Research on Legal Education Enabling New Quality Productivity Development in the Context of Big Data

With the deepening of global economic integration, countries have taken education reform as a core strategy to enhance national competitiveness. Especially in the context of the new era, the Chinese government attaches great importance to the development of vocational education and regards it as an important way to promote high-quality development [1-4]. It also clearly puts forward to accelerate the construction of modern vocational education system, and cultivate more high-quality skilled talents with innovative spirit and practical ability. In this context, legal education to empower the development of new quality productivity has become an important issue in the current reform of vocational education [5-8].

Law is an indispensable part of the field of daily life of the society at large, therefore, legal education is extremely important to people [9-10]. Good legal education can provide students with correct knowledge of legal concepts, legal philosophy, legal means, etc., improve their legal literacy and legal awareness, and play a role in promoting the new quality productivity [11-14]. The new quality productivity refers to the mode of production based on scientific and technological innovation, led by information technology, biotechnology, new materials and other emerging industries, with knowledge, information, data and other factors, and driven by innovation as the core [15-18]. The development of new quality productivity has put forward new challenges and requirements for vocational education, which requires vocational education to keep pace with the development of the times and continuously improve the quality and level of education to meet the demand for skilled personnel in the development of new industries. The benign development of new quality productivity has a positive impact on the improvement of legal education [19-22].

The close relationship between education and new quality productivity has been widely discussed by scholars. In this paper, 2500 panel data of 19 cities, i.e., 229 prefecture-level cities, are selected from 2013-2023, and the unit root test and over-cointegration test are conducted on the panel data before data modeling. Subsequently, individual effect and time-effect analysis models are established to explore the causal relationship between legal education and the development of new quality productivity. Data calibration was performed based on the identification of condition and outcome variables. The necessity and sufficiency of calibrated data were analyzed using the fuzzy set qualitative comparative analysis method (fsQCA). The development path of legal talent, which drives new quality productivity, is analyzed.

2

Scientific and technological jurisprudential responses to the development of new qualitative productivity

The essential requirement for the development of new-quality productive forces lies in the all-round “quality” development of all elements of the social system, such as original and disruptive scientific and technological innovations, the efficiency of total factor production, new types of production relations, and changes in the economic and social system. Although law is a product of solidified concepts, as an independent system, it still needs to pay attention to the impact of new conceptual forces in the environment and constantly seek ways to cope with new things. The development of new productivity will certainly trigger deep changes in legal value, legal relationship and legal behavior, which will pose a brand new challenge to the existing legal framework in balancing the development of new productivity and security issues. In the era of big data, jurisprudence should shoulder the important task of empowering the development of new productivity: first, give full play to the catalog function of science and technology jurisprudence, and abstract the jurisprudence theoretical system to escort the development of new productivity. The second goal is to fully utilize the normative function of science and technology jurisprudence and clarify the legal rule system to facilitate the development of new quality productivity. Thirdly, fully utilize the aggregative function of science and technology law to establish a system of legal talent that can assist in the development of new quality productivity. According to the new trend of scientific and technological development, we should optimize the setting of disciplines and the mode of talent cultivation in higher education institutions, so as to cultivate urgently-needed talents for the development of new productive forces and the promotion of high-quality development. Science and technology jurisprudence should form a multidisciplinary talent aggregation, cultivate comprehensive legal talents conducive to escorting the development of new quality productivity, cultivate compound legal talents with innovative thinking and cross-border thinking, able to communicate and cooperate effectively with high-tech professionals, able to understand and apply the knowledge and methods of other fields to solve the problems of the intersection area of science and technology and law, and able to put forward new concepts and new ideas and new programs in the intersection area of science and technology and law. Legal talents.

3

Panel data and effects models

3.1

Panel data

Panel data can integrate and utilize a lot of information and are widely used in econometrics, social sciences and other fields. Panel data are discrete variables with respect to time t and can be represented by double subscripted variables such as X_n(i = 1, 2, …, N, t = 1, 2, …, T) where N denotes that the panel data contains N individuals and T denotes the length of the time series.

