Different from the cross-section (Cross-sectional) and time sequence (time-server) data，Panel data (Panel Data) (or vertical data、Tracking data) At the same time, it contains multiple cross -sectional analysis units and time sequences。Based on its own characteristics，Panel data not only provides more comprehensive data information，and is conducive to scientific causality analysis。This article uses panel data as a small incision，It aims to appeal: with the advent of the era of big data and the extensive promotion of various statistical software，Big data analysis has important strategic significance in the application of scientificization and expansion of political science research in Chinese political science research。

bet365 Play online games

　　Regardless of natural science or social science，Human exploration of scientific knowledge starts with questioning。But，The end point of scientific exploration will not only stay on question，Great social scientists not only proposed but also successfully answered the classic questions of "why A can cause B"。Carl Marx and Max Weber successfully answered "Why the development of capitalist economic development worsening the living conditions of ordinary workers" and "how the new teaching ethics affects the development of Western capitalism"。The contemporary world is catching up with the fourth industrial revolution，How to use informatization and intelligent technology to respond to major social problems and challenges, it has become the time of the era of social science research practitioners。As one of the pillar disciplines of social sciences，Shoulder the same responsibility。

Starting since the middle and late 20th century，Deeply influenced by the "Behavioral Revolution"，The focus of modern political studies began to turn on the occurrence。The purpose of academic research and the mode of academic training has gradually changed from macro theory construction and system description to the use of scientific methods to provide reliable evidence for traditional theory and political practice。In the 21st century to drop，Innovation of information and computer science and technology，Get and use big data to promote the deep development of political research into a new trend。At present, the core of empirical political research with data -driven data is to transcend the correlation between research variables，Seeking to establish political propositions and theories with strict causal relationships，Then accumulate scientific knowledge for Chinese political research，Provide a reliable empirical basis for the introduction of national policy。

"Causal Revolution" of scientific thinking

Cause and effect reasoning is the core of scientific research。The causal relationship that seeks theoretically and experienced is the core task of political science and even social sciences。The Turing Award winner、The father of "Bayesian Network" Judea Pearl (Judea Pearl) referred to "causal inference" as the "new revolution" in human scientific thinking，that is "Causal Revolution"。Specifically，Cause and effect inference aims to advance the research assumption (assuming variables or the relationship between bet365 live casino games events) from correlation description to causal relationship。

Establishing a research assumption that the existence of causality generally needs to meet three conditions。First，Research hypothesis "Explanation variables" or "independent variables" and "result variables" or "due to variables" have a high degree of correlation。A classic example is，The two incidents of the sales of ice cream and the frequency of being attacked by shark are often highly correlated on the beach of the midsummer。So，How can there be a causal relationship between the two incidents? second，The order of the two incidents that require causality must be in line with the reason for the reason.，The result is later。Very obvious，The sales volume of ice cream and the shark attack tourists do not exist such a cause and consequences。So，We can exclude two incidents to have a causal relationship。third，Cause and effect requires a real and reliable causality mechanism between two events or variables。In other words，illegal variables or other interference factors may jointly affect two events or variables，Make a "common change" relationship between the two。For example，High -temperature weather may be behind the scenes that lead to the increase in the sales volume of ice cream and the number of tourists being attacked by the number of tourists.。

　　The advantages and strategies of panel data causal reasoning

According to data features，Observation data is mainly divided into three types: cross -section data、Time serial data and panel or vertical data。Cross -section data is a given time for individuals、Collective、Enterprise、City、Sample data collected by the country or a series of other units，For example, the GDP of several countries Bet365 lotto review in East Asia in 2021。Time sequence data is the observation value of collecting one or several variables at different times，For example, the changes in the GDP of China in the past 10 years。Panel data is composed of a time sequence of a data concentration of each cross section unit，For example, the domestic GDP information of all countries in East Asia in the past 10 years is a panel data。

Panel data Multiple observations for the same unit，It helps us to control some of the observations that are not observed in different observations，It can also help us study and evaluate the lag of behavior and results，For example, analyzing the government's economic policy affects the economic growth of East Asian countries。Even if East Asian countries can control the variables of East Asian countries，Including population、Capital、Education、Age and other factors，We still cannot examine and observe the possible interference factors，Especially between the country due to history、Culture and other differences。Panel data analysis technology can be ignored by ignoring the differences between the country，Only observe the differences in the internal policy of one country (a national policy changes at different times) for causal inference。This analysis method is called a panel fixed effect model (Panel Fixed Effects Model)。Because the policy effect of a certain policy at different times in the same country，The unit observed is homogeneous，Therefore, other interference factors，Especially the observed observation and observed confusion variables that are "not changing over time" are controlled。

The high-level methods in inferred by panel data causal causality include the Difference-in-Difference Bet365 lotto review Method and the Synthetic Control Method。Double differential methods The effect of one policy cross -time between two different units through two fixed effects。When the two units meet the anti -factual hypothesis，This valuation is the effect of policy。Classic example is to evaluate the impact of minimum wages on local employment rates。Yunnan Province and Guizhou Province are neighbor provinces，The impact on the implementation of the minimum wage policy on Yunnan can be estimated through the dual difference method。Specifically，The difference between Yunnan Province before and after policy introduction and then minus the difference between Guizhou, which has not implemented the new policy.。If the employment rate of the two provinces before the policy is introduced, it remains parallel (in line with the anti -factual hypothesis)，This model is estimated to have a causal effect。essentially，This method is to simulate the control group and experiment group in the simulation experiment to evaluate the effect of the policy。

Comprehensive control case method can be said to be an extension of dual differential methods。In social science research，Finding a control team that can meet parallel assumptions is often difficult。Comprehensive control case methods By giving other comparison units a certain weight，Human constructing an analog unit similar to the experimental group for causal inference。In addition to the two high -level methods above，Panel data can also be used by nested tool variables (Instrumental Variable Method)、Lagged Dependent Variable Method, etc. to deal with other problems facing causal inference。In short，The characteristics and advantages based on the data structure itself，Panel data analysis has Bet365 lotto review greatly expanded causal inference in social science，Applications in political studies。

　　Big Data Analysis Promoting Political Study Style Transformation

The advent of the era of big data to promote the change of political research paradigm。Big data analysis further promotes the political scientific research mode of "theoretical hypothesis drive"，Provides reliable for political decision -making and policy implementation、Solid evidence。Big data analysis helps causality becomes a powerful tool for political scientific research。Combined with machine learning，Big data cutting -edge research will play a vital role in prediction and warning of major political issues。Combination of big data analysis and cause and effect relationships to enhance the degree of science of the study of political disciplines。Use standardized research design and scientific data analysis，Display the logic (mechanism) behind the political phenomenon clearly，Then bring innovation of political knowledge。Scientific knowledge production methods have the scientific accumulation of scientific accumulation of political science knowledge。

As General Secretary Xi Jinping said，"Current，Speed ​​evolution without a large change in the world for 100 years，The change of the world、Changes in the era、The characteristics of historical changes are more obvious，my country's development faces new strategic opportunities、New strategic tasks、New Strategic Phase、New strategic requirements、New Strategic Environment "。This requires that politicians faced many domestic and foreign challenges at home and abroad。Big Data Analysis of the Scientific Prevention of the New Crown Pneumonia Epidemic、Rural grassroots bet365 best casino games governance、Social Security and Distribution、National image construction、China's rise and the construction of the "Belt and Road" have huge application space。

　(Author Unit: Department of Political Science, School of Government Management, Yunnan University)