📚 Quantitative Methods and Data Analysis in Economics | 经济学中的定量研究方法与数据分析
Quantitative methods are the backbone of empirical economics. They allow economists to test theories, estimate relationships, and forecast future trends using numerical evidence. From measuring inflation to evaluating the impact of a tax change, quantitative data analysis transforms abstract concepts into testable, decision-ready facts.
定量方法是实证经济学的核心支柱。借助数值证据，经济学家能够检验理论、估计变量之间的关系并预测未来趋势。从衡量通胀率到评估税收改革的影响，定量数据分析将抽象概念转化为可检验且利于决策的事实。
1. Why Quantitative Methods Matter | 定量方法的重要性
Economics is not merely a theoretical discipline; it requires empirical validation. Quantitative methods provide a structured framework for collecting, summarising, and interpreting numerical data. They help answer questions such as 'How does a 1% rise in interest rates affect investment?' or 'What is the price elasticity of demand for petrol?'
经济学并非纯粹的理论学科，它需要实证检验。定量方法为收集、汇总和解读数值数据提供了系统化框架，帮助回答诸如“利率上升1%如何影响投资？”或“汽油的需求价格弹性是多少？”等问题。
Unlike qualitative opinions, quantitative evidence is replicable and objective. When presented with clear definitions and appropriate techniques, statistical results can guide government policy, business strategy, and individual choices with measurable confidence.
与定性观点不同，定量证据具有可复制性和客观性。在明确定义和恰当技术的支持下，统计结果能够以可衡量的置信度指导政府政策、企业战略和个人选择。
2. Types of Economic Data | 经济数据的类型
Economic data can be classified into three main types. Cross-sectional data capture observations of multiple units (e.g., households, firms, countries) at a single point in time. Time-series data track one variable (e.g., GDP, unemployment) over successive periods, such as months or years. Panel (longitudinal) data combine both dimensions, following many units over time.
经济数据可分为三类主要形式。截面数据在某一时点上收集多个单位（如家庭、企业、国家）的观测值。时间序列数据在连续时间段（如月度、年度）追踪一个变量（如GDP、失业率）。面板（纵向）数据结合了前两者，在多个时期内追踪多个单位。
Cross-sectional example: household income and spending across 1,000 families in 2024.
截面数据示例：2024年1000户家庭的收入与支出。
Time-series example: UK inflation rate, monthly from 2010 to 2024.
时间序列数据示例：英国2010年至2024年的月度通胀率。
Panel example: GDP growth for 20 European countries from 2000 to 2023.
面板数据示例：2000年至2023年20个欧洲国家的GDP增长率。
3. Sources and Sampling of Data | 数据来源与抽样
Reliable data sources are essential for valid conclusions. Economists use official statistics from national agencies (e.g., Office for National Statistics), international organisations (e.g., World Bank, IMF), central banks, and academic surveys. Each source has its own strengths and limitations in coverage, frequency, and accuracy.
可靠的数据来源是有效结论的前提。经济学家利用国家机构（如国家统计局）、国际组织（如世界银行、国际货币基金组织）、中央银行及学术调查的官方统计数据。每个来源在覆盖面、频率和准确性上各有优缺点。
When a full population cannot be studied, sampling is used. A random sample gives every member of the population an equal chance of selection, reducing bias. However, sample size and selection method matter: small or non-random samples may produce unrepresentative estimates.
当无法对全体总体进行研究时，就需要抽样。随机抽样使总体中每个成员被选中的机会均等，从而减少偏差。但样本量和抽样方法至关重要：样本过小或非随机抽样可能产生不具有代表性的估计值。
4. Descriptive Statistics: Central Tendency | 描述性统计：集中趋势
Descriptive statistics summarise data. Measures of central tendency locate the 'typical' value. The mean is the arithmetic average, the median is the middle value when data are ordered, and the mode is the most frequent value.
描述性统计用于概括数据。集中趋势指标确定“典型”值的位置。均值是算术平均数，中位数是数据排序后位于中间位置的值，众数是出现频率最高的值。
For a dataset of wages: £20,000; £22,000; £25,000; £30,000; £80,000, the mean is £35,400, while the median is £25,000. The median is often more robust because extreme values (outliers) skew the mean upward or downward.
对于一组薪资数据：20,000英镑；22,000英镑；25,000英镑；30,000英镑；80,000英镑，均值为35,400英镑，而中位数为25,000英镑。中位数通常更具稳健性，因为极端值（离群值）会使均值偏上或偏下。
Mean = Σx ÷ n, where Σx is the sum of all values and n is the number of observations.
均值 = Σx ÷ n，其中Σx是全部数值之和，n是观测次数。
5. Dispersion and Spread | 离散程度与波动
Dispersion measures how spread out the data are. The range is the difference between the maximum and minimum values. The variance and standard deviation capture the average distance of observations from the mean.
离散程度衡量数据的分散情况。极差是最大值与最小值之差。方差和标准差反映各观测值偏离均值的平均距离。
A low standard deviation indicates data points cluster closely around the mean; a high standard deviation implies greater volatility. For example, two regions may have the same average income, but one may have far higher inequality, as shown by a larger standard deviation.
标准差较低说明数据点紧密围绕均值分布；标准差较高则意味着波动较大。例如，两个地区可能拥有相同的平均收入，但其中一个地区的不平等程度可能更高，这通过较大的标准差得以体现。
Variance = Σ(xᵢ − μ)² ÷ n, and Standard Deviation = √Variance.
方差 = Σ(xᵢ − μ)² ÷ n，标准差 = √方差。
6. Index Numbers | 指数
Index numbers measure changes in a variable relative to a base year, which is assigned a value of 100. They are indispensable for tracking inflation, output growth, and share prices. The consumer price index (CPI) and retail price index (RPI) are common examples.
