五楼自拍

  2026年五楼自拍 国际小学期课程

社会统计学：原理、理解与前沿

Social Statistics: Fundamentals, Interpretation, and Advances

授课教师：傅强博士

、

授课教师简介：傅强博士，加拿大英属哥伦比亚大学（The University of British Columbia）公共政策学院与全球事务学院讲席教授、社会学系合聘副教授。与齐慕实院士共同担任UBC中华研究中心联合主任，曾任UBC大数据与计算社会科学研究卓越团队共同负责人，加拿大人口学会常委以及社会学会司库、常委、以及研究道德与职业伦理委员会主席等职。分别于暨南大学，北京大学以及美国杜克大学获得本科、硕士和博士学位。研究成果发表于American Journal of Epidemiology, American Journal of Sociology, Annuals of the American Association of Geographers, Annals of Epidemiology, Journal of the Royal Statistical Society: Series A, Justice Quarterly, Social Networks, Social Science & Medicine以及Sociological Methods & Research等地理学、流行病学、犯罪学、统计学、社会学、健康领域的顶尖期刊。近年来的研究重点转向制度变迁、互动、以及社会史等主题的中国本土化质性和田野研究。

课程简介：社会统计学不仅是社会学家理解社会结构、解释社会现象的重要工具，也广泛应用于经济学、人口学、政治学等多个学科领域。本课程将在介绍描述统计、推断统计等基础内容的基础上，进一步探讨因果推断、回归分析、简化数据、分层线性模型、年龄-世代-同期群分析、泊松模型、负二项模型等社会科学前沿方法，并结合经典研究案例与实际数据操作，引导学生提升数据分析能力与学术研究素养，培养运用定量方法分析现实社会问题的能力。本课程不仅聚焦统计模型是什么，更试图深入浅出地阐释模型背后基本的统计原理，使学生真正能做到学以致用，用实证的方法揭示和理解社会运行的本质及其规律。

课程时间：7月7号全天、7月8号下午、7月9号全天

(7月8号上午9点半五楼自拍 楼144讲座：机器学习视角下的核心讨论:网络规模与测量偏差【自选】）

课程地点：津南校区公共教学楼C区220

课程内容安排：

Session 1（July 7^th, Morning）: Epistemological Foundations & Fundamentals of Social Statistics

·The Paradigm of Empirical Inquiry:Distinguishing empirical research from theoretical speculation within the social sciences.

·Taxonomy of Measurement:Navigating and coding nominal, ordinal, and interval-ratio variables for quantitative modeling.

·Descriptive vs. Inferential Paradigms:The mathematical transition from sample description to population generalization.

·Anatomy of Variance:Deconstructing the formulas and theoretical intuitions behind measures of central tendency and data dispersion.

Session 2（July 7^th, Afternoon）: Probability Distributions & The Logic of Statistical Inference

·Asymptotic Theory:Utilizing the Central Limit Theorem to understand the behavior and convergence of sampling distributions.

·Parameter Estimation:Constructing confidence intervals and mathematically quantifying statistical uncertainty.

·The Hypothesis Testing Framework:Formulating null models, calculating test statistics (Z and t-distributions), and interpretingp-values.

·Statistical Errors:Navigating the critical trade-offs between Type I and Type II errors in empirical decision-making.

Session3（July 8^th, Afternoon）: Ordinary Least Squares (OLS) Regression Mechanics

·Bivariate Associations:Geometric and mathematical interpretations of Pearson's Correlation Coefficient and covariance.

·The Calculus of OLS:Deriving the best-fitting linear regression model by minimizing the sum of squared errors (SSE).

·Model Diagnostics & Evaluation:Partitioning variance (SST, SSR, SSE) and interpreting the Coefficient of Determination.

·The Gauss-Markov Theorem:Evaluating the fundamental assumptions of classical linear regression models to ensure Best Linear Unbiased Estimators (BLUE).

Session 4（July 9^th, Morning）: Matrix Algebra & The Geometry of Regression

·Fundamentals of Linear Transformations:Matrix operations, transposes, inverses, and the geometric interpretation of determinants.

·OLS in Matrix Notation:Formulating and solving the classical regression estimator through linear algebra.

·Vector Spaces & Projections:Understanding linear regression visually and mathematically as orthogonal projections onto column spaces.

·Diagnosing Rank Deficiency:The mathematical causes, consequences, and singularity issues resulting from perfect multicollinearity.

Session 5（July 9^th, Afternoon）: Advanced Regularization & Dimensionality Reduction

·Spectral Decomposition:The role of eigenvalues and eigenvectors in identifying matrices' directions of maximum variance.

·Principal Component Regression (PCR):Feature extraction, orthogonal transformations, and the targeted removal of null spaces.

·Ridge Regression:Imposing continuous shrinkage penalties to stabilize ill-conditioned matrices and mitigate multicollinearity.

·LASSO:Utilizing sparse shrinkage constraints and absolute value penalties for automated feature selection.

课程联系人：杜思佳

联系方式：[email protected]