by the economic recession, the staff of public libraries decreased by 3.9%. These comprehensive factors had a great impact on the development of public libraries. On the whole, the development of public libraries in the United States has shown an upward trend since 1996.

Model construction

According to the above selected explained variable and explanatory variable, the time series model of S&E publications output (Y) and the development level of public libraries in the United States (X) is constructed as follows:

Y_{t} = C + β X_{t} + μ_{t}

(3)

where $C$ and $β$ are unknown parameters, $µ_{t}$ is an unknown residual time series.

Whether the model is correct depends on the stability of the time series to a large extent. If the time series variables are non-stationary, spurious regression will occur. In this paper, stationarity test, cointegration test, Granger causality test, and impulse response function are used to investigate the possible relationships between variables.

Specific steps are as follows. First, the Augmented Dick-Fuller (ADF) unit root test (Dickey and Fuller, 1979) is performed on the time series data to examine the stationarity of the variables. If the variables are same-order difference stationary series, the Engle-Granger two-step procedure (E-G) for cointegration is used to test the unit root of the residuals of the regression equation to investigate whether there is a long-term cointegration relationship between the variables. The Granger causality test is used to examine whether there is a causal effect between the variables. Finally, we investigate the dynamic change relationship between variables with the impulse response function to further explore the influence between the development level of public library and the output S&E publications.

Tests and analyses

Stationarity test

Figure 2 shows a publishing trend of S&E publications. Although there are occasional fluctuations in the number of publications, the overall trend is on the rise. It is determined that it is not a stationary time series. It can be judged from Figure 1 that the development level of American public libraries is not a standard stationary time series, either. To avoid spurious regressions in the regression equation, the stationarity of the two time series are tested by the ADF unit root test. According to modified Schwarz information criterion, Akaike info criterion (AIC), and Hannan-Quinn criterion (HQ), lag orders are determined, and the test results are shown in Table 4.

Figure 2.

Publishing trend of S&E publications.

Table 4.

Stationarity test.

Variables	Test form	ADF test t-statistic	5% critical level	Prob.	Result
X	(C, N, 1)	−2.615335	−3.004861	0.1050	Non-stationary
Y	(N, N, 1)	1.352952	−1.957204	0.9507	Non-stationary
DX	(C, T, 0)	−3.718607	−3.632896	0.0424	Stationary**
DY	(N, N, 0)	−2.159870	−1.957204	0.0325	Stationary**

DX, DY represent the first-order difference of X, Y.

Represents rejection of the original hypothesis at 5% critical level.

The ADF t-statistics of the two variables X and Y are both greater than test values at 5% critical level, so both of them are non-stationary time sequences. We make first-order difference for X and Y respectively to obtain DX and DY, both of which have t-statistics less than test values at 5% critical level. Thus, the original hypothesis is rejected. DX and DY are stationary time series. Both variables X and Y are first-order difference stationary series, which satisfy the condition of cointegration test.

Cointegration test

For two time series that are non-stationary but same-order difference stationary series, their linear combination may be stationary, the Engle-Granger (E-G) two-step procedure for cointegration is used to determine whether there is a long-term stable equilibrium relationship between the two series. Since X and Y are both first-order difference stationary series, the ordinary least squares (OLS) method is used to regress X and Y. The regression results are as follows:

\begin{array}{l} Y_{t} = 378300.3 + 62160.56 X_{t} \\ (98.84662) (12.54966) \end{array}

$R^{2}$ = 0.877433, $D W$ = 0.368909, $F$ = 157.4939

where, the numbers in parentheses are the values of the corresponding t-statistics, $R$ -squared is coefficient of determination, $D W$ is Durbin-Watson statistic, and $F$ is $F$ -statistic.

The goodness of fit $R$ -squared of the regression model is 0.877433, the probability of $F$ -statistic test is 0.0000, which indicates that the equation is overall highly significant. The probabilities of t-statistic tests for the constant term and the explanatory variable coefficient are both 0.0000 < 0.01, which implies to pass the significance test of variables. Hence, there is a linear relationship between the development level of American public libraries and the output of S&E publications.

According to the Durbin Watson ( $D W$ ) test table, it is obtained that at the significance level of 5%, for the regression model, $D W$ = 0.37 < d_L = 1.273, implies that there is a first-order positive autocorrelation in the model. The Lagrange Multiplier (LM) test is used for its high-order autocorrelation test. The results show that LM (1) = 12.83277, probability is 0.0003, LM (2) = 14.36960, probability is 0.0008, that is, there is a high-order serial autocorrelation in the model. Considering the AIC, SC, and HQ information criteria, and the significance of the estimated values RESID (−3), RESID (−2), RESID (−1) of the corresponding parameters of the lag terms, it is reasonable to choose the second-order lag. Consequently, there is a second-order positive autocorrelation in the disturbance term.