3.2

Unit root test for panel data

The data in the LLC test is generated in the form: (1) $y_{i t} = β y_{i, t - 1} + μ + δ_{1} t + α_{i} + u_{i t}, i = 1, \dots, N, t = 1, \dots, T$ $${y_{it}} = \beta {y_{i,t - 1}} + \mu + {\delta _1}t + {\alpha _i} + {u_{it}}\>,\>i = 1, \ldots ,N,t = 1, \ldots ,T$$

The ADF test equation is of the form: (2) $Δ y_{i t} = α_{m i} z_{m t} + β y_{i, t - 1} + \sum_{j = 1}^{p_{i}} ϕ_{i j} Δ y_{i, t - j} + u_{i t}$

The original assumption that each cross-section series of the panel data has the same unit root is H₀ : β = 1. This is done as follows:

Firstly, ADF test type regression was performed for each individual and further two residuals were obtained using auxiliary equations, i.e., doing Δy_it least squares regression on Δy_i,t−j and z_mt to get residual ${\hat{u}}_{i t}$ and then y_i,t−1 least squares regression on Δy_i,t−j and z_mt to get the second residual ${\hat{v}}_{i, t - 1}$ . Standardizing the above two residuals was obtained: (3) ${\tilde{u}}_{i t} = \frac{{\hat{u}}_{i t}}{{\hat{σ}}_{ε i}}, {\tilde{v}}_{i, t - 1} = \frac{{\hat{v}}_{i, t - 1}}{{\hat{σ}}_{ε i}}$

where ${\hat{σ}}_{s i}$ is the standard error of the regression of $Δ y_{i t} = α_{m i} z_{m t} + β y_{i, t - 1} + \sum_{j = 1}^{r_{l}} ϕ_{i j} Δ y_{i, t - j} + u_{i t}$ and ${\tilde{u}}_{n}$ and ${\tilde{ν}}_{i, t - 1}$ are two residual series that are orthogonal to each other.

Next, a least squares regression ${\tilde{u}}_{i t}$ about ${\tilde{v}}_{i, t - 1}$ is done, and then the t statistics of the regression coefficients are computed, which in turn estimate the ratio ${\hat{s}}_{i}$ of their standard deviation to the short-term standard deviation and the mean ratio is: (4) ${\hat{S}}_{N} = \frac{1}{N} \sum_{i = 1}^{N} {\hat{s}}_{i}$

where ${\hat{s}}_{i} = \frac{{\hat{σ}}_{y i}}{{\hat{σ}}_{u i}}$ , ${\hat{σ}}_{u i}$ are the standard deviations of the regression residuals u_i of the ADF test equation for each cross-section, and ${\hat{σ}}_{y i}$ are obtained by nonparametric estimation methods with: (5) ${\hat{σ}}_{y i}^{2} = \frac{1}{T - 1} \sum_{t = 2}^{T} Δ y_{i t}^{2} + 2 \sum_{l = 1}^{K} w_{K l} [\frac{1}{T - 1} \sum_{t = 2 + l}^{T} Δ y_{i t} Δ y_{i, t - 1}]$

where ${\hat{σ}}_{y i}$ takes the value of K when consistent estimation is performed, w_xi is the Bartlett kernel, and $w_{x i} = 1 - \frac{l}{K + 1}$ finally, the statistics of the LLC test are obtained by calculation. When the pair: (6) ${\tilde{u}}_{i t} = β {\tilde{v}}_{i, t - 1} + ε_{i t}$

