MIT OpenCourseWare


» 進階搜尋
 課程首頁
 教學大綱
 教學時程
 相關閱讀資料
 作業
 討論群組
 下載課程

教學大綱


本頁翻譯進度

燈號說明

審定:無
翻譯:翁于棻(簡介並寄信)
編輯:侯嘉玨(簡介並寄信)

概觀

當代對組織、策略與管理的研究,大部份依據定量研究方法。本課程將介紹一些最常使用的定量方法,包括對數轉換值/機率轉換值模型、計算模型、事件歷史模型,並將各技術做跨區域整合。

這是關於研究過程的一門課程。我明確的目標是在幫助你理解理論、數據和統計方法之間的關係。這不是一門統計學理論課程;雖然會明確地依據統計過程訂定假說,但並不會花很多時間在推導可能的函數…等等,而是如何使用統計方法來回答研究問題。我們會花相當多的時間在思考如何將理論轉化為可驗證的假說,以及如何最有效地驗證這樣的假說。我們將藉由討論組織研究領域中的主要期刊,並讓你使用數據和估計模型加以進行。這些技巧中我主要目標是幫助你能輕易地使用統計方法提出與解決研究問題,並且發展評估他人研究的關鍵性技能,這樣就可以將這些技能應用於自己的研究上。

形式與要求

本課程之架構(大致上)為一週的理論授課與隔週的理論應用相互交替。我們將使用以下兩種方式:

  • 首先,討論使用特定方法的報告(從網路和其他地方取得),為了評估這些數據與方法對於研究問題來說是否得宜。

  • 其次,我會要求學生處理數據與估計模型(使用STATAR),寫下結果並闡明原因,然後(偶爾)在課堂中報告。

數據作業中之文句須得宜。最重要的是這些作業都沒有特定答案。我不會指定你該用什麼相依變數、該解釋什麼…等等。這樣的部份目的是要讓你練習如何使用數據、估計模型和使答案合理化。利用課程15.347所學,你要規劃出數據的概念性模型和統計模型。這門課較大的限制是你只能使用當週課堂中討論的方法(例如:計算模型)。我都已經將適當的數據資料設定好,不過你也可以使用自己取得的數據。

每次「應用」課的開始都要交一份分析結果概要,本文最多兩頁,加上圖表,並根據推衍出的概念性模型討論其結果。這些作業最多可由兩人一組完成。我每週會請兩或三組提出他們的作品結果,並在課堂中討論。

評分分配標準如下:

事項 百分比
課堂參與 30%
指定報告之一的討論領導力 20%
五個含數據工作的簡短作業 50%


軟體

若你沒用過STATAR,我希望你在開課的前幾週作以下事項:

  1. 安裝軟體(比方在計算機實驗室),以便讓你知道如何運作。

  2. 2. 至少先瀏覽使用手冊,特別是閱讀「開始使用STATA®」。

網路上也可取得資源,例如:UCLA提供的STATA®啟動工具。

教科書

Alrich, John H.和Forrest D. Nelson著,《線性機率, 對數轉換值/機率轉換值模型》,Newbury Park,CA:Sage出版,1984,ISBN: 0803921330.

Singer, Judith D.和John B. Willett著,《應用縱向數據分析:模型改變與事件產生》,紐約:牛津大學出版社,2003,ISBN: 0195152964.


Overview

A large proportion of contemporary research on organizations, strategy and management relies on