In order to overcome its autocorrelation, on the basis of the regression model, the Conditional Least Squares (CLS) and Cochrane-Orcutt iterative method are used for correction, and the results are as follows:

\begin{array}{l} Y_{t} = 375026.3 + 69305.23 X_{t} + 1.091877 μ_{t - 1} - 0.478387 μ_{t - 2} \\ (66.76856) (7.131352) (5.025351) (- 2.372172) \end{array}

$R^{2}$ = 0.970218, $D W$ = 1.995553, $F$ = 195.4627

The goodness of fit of the revised model is increased to 0.97, indicating that the fitting degree of the model to the sample data reaches 97%. After looking up the table, we obtain $d_{U}$ = 1.446. DW satisfies $d_{U}$ < DW < 4− $d_{U}$ , that is, 1.446 < DW = 1.995553 < 2.554, indicating that there is no first-order serial autocorrelation. For the model, a high-order autocorrelation test is carried out by LM test. The results show that LM (1) = 0.003188, probability is 0.9550, LM (2) = 0.014910, probability is 0.9926. Both probabilities are greater than 0.05, and the original hypothesis is accepted, implying that there is no high-order serial autocorrelation in the model. The model is tested with Ramsey RESET. Each added variable was generated by the power of the fitted value of the original regression quantity. The probabilities of the statistics are 0.3436 and 0.6404, which are both greater than 0.05. It implies that the original hypothesis is accepted, and there is no setting error in the model.

We make the stationarity test for the residual series of the revised model and the results are shown in Table 5. It should be noted that the ADF test here is for the residual term calculated from the regression equation. It is needed to calculate the corresponding critical level. The formula is as follows:

C (α) = ϕ_{\infty} + ϕ_{1} T^{- 1} + ϕ_{2} T^{- 2} + \dots + ϕ_{n} T^{- n}

(4)

where $T$ is the sample size, n is the number of variables, and α is critical level.

Table 5.

Stationarity test for residual.

	Test form	ADF test t-statistic	5% critical level	1% critical level	Result
Residual	(N, N, 0)	−4.462452	−3.601915	−4.391152	Stationary***

***

Represents rejection of original hypothesis at 1% critical level.

$T$ The residual series is stationary at 1% critical level. It implies there is a cointegration relationship between X and Y, that is, there is a long-term equilibrium relationship between the development level of American public libraries and the output of S&E publications. Since in the cointegration equation, the coefficient of the explanatory variable X is a positive number, it shows that the improvement of the development level of public libraries can promote the output of S&E publications in the United States. Specifically, the output of S&E papers increases by 69,305.23 papers for increasing one unit of development level of American public libraries.

Granger causality test

From the cointegration test, it can be found that the development level of American public libraries has a positive impact on the output of S&E publications. However, although the cointegration test shows that there is a long-term stable relationship between them, it is only a numerical conclusion, and the causal relationship between them needs to be further determined. This paper constructs a Granger causality test to investigate the sequence of influence between variables. The principle of the test is that if a change in X causes a change in Y, then the change in X should occur before the change in Y. Table 6 shows the results of the Granger causality test.

Table 6.

Granger causality test.

Original hypothesis	Lag	1	2	3	4	5	6	7
X is not the Granger cause for Y	F-statistic	13.2691	7.50843	2.957	1.26006	1.0933	0.36939	0.18347
X is not the Granger cause for Y	Prob.	0.0016***	0.0046***	0.0688*	0.3427	0.4325	0.8713	0.9626
Y is not the Granger cause for X	F-statistic	0.8541	0.64304	1.01689	1.45346	2.93897	5.75309	10.5858
	Prob.	0.3664	0.538	0.4147	0.2813	0.085*	0.0371**	0.0890*

, **, ***Represents rejection of original hypothesis at 10%, 5%,and 1% critical level respectively.