When performing rounded least squares estimation, the statistic $t_{β} = \hat{β} / S (\hat{β})$ can be obtained for parameter β, Eq: (7) $\begin{array}{l} \hat{β} = \frac{\sum_{i = 1}^{N} \sum_{t = 2 + p_{i}}^{T} {\tilde{v}}_{i, t - 1} {\tilde{u}}_{i t}}{\sum_{i = 1}^{N} \sum_{t = 2 + p_{i}}^{T} {\tilde{v}}_{i, t - 1}^{2}} \\ S (\hat{β}) = \frac{{\hat{σ}}_{u}}{\sqrt{\sum_{i = 1}^{N} \sum_{t = 2 + p_{i}}^{T} {\tilde{v}}_{i, t - 1}^{2}}} \\ {\hat{σ}}_{u}^{2} = [\frac{1}{N T} \sum_{i = 1} \sum_{t = 2 + p_{i}} {({\tilde{u}}_{i t} - \hat{β} {\tilde{v}}_{i, t - 1})}^{2}] \end{array}$ $$\matrix{ {\hat \beta = {{\mathop \sum \limits_{i = 1}^N \mathop \sum \limits_{t = 2 + {p_i}}^T {{\tilde v}_{i,t - 1}}{{\tilde u}_{it}}} \over {\mathop \sum \limits_{i = 1}^N \mathop \sum \limits_{t = 2 + {p_i}}^T \tilde v_{i,t - 1}^2}}\,} \hfill \cr {S(\hat \beta ) = {{{{\hat \sigma }_u}} \over {\sqrt {\mathop \sum \limits_{i = 1}^N \mathop \sum \limits_{t = 2 + {p_i}}^T \tilde v_{i,t - 1}^2} }}\,} \hfill \cr {\hat \sigma _u^2 = [{1 \over {NT}}\mathop \sum \limits_{i = 1} \mathop \sum \limits_{t = 2 + {p_i}} {{({{\tilde u}_{it}} - \hat \beta {{\tilde v}_{i,t - 1}})}^2}]\,} \hfill \cr } $$

The test statistic for LLC is: (8) $t_{β}^{*} = \frac{t_{β} - I T {\hat{S}}_{I} {\hat{σ}}_{u}^{- 2} - S (\hat{β}) μ_{m T}^{*}}{σ_{m T}^{*}}$

where $σ_{m T}^{*}$ is the adjustment factor for the mean and $μ_{m T}^{*}$ is the adjustment factor for the variance. As the values of N and T converge to infinity at the same rate, the distribution of $t_{β}^{*}$ also converges to N(0, 1).

3.3

Cointegration test for panel data

Smooth series can be directly modeled in data, but when the series is not smooth, pseudo-regression phenomenon will be generated if the data is modeled directly. Although the use of difference can be transformed from non-smooth to smooth data, but this will lead to a large loss of data information, which is not conducive to modeling research and prediction. Some of the series itself is not smooth, but it is a certain linear combination is smooth, this smooth linear combination also indicates that there is a long-term stable equilibrium relationship between the non-smooth variables, this relationship is called cointegration.

3.4

Classification of panel data models

The models can be further categorized into individual fixed-effects models, point-in-time fixed-effects models, and point-in-time individual fixed-effects models based on changes in the intercept term. 1)

Individual fixed effects model. The coefficients of the individual fixed-effects model are constant while the intercept term varies only with individuals and not with time, i.e., α_it = δ_i, which corresponds to the model form: (9) $y_{i t} = δ_{i} + \sum_{k = 2}^{K} β_{k} x_{k i t} + u_{i t}$

2)

Point-in-time fixed effects model. The point-in-time fixed effects model has a constant coefficient term and an intercept term that varies only with time, i.e., α_it = η_t, and corresponds to a model of the form: (10) $y_{i t} = η_{t} + \sum_{k = 2}^{K} β_{k} x_{k i t} + u_{i t}$

3)

Individual point-in-time fixed effects model. The coefficient term of the individual time-point fixed-effects model remains unchanged, while the intercept term contains both the period effect η_t, which is affected by time, and the individual effect δ_i, which is affected by individuals, i.e., α_it = δ_i + η_t, corresponding to a model of the form: (11) $y_{i t} = δ_{i} + η_{t} + \sum_{k = 2}^{K} β_{k} x_{k i t} + u_{i t}$

3.5

Tests for moderating effects

The commonly used tests for the moderating effect are ANOVA and regression analysis, both of which are described below. 1)