指数是衡量某一变量相对于基期（设为100）的变化。它们在追踪通胀、产出增长和股价方面不可或缺。消费者价格指数（CPI）和零售价格指数（RPI）是常见例子。
A weighted index reflects the relative importance of different items. For instance, if housing accounts for 30% of household spending and food for 15%, a rise in house prices has a greater impact on the overall CPI than an equal percentage rise in food prices.
加权指数反映不同项目的重要性。例如，若住房占家庭支出的30%，食品占15%，则房价上涨对整体CPI的影响大于同等百分比的食品价格上涨。
Index in year t = (Value in year t ÷ Value in base year) × 100
第t年的指数 = (第t年的数值 ÷ 基年数值) × 100
7. Correlation vs Causation | 相关与因果
Correlation measures the strength and direction of a linear relationship between two variables. A scatter diagram illustrates this visually, while Pearson's correlation coefficient (r) ranges from −1 to +1. r = +1 indicates perfect positive correlation, r = −1 indicates perfect negative correlation, and r = 0 indicates no linear relationship.
相关衡量两个变量之间线性关系的强度和方向。散点图可以直观显示这种关系，皮尔逊相关系数（r）的取值范围为−1到+1。r = +1表示完全正相关，r = −1表示完全负相关，r = 0表示不存在线性关系。
Correlation does not mean causation. Ice cream sales and drowning incidents may both rise in summer, showing positive correlation, but ice cream does not cause drowning. The true cause is the confounder 'temperature'. Economists must carefully control for omitted variables before claiming causality.
相关不等于因果。夏季冰淇淋销量与溺水事件可能同时上升，呈现正相关，但冰淇淋并不会导致溺水，真正的混杂因素是“气温”。经济学家在断言因果关系之前，必须仔细控制遗漏变量。
8. Regression Analysis | 回归分析
Regression analysis estimates how a dependent variable (Y) changes when an independent variable (X) changes, holding other factors constant. A simple linear regression is written as Y = a + bX, where a is the intercept and b is the slope coefficient.
回归分析用于估计在其他因素保持不变的情况下，因变量（Y）如何随自变量（X）变化而变化。简单线性回归的表达式为Y = a + bX，其中a是截距，b是斜率系数。
The coefficient b tells us the marginal effect: if b = 0.8, then a 1-unit increase in X is associated with a 0.8-unit increase in Y on average. The ordinary least squares (OLS) method minimises the sum of squared vertical distances between the data points and the regression line.
系数b表示边际效应：若b = 0.8，则X每增加1个单位，Y平均增加0.8个单位。普通最小二乘法（OLS）使数据点与回归线之间垂直距离的平方和最小化。
Interpretation: A positive b suggests a direct relationship; a negative b suggests an inverse relationship.
解读：b为正表示正向关系；b为负表示反向关系。
Goodness of fit: R² measures the proportion of variation in Y explained by X, ranging from 0 to 1.
拟合优度：R²衡量Y的变动由X解释的比例，取值范围为0到1。
9. Hypothesis Testing and Significance | 假设检验与显著性
Hypothesis testing is a formal procedure to decide whether an estimated relationship is statistically meaningful or just due to chance. The null hypothesis (H₀) usually states that there is no effect (e.g., b = 0), while the alternative hypothesis (H₁) states there is an effect (e.g., b ≠ 0).
假设检验是一种正式程序，用于判断估计关系是否具有统计意义，还是仅仅源于偶然。原假设（H₀）通常表明没有效应（如b = 0），备择假设（H₁）则说明存在效应（如b ≠ 0）。
A p-value measures the probability of obtaining the observed result if H₀ were true. If the p-value is below 0.05, economists typically reject the null hypothesis and call the result statistically significant. However, significance does not imply practical importance or causality.
p值衡量在原假设为真的情况下得到当前观测结果的概率。若p值低于0.05，经济学家通常会拒绝原假设，并称结果具有统计显著性。然而，显著性并不代表实际重要性或因果关系。
10. Limitations and Common Pitfalls | 局限性与常见误区
Quantitative methods are powerful but not infallible. Measurement errors may arise from inaccurate reporting or flawed survey design. Outliers and small sample sizes can distort estimates. Extrapolating a trend beyond the observed range is risky, especially during structural changes.
定量方法强大但并非万无一失。测量误差可能源于报告不准确或调查设计缺陷。离群值和小样本量会扭曲估计结果。将趋势外推到观测范围之外是有风险的，尤其是在结构性变化时期。
Spurious correlation: two unrelated series may appear correlated by coincidence; always check for logical connection.
伪相关：两个无关序列可能因巧合而看似相关；始终要检验逻辑联系。
Simpson's paradox: an overall trend may reverse when data are split into groups, misleading careless analysts.
辛普森悖论：当数据分组后，总体趋势可能反转，从而误导粗心的分析者。
Time lag: the effect of an economic policy may take years to appear, so current data may not reflect its full impact.
时滞：经济政策的效果可能需要数年才显现，因此当前数据可能无法反映其全部影响。
To conclude, quantitative methods and data analysis are indispensable tools for economic inquiry. Mastering descriptive statistics, index numbers, correlation, regression, and hypothesis testing enables students and practitioners to interpret evidence critically, avoid common errors, and make more informed decisions in a complex world.
总而言之，定量研究方法与数据分析是经济探究不可或缺的工具。掌握描述性统计、指数、相关、回归和假设检验，能够使学生和从业者批判性地解读证据、避开常见误区，并在复杂世界中做出更明智的决策。
Published by TutorHao | Economics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图