Table 6 shows that there is a one-way Granger causality from the development level of American public libraries (X) to the output of S&E publications (Y) in lag period 1, lag period 2, and lag period 3. It indicates that the improvement of the development level of American public libraries can promote the output of S&E publications within 1–3 years of lag. That is to say, American public libraries can comprehensively promote the output of S&E publications through the development of the four aspects in infrastructure, literature resources, user services, operating revenues and expenditures. The improvement of the development level of public libraries implies that the libraries better meet the needs of users, prompt more people to pay attention to books, periodicals, magazines, newspapers, electronic publications, etc., in the meanwhile, improve the cultural level of users, and better help users for scientific research and paper writing, thus affecting the output of the S&E publications. In lag period 5, lag period 6, and lag period 7, there is a one-way Granger causality from the output of S&E publications to the development level of American public libraries. It implies that the output of S&E publications can affect the development level of American public libraries within 5–7 years of lag. It shows that it takes a relative long time for the output of S&E publications to promote the development of public libraries. The causal relationship does not exist in lag periods 1–4.

Impulse response function analysis

Since X and Y are first-order difference stationary series, the VAR model can be used to examine the influence of a shock of random disturbance term on them. In order to better reflect the dynamic characteristics of the model, it is necessary to determine the optimal lag orders of the model before building the VAR model. According to the five information criteria, we consider the lag orders 0–3 of the model to obtain that the optimal lag orders of the model are 2, that is VAR (2). The results are shown in Table 7.

Table 7.

Lag selection of VAR model.

Lag	Log L	LR	FPE	AIC	SC	HQ
0	−248.339	NA	77,531,553	23.84181	23.94128	23.86339
1	−194.5416	92.22399	678,326.6	19.0992	19.39764	19.16397
2	−188.1557	9.730888*	548,452.0*	18.87197*	19.36937*	18.97992*
3	−186.7833	1.829955	728,995.5	19.12222	19.81856	19.27334

Indicates lag order selected by the criterion.

The stationarity of the VAR model is tested and the results are shown in Figure 3. The reciprocals of the eigenvalues of the VAR model are all located within the unit circle, indicating that the established model is stable. Considering that the VAR model is a non-theoretical model, its independent variable coefficients have little practical significance and its model is not explained in detail here. Instead, an impulse response function is used to analyze the relationship between variables. In order to fully show the impulse responses for the development level of American public libraries and the output of S&E publications, this paper make 30-periods impulse response function graphs (see Figures 4 and 5). The horizontal axis is the time period, and the vertical axis is the response of the variable to a shock.

Figure 3.

VAR model stationarity test.

Figure 4.

Shock influence on the output of S&E publications.

Figure 5.

Shock influence on the development level of American public libraries.

In Figure 4, the output of S&E publications is positively influenced in period 1 after receiving its own shock, and the positive effect is relatively obvious in former periods. This shows that the output of S&E publications is significantly affected by itself in former periods, and there is certain path-dependent characteristics. After the output of S&E publications receives a shock from the development level of American public libraries, there is a small negative effect (−332) in the second period, at third period it begins to rapidly rise to a positive effect (1882), then exceeds the effect of S&E publications on itself, in the third period, goes up to a positive effect (3564), reaches a maximum (6586) in sixth period, then gradually declines, and finally becomes stable. The above shows that the shock of the development level of American public libraries on the output of S&E publications has a certain lag. The deficiencies of American public libraries in the quality of computer and internet infrastructure (Noh, 2019), the application of management innovation (Potnis et al., 2021), and the uneven development level (Mehra et al., 2017) restricted the output of S&E publications in the initial period. With the gradual emergence of the information aggregation effect of American public libraries, the supporting effect on the output of S&E publications begins to appear. This influence is continuous, accumulates during periods 2–3, after the third period, exceeds the influence of S&E publications on itself and becomes the main influencing factor.

In Figure 5, the development level of American public libraries is positively influenced in the first period after receiving its own shock, and the positive effect reaches a peak in the second period. Since then, although the impact continues to fall, the positive impact has always dominated. After the development level of American public libraries receives a shock from the output of S&E publications, a positive impact of 0.003 occurs in the first period, and the impact reaches the maximum 0.032 in the third period. As time goes on, the influence gradually weakens, and becomes stable after the 20th period. It can be seen that the development level of American public libraries is greatly affected by itself. The output of S&E publications has a long-term positive impact on the development of American public libraries. With time, the impact gradually increases, it peaks in the third period, and after that gradually weakens. But on the whole, the impact value is small.