Analysis of variance

If variables X and U are both categorical variables, then the significance of the moderating effect will be determined by performing an ANOVA on both variables. In order to analyze the moderating effect of X and U on Y, the total squares and SS_T need to be decomposed as follows: (12) $S S_{T} = S S_{X} + S S_{U} + S S_{X \times U} + S S_{E}$

where SS_X denotes the sum of squares of the effects of X, SS_U denotes the sum of squares of the effects of U, SS_X×U denotes the sum of squares of the effects of the interaction terms X and U, and SS_E is the sum of squares of the errors. The degrees of freedom corresponding to each of the above sums of squares have the following decomposition: (13) $d f_{T} = d f_{X} + d f_{U} + d f_{X \times U} + d f_{E}$

Use statistics: (14) $F = \frac{S S_{X \times U} / d f_{X \times U}}{S S_{E} / d f_{E}}$

To conduct the F test as a way to determine whether the moderating effect of U between X and Y is significant (the methods of sum-of-squares decomposition and degrees-of-freedom decomposition of the variables given here are only applicable to ANOVAs of between-subjects designs). 2)

Regression analysis method

When variables X and U are both continuous variables, a regression model with XU terms can be used when analyzing interaction effects: (15) $Y = c_{0} + c_{1} X + c_{2} U + c_{3} X U + ε$

where X, U is the main effect term and XU is the interaction effect term. The test statistic is to determine whether the interaction effect of X, U is significant by doing a t-test on hypothesis H₀ : c₃ = 0: (16) $t = \frac{{\hat{c}}_{3}}{s e ({\hat{c}}_{3})}$

where $s e ({\hat{c}}_{3})$ is the standard error of ${\hat{c}}_{3}$ .

In order to reduce the possibility of X and U showing multiple linear correlation with XU, the centered processed data from X and U were used to generate the product term XU, and the hierarchical regression analysis was done according to the following steps: 1)

Do a regression of Y on X and U to obtain a squared compound correlation coefficient of $R_{l}^{2}$ .

2)

Do a regression of Y on X, U and XU to obtain a squared compound correlation coefficient of $R_{2}^{2}$ . If the regression coefficient for XU is significant at this point, then $R_{2}^{2}$ is significantly higher than $R_{l}^{2}$ , where the change in R² is an additional contribution from $Δ R^{2} = R_{2}^{2} - R_{1}^{2}$ to measure the interaction effect.

4

Study on legal education as an enabler of new quality productivity development

4.1

Descriptive statistics

Between obtaining data and conducting empirical analysis, descriptive statistics of the data are needed, and the basic statistical information of the observed acquired data is shown in Table 1. Therefore, the data of all 229 prefecture-level cities covered by 19 city clusters are selected from 2013-2023, a total of 11 years, to form a short panel with 2,500 records. All the data in this paper come from official published data from open sources such as China Urban Statistical Yearbook, provincial statistical yearbooks, and statistical bulletins on innovation, science and technology, and social development of each city.

Table 1.

Variable descriptive statistics

Variable	N	Mean	SD	Min	Max
New productivity(NP)	2500	0.041	0.041	0.04	0.672
Legal education(LE)	2500	0.52	0.224	0	1
Education discipline setting (EDS)	2500	0.074	0.105	0	1
Talent training mode (TTM)	2500	0.043	0.0628	0	1
Complex legal personnel (CLP)	2500		0.0239	0	1
High-tech (HT)	2500	0.224	0.322	0	3.124
Science and technology (SAT)	2500	0.0826	0.0358	0	0.418
Large number according to education (LNATE)	2500	0.394	0.129	0.024	1.022

4.2

Benchmark panel regression

The results of the fixed effects model are shown in Table 2.

Table 2.

Fixed effect model results

Variable	Individual fixation effect	Individual fixation effect	Individual fixation effect -LSDV	Individual fixation effect -LSDV	Two-way fixed effect - individual/time	Two-way fixed effect - individual/time
Variable	NP	NP	NP	NP	NP	NP
LE	0.0185***(0.00463)	0.0168***(0.00375)	0.0182***(0.00232)	0.0179***(0.00215)	0.245***(0.0620)	0.213***(0.0516)
CLP		-0.00422***(0.00124)		-0.00428(0.00587)		-0.00268**(0.00117)
HT		-0.0442**(0.0195)		-0.0442***(0.0108)		-0.0436**(0.0187)
SAT		-0.0385(0.0472)		-0.0385(0.0348)		-0.0122(0.0467)
LNATE		-0.0214***(0.00756)		-0.0214***(000620)		-0.0147*(0.00796)
2014					-0.0285**(0.00678)	-0.0254***(0.0542)
2015					-0.0524***(0.0136)	-0.0478***(0.0114)
2016					-0.0635***(0.0162)	-0.0568***(0.0132)
2017					-0.0798***(0.0215)	-0.0705***(0.0175)
2018					-0.0945***(0.0259)	-0.0845***(0.0216)
2019					-0.125***(0.0305)	-0.0985***(0.0285)
2020					-0.119***(0.0326)	-0.115***(0.0263)
2021					-0.128***(0.0336)	-0.112***(0.0274)
2022					-0.134***(0.0362)	-0.115***(0.0294)
2023					-0.145***(0.0425)	-0.134***(0.0322)
_cons	0.0272***(0.00258)	0.0489***(0.00685)	0.378***(0.0415)	0.436***(0.0415)	0.00512(0.00867)	0.0240***(0.00785)
N	2500	2500	2500	2500	2500	2500
R²	0.063	0.163	0.879	0.895	0.148	0.227
Adj.R²	0.063	0.162	0.874	0.896	0.137	0.225

Although the R² and the adjusted R² are small, the F-statistic is significant at the 0.01 level, so statistically it can be said to be very significant.

There are two kinds of fixed effect models, individual fixed effect and time fixed effect, individual fixed effect mainly solves the problem of omitted variables that do not change over time but vary with individuals, there are two measures about individual fixed effect, one is to take the average and then do the deviation to get the deviation variable within the group, and then regression, and the other is to use dummy variables, introduce (n-1) dummy variables to represent the different individuals (if the constant term is not retained, the n dummy variables are introduced), also known as the least squares dummy variable model, or LSDV method, which has the advantage of obtaining an estimate of individual heterogeneity. Time effects mainly address the problem of omitted variables that do not vary with individuals but vary over time. At the same time, it is possible for panel data to have both of these effects, so the examination of individual time series with two-way fixed effects has gradually increased in recent years.

In this paper, we examine individual fixed effects, time fixed effects, and two-way fixed effects one by one, and distinguish between the effects before and after adding control variables. From the results, no matter which fixed effects, the effect of legal education empowering the development of new quality productivity is very significant, and the two have the same effect, the regression coefficient is roughly around 21.3%, i.e., for every 1% increase in the index of legal education in the city on average, the index of new quality productivity in the city will be increased by about 21.3% accordingly, which has the effect of amplification. The inclusion of control variables does not affect the significance of the coefficients, although it makes the regression coefficients smaller. The two measures of individual fixed effects yield coefficients for the main explanatory and control variables. The results are the same, with the LSDV method estimate being slightly larger under the constant. However, all coefficients are highly significant, which proves that there are strong individual effects on the effectiveness of legal education in empowering the development of new qualitative productivity, with non-negligible differences between cities.

4.3

Random effects model

The results of the random effects model are shown in Table 3 with two estimation methods: generalized least squares (FGLS) and maximum likelihood estimation (MLE). Like the fixed effects, the random effects are contrasted with the inclusion of control variables or not, and the result is that the coefficients on legal education with the inclusion of control variables are a little bit smaller. Without the inclusion of control variables, the coefficient on legal education estimated by FGLS is as large as that of MLE, but the results of FGLS are highly significant and those of MLE are insignificant, but with the inclusion of control variables, the results of MLE are a little bit larger: the MLE approach yields that on average, for every 1% increase in legal education, the city’s index of NQP improves by 27.8% on average.The results of the FGLS approach yield result that for every 1% increase in legal education, the city’s NQP index increases by an average of 3.57%. The coefficients are significant in both cases. Generally speaking, the differences in the coefficients of the main explanatory variables of these two methods are common, but they are still similar in nature, and the impact of legal education on the development of NQP is positive.

Table 3.