Conclusions

The relationship between the development of American public libraries and the output of S&E publications has been deeply explored through time series analysis methods such as stationarity test, cointegration test, Granger causality test, and impulse response function, and the following conclusions are obtained. First, the co-integration relationship between them is verified, that is, there is a long-term equilibrium relationship between them. The elevation of the development level of public libraries can promote the output of S&E publications in the United States. Second, there is a two-way Granger causality between them. In the lag of 1 and 3 years, the development level of American public libraries has a relatively obvious ability to predict the output of S&E publications. In the lag of 5–7 years, the output of S&E publications has a relatively obvious ability to predict the development level of American public libraries. Third, the results of impulse response function analysis are consistent with the results of Granger causality test. This also confirms the correctness of the results of Granger causality test from another perspective.

To promote the coordinated development of public libraries and other aspects of society requires joint efforts. In terms of the construction and development of public libraries, it is necessary to enhance scientific and technological service capability, improve management services, try service diversity, and expand service businesses. Librarians with professional backgrounds form information service teams, which visit enterprises, institutions and rural areas to understand their requirements and provide corresponding supports. Also, public libraries can cooperate with governments, institutions, research groups, and other organizations to undertake the services of quantitative analysis for literature resources and information, provide relevant analysis reports, which provide scientific grounds for their decision-making or other work. This not only increases the utilization of literature, but also increases the depth and breadth of services.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors received financial support from the Research Project of Library Society of China for the research of this article. (No.2022LSCKYXM-ZZ-QN010).

ORCID iD

Jiahe Jiang

Author biographies

Yangyang Jiang received the B.E. degree from Northeastern University, Shenyang, China, and the M.S. degree from Beihang University, Beijing, China, in 2014. She is currently a librarian in the Library, Beihang University. Her research interests include information science & library science and bibliometric analysis.

Jia He received the Master’s degree in communication and information systems from Beihang University in 2004. She is currently a subject librarian in Beihang University Library. Her research focuses on bibliometrics and content analysis.

Bo Jin is a lecturer in the Beijing University of Civil Engineering and Architecture. He earned the M.S. degree from the University of Southampton. He is currently a PhD candidate in the Beihang University. His research interests include Machine Learning, Prognostication and Health Management (PHM) Systems, and international education research.

Jiahe Jiang received the M.S. and Ph.D. degrees from Beihang University, Beijing, China, in 1997 and 2000, respectively. He is currently an Associate Professor in the School of Automation, Beihang University. His research interests include information science and flight control.

References

Allison

(2015) Checkpoints. Cell 162(6): 1202–1205.

Banerjee

Barnhardt

Duflo

(2018) Can iron-fortified salt control anemia? Evidence from two experiments in rural Bihar. Journal of Development Economics 133: 127–146.

Banerjee

(2004) Educational policy and the economics of the family. Journal of Development Economics 74(1): 3–32.

Barron

Williams

Bajjialy

, et al. (2005) The economic impact of public libraries on South Carolina. Available at: http://www.libsci.sc.edu/sceis/exsummary.pdf (accessed 28 July 2022).

Cai

(2013) Theoretical and empirical analysis of the influence of public library on scientific and technological innovation. BioTechnology: An Indian Journal 8(8): 1124–1129.

Chamon

Kremer

(2009) Economic transformation, population growth and the long-run world income distribution. International Economic Journal 79(1): 20–30.

Dickey

Fuller

(1979) Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association 74(366): 427–431.

Duflo

(2004) The medium run effects of educational expansion: Evidence from a large school construction program in Indonesia. Journal of Development Economics 74(1): 163–197.

Duflo

Dupas

Kremer

(2015) School governance, teacher incentives, and pupil–teacher ratios: Experimental evidence from Kenyan primary schools. Journal of Public Economics 123: 92–110.

10.

Egen

Allison

(2002) Cytotoxic T lymphocyte antigen-4 accumulation in the immunological synapse is regulated by TCR signal strength. Immunity 16(1): 23–35.

11.

Elsevier (2021) Honoring the 2021 Nobel Laureates with free access to their research. Available at: https://www.elsevier.com/connect/honoring-the-2021-nobel-laureates-with-free-access-to-their-research

12.

Gao

Pan

(2013) Empirical analysis of relationship between development of public libraries and economic growth. In: Du

(ed.) Informatics and Management Science IV. Lecture Notes in Electrical Engineering, vol. 207. London: Springer, pp.359–364.

13.

Goodenough

(2007) Cathode materials: A personal perspective. Journal of Power Sources 174(2): 996–1000.

14.

Goodenough

(2015) Energy storage materials: A perspective. Energy Storage Materials 1: 158–161.

15.

Goodenough

Huang

(2007) Alternative anode materials for solid oxide fuel cells. Journal of Power Sources 173(1): 1–10.

16.