Random effect model

	FGLS	FGLS	MLE	MLE
LE	0.0475***(0.0144)	0.0357***(0.00972)	0.0468(.)	0.278***(0.0143)
CLP		0.0214*(0.0118)		0.0423(0.0345)
HT		0.0496***(0.00497)		0.0277***(0.00272)
SAT		0.322***(0.0765)		0.242***(0.0249)
LNATE		0.00245(0.00536)		-0.00998*(0.00639)
_cons	0.0132(0.00896)	-0.0189(0.0122)	0.0138***(0.00215)	-0.125***(0.0178)
Sigma_u_cons			0.0597***(0.00163)	0.0522***(0.0122)
Sigma_e			0.0597***(0.00163)	0.0522***(0.0122)
_cons			0.0415***(0.000586)	0.0346***(0.000477)
N	2500	2500	2500	2500

On the whole, for every 1 per cent increase in the legal education index, the city’s new productivity index increases by an average of 20 per cent, and legal education as a whole has a positive amplifying effect on the development of new productivity. When legal education is divided into three aspects: breadth of coverage, depth of use and level of digitization, the effect of each aspect is smaller than that of the whole, so when actually taking the path of developing legal education to empower the development of new quality productivity, attention should be paid to the overall planning and synergistic promotion of legal education, and no one aspect should be neglected.

5

fsQCA analysis steps

Different from the traditional quantitative regression analysis method, this paper adopts a multi-case grouped comparative analysis method based on the logic of set relationship, i.e., qualitative comparative analysis (QCA). fsQCA is to assign the variables between [0,1], i.e., the variables are in any value between fully affiliated and fully unaffiliated, and the assignment of the variables is not fixed-distance, but continuous and systematic. 1)

Case selection. fsQCA method is based on quantitative research with cases as research units. After clarifying the direction and content of the study, selecting the research object, and choosing the appropriate case as the research object is an important prerequisite for conducting the study, so the scope and number of selected cases need to be clarified.

2)

Variable selection. Variables can be selected through literature induction method, theoretical perspective method, problem-oriented method, expert scoring method and other research methods, so as to determine the outcome variables and condition variables and their number and measure.

3)

Data collection. According to the research case and metric conditions, start collecting relevant data. It can be inquired and collected through questionnaires, enterprise interviews, finding databases, and organizing secondary data.

4)

Data calibration. Due to the different units of data measurement of each variable, it is necessary to calibrate the data of each variable when performing the fsQCA method operation, and transform all the data into the concept of set, and all the values after the transformation are fuzzy affiliation scores located between [0,1].

5)

Necessity analysis. Necessity analysis is performed on individual conditional variables to determine whether they are necessary to constitute the occurrence of the outcome variable. When the necessity analysis consistency value is greater than or equal to 0.9, the condition is considered a necessary condition for the outcome.

6)

Construction of truth table and sufficiency analysis. The truth table is generated by fsQCA software, if there are k condition variables, logically there are 2k potential combinations of conditions in the truth table, the truth table is screened by setting the frequency threshold and the consistency threshold, and then standardized analysis is carried out to obtain the simple solution, the complex solution, and the intermediate solution, and the intermediate solution is usually analyzed.

7)

Robustness test. The robustness test is used to test the stability of the results, which is usually carried out by adjusting the calibration anchor point, changing the case frequency and consistency threshold, and supplementing or reducing the cases.

8)

Results are further analyzed. The intermediate solution is analyzed to derive the corresponding several groupings, and the individual groupings are compared horizontally to observe the complementary, alternating, or mutually inhibiting relationships that exist between the groupings. For different types of group states, the case dialog is realized through typical case explanations to deepen the analysis of results.

6

Study on the path of development of new quality productivity driven by legal talents

6.1

Individual Conditional Necessity Analysis

In the context of big data, legal talent is the link between higher education and new-quality productivity, and the interaction between higher education and new-quality productivity cannot be separated from the cultivation of legal talent. First of all, we should clarify what qualities innovative talents should have in the context of accelerating the formation and development of new quality productivity. Secondly, the cultivation of innovative talents is a systematic and long-term project, and it is far from enough to carry out reforms in the field of higher education alone. How to find and cultivate innovative talents in the stage of basic education, so as to lay a good learning foundation for them to receive higher education in the future, is also a question we need to think about. Universities are responsible for running schools that reflect their own development and cultivate personalized talents with special characteristics. Finally, it is important to encourage a group of qualified colleges and universities to actively explore the practice of legal talents in order to provide valuable practical experience for the cultivation of legal talents. The theoretical framework of legal talents driving the development of new quality productivity is shown in Figure 1.

Based on the fsQCA analysis procedure, the necessity of individual conditions (including non-set ~) needs to be examined before conducting the condition grouping sufficiency analysis, and the results are shown in Figure 2. The level of technological innovation in traditional industries (Y1), the level of technological innovation in strategic emerging industries (Y2) and the level of technological innovation in future industries (Y3).

In the legal talent agglomeration effect: leading talents (T1) are measured using the standardized average of the number of academicians of the two academies and the number of scientific and technological innovation leaders in each city. Legal basic talents (T2) are measured using the number of experts with full senior titles in each city. Legal technical talents (T3) are measured using the number of R&D personnel in each city. Legal Skilled Talent (T4) is measured using the number of employed persons with postgraduate degrees in each city. The indicators of social environment (E1), education environment (E2), innovation support (E3), and policy environment (E4) in the talent ecosystem. con denotes consistency, cov denotes coverage, and ~ denotes “not” in the set operation.

It is found that the consistency level of each conditional variable for high-level traditional industry technological innovation, high-level strategic emerging industry technological innovation and high-level future industry technological innovation is not higher than 0.90, which indicates that each conditional element of the effectiveness of legal talent plays an important role in the development of new quality productivity in the city, but none of the single factors constitutes a necessary condition. Therefore, it is necessary to carry out further conditional grouping analysis with the help of the fsQCA method to explore which combination of factors is the grouping condition for the formation of new quality productivity levels.

6.2

Conditional Configuration Sufficiency Analysis

● indicates that the core condition exists, ○ indicates that the marginal condition exists, × indicates that the core condition is missing, x indicates that the marginal condition is missing, and a space indicates that the condition is dispensable. As shown in Table 4, the eight conditional factors of legal talent effectiveness constitute the group configuration of high-level technological innovation in different industries. For traditional industry technological innovation, there are 4 groups of high-level grouping paths, and the overall consistency is 0.9253, which is higher than the consistency standard of 0.800 0, implying that 92.53% of the cities that meet the above 4 conditional groupings have demonstrated high-level traditional industry technological innovation. Among the 4 combinations of legal talent elements (A1, A2, A3, A4) for realizing high-level traditional industry technological innovation, A1 has the highest coverage rate (0.5143), indicating that A1 is the main path for realizing high-level traditional industry technological innovation. For strategic emerging industry technological innovation, there are 3 groups of high-level grouping paths, with an overall consistency of 0.9635, implying that 96.35% of the cities that meet the above 3 conditional groupings have demonstrated high-level strategic emerging industry technological innovation. In terms of the coverage of the 3 conditional groupings, B1 is the highest (0.5128), indicating that B1 is the main path to realize high-level strategic emerging industry technology innovation.

Table 4.

The condition group state sufficiency analysis

Conditional variable	High level traditional industrial technology innovation				High-tech innovation of high level strategic emerging industries			High level future industrial technology innovation
Conditional variable	A1	A2	A3	A4	B1	B2	B3	C1	C2
Leading talent	○	×	×	○	○	×	○	●	●
Basic science	○	×	×	×	○	×	×	○	×
Talent	○	○	×	○	●	●	●	●	●
Technical talent	●	●	●	●	●	●	●	●	●
Social environment		×	○	○		×	○		○
Education environment	○	×	×	×	○	×	×	○	×
Innovation support	○	○	○	○	●	●	●	●	●
Policy environment	○	×	×	○	○	×	○	●	●
Original coverage	0.5143	0.1332	0.0826	0.0937	0.5128	0.1252	0.0922	0.5143	0.0915
Unique coverage	0.4336	0.0846	0.0326	0.0341	0.4534	0.0936	0.0306	0.4639	0.0352
consistency	0.9445	0.8667	0.9563	0.9847	0.9925	0.8475	0.9956	0.9985	0.9979
Overall resolution	0.6639				0.6422			0.5528
Overall solution consistency	0.9253				0.9635			0.9977