Goodenough

Kim

(2011) Challenges for rechargeable batteries. Journal of Power Sources 196(16): 6688–6694.

17.

IMLS (2021) Public Libraries Survey. https://www.imls.gov/research-evaluation/data-collection/public-libraries-survey

18.

Inglesi-Lotz

Balcilar

Gupta

(2014) Time-varying causality between research output and economic growth in US. Scientometrics 100(1): 203–216.

19.

Iwasato

Shimizu

Honjo

, et al. (1990) Circular DNA is excised by immunoglobulin class switch recombination. Cell 62(1): 143–149.

20.

Kumar

Stauvermann

Patel

(2016) Exploring the link between research and economic growth: An empirical study of China and USA. Quality & Quantity 50(3): 1073–1091.

21.

Lee

Lin

Chuang

, et al. (2011) Research output and economic productivity: A Granger causality test. Scientometrics 89(2): 465–478.

22.

Linhartová

Stejskal

(2017) Public libraries’ services and their economic evaluation. Scientific Papers of the University of Pardubice. Series D: Faculty of Economics and Administration 24(41): 90–101.

23.

Lipkova

Diviak

Jarolimkova

, et al. (2019) Assessing libraries’ community roles. Proof of concept. Communications in Computer and Information Science 989: 83–93.

24.

Mehra

Bishop

Partee

(2017) Small business perspectives on the role of rural libraries in economic development. The Library Quarterly 87(1): 17–35.

25.

Morris

Ayre

Jones

(2006) Audiovisual materials in UK public libraries: economic sense?. Journal of Documentation 62(5): 555–569.

26.

Nishimura

Honjo

(2001) PD-1: An inhibitory immunoreceptor involved in peripheral tolerance. Trends in Immunology 22(5): 265–268.

27.

Noh

(2019) A comparative study of public libraries’ contribution to digital inclusion in Korea and the United States. Journal of Librarianship and Information Science 51(1): 59–77.

28.

Okazaki

Honjo

(2005) Pathogenic roles of cardiac autoantibodies in dilated cardiomyopathy. Trends in Molecular Medicine 11(7): 322–326.

29.

Okazaki

Honjo

(2006) The PD-1–PD-L pathway in immunological tolerance. Trends in Immunology 27(4): 195–201.

30.

Potnis

Winberry

Finn

(2021) Best practices for managing innovations in public libraries in the USA. Journal of Librarianship and Information Science 53(3): 431–443.

31.

Shimizu

Honjo

(1984) Immunoglobulin class switching. Cell 36(4): 801–803.

32.

Solarin

Yen

(2016) A global analysis of the impact of research output on economic growth. Scientometrics 108(2): 855–874.

33.

Ugwoke

(2013) Promoting Nigerian agriculture through library and information services. International Journal of Information Management 33(3): 564–566.

34.

White

(2021) Publications output: U.S. Trends and International Comparisons. Available at: https://ncses.nsf.gov/pubs/nsb20214/executive-summary

35.

Whittingham

(2000) Insertion electrodes as SMART materials: The first 25 years and future promises. Solid State Ionics 134(1–2): 169–178.

36.

Whittingham

(1979) Intercalation chemistry and energy storage. Journal of Solid State Chemistry 29(3): 303–310.

37.

Whittingham

Gamble

(1975) The lithium intercalates of the transition metal dichalcogenides. Materials Research Bulletin 10(5): 363–371.

38.

Whittingham

Guo

Chen

, et al. (1995) The hydrothermal synthesis of new oxide materials. Solid State Ionics 75: 257–268.

39.

Yin

Dong

Wang

, et al. (2021) Science as a public good: Public use and funding of science. Working Paper 28748, Cambridge, MA: National Bureau of Economic Research. Available at: https://www.nber.org/papers/w28748?utm_campaign=ntwh&utm_medium=email&utm_source=ntwg15

40.

Yoshino

(2014) 1 - Development of the lithium-ion battery and recent technological trends. Lithium-Ion Batteries Advances and Applications 2014: 1–20.

The relationship between the development level of American public libraries and the output of science and engineering publications

Abstract

Keywords

Introduction

Literature review

Model construction and data analysis

Construction of evaluation index system and data sources

Analysis of development level indexes of public libraries in the United States

Standardization

Validity analysis

Factor analysis

Comprehensive evaluation score for the development of public libraries in the United States

Model construction

Tests and analyses

Stationarity test

Cointegration test

Granger causality test

Impulse response function analysis

Conclusions

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

Author biographies

References