For future industrial technological innovation, there are two groups of high-level grouping paths, with an overall consistency of 0.9977, implying that 99.77% of the cities in the cases that meet the above two conditional groupings exhibit high-level future industrial technological innovation.C1 (0.5143) is the main path to realize high-level future industrial technological innovation. Specifically, A1, B1 and C1 are the vertical talent agglomeration and talent ecosystem and heavy-duty driving paths, which are manifested as a good positive feedback effect between legal talent agglomeration and talent ecosystem, which jointly promote the city’s industrial development.

The representative cities are shown in Table 5. From the perspective of specific cities such as Beijing, Shanghai, Tianjin, Chongqing, Hangzhou, etc. are the representative cities of the main paths for each industry to realize high-level technological innovation, so it is evident that a high-quality talent ecological environment can prompt the gathering of multiple types of talents and give full play to their own efficacy, thus promoting urban innovation.

Table 5.

Representative city

	Conditional variable
High level traditional industrial technology innovation	A1	Beijing, Shanghai, Guangzhou, Hangzhou State, Suzhou, Nanjing, Wuhan, Xi ‘an, Chengdu, Hefei and heavy Qing, Qingdao, Tianjin, Zhengzhou
	A2	Dongguan, Foshan, Nantong
	A3	Shaoxing, Wenzhou
	A4	Shenzhen, Wuxi, Xiamen
High-tech innovation of high level strategic emerging industries	B1	Beijing, Shanghai, Guangzhou, Hangzhou, Nanjing, Suzhou, Wuhan, Chengdu, Xi ‘an, Hefei, Chongqing, Qingdao, Tianjin
	B2	Dongguan, Foshan, Nantong
	B3	Shenzhen, Wuxi, Xiamen
High level future industrial technology innovation	C1	Beijing, Shanghai, Hangzhou, Guangzhou, Nanjing, Wuhan, Suzhou, Chengdu, Xi ‘an, Hefei, Qingdao, Chongqing, Jinan,Tianjin
High level future industrial technology innovation	C2	Shenzhen, Wuxi, Xiamen

7

Conclusion

Scientific and technological innovation is driven by talents, and the cultivation of talents is realized through education. This paper will utilize panel data theory and effect analysis models to conduct research on the development of new quality productivity empowered by legal education. Based on the determination of the type of outcome variables and condition variables, the fuzzy set qualitative comparative analysis method is used to carry out the research on the development path of legal talent-driven new quality productivity. The study finds: 1)

From the results of individual effect and time effect, the regression coefficient of legal education empowering the development of new quality productivity is roughly 21.3%. Therefore, the impact of legal education on the development of new quality productivity is significant. The two have a positive correlation, and there is a strong individual effect.

2)

Between legal talent agglomeration and talent ecological environment to promote the development of new quality industry. A good talent ecosystem can promote the further aggregation of multiple types of talent, thus promoting the development of innovation.

Język:: Angielski

Częstotliwość wydawania:: 1 razy w roku
Dziedziny czasopisma:: Nauki biologiczne, Nauki biologiczne, inne, Matematyka, Matematyka stosowana, Matematyka ogólna, Fizyka, Fizyka, inne

Kanał RSS czasopisma

Research on Legal Education Enabling New Quality Productivity Development in the Context of Big Data

Ping Liu

Data publikacji: 24 mar 2025

Otrzymano: 23 paź 2024

Przyjęty: 18 lut 2025

DOI: https://doi.org/10.2478/amns-2025-0759

Słowa kluczoweNew quality productivity, Individual effect, Time effect, Qualitative comparative analysis

© 2025 Ping Liu, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Słowa kluczowe
New quality productivity, Individual effect, Time effect, Qualitative comparative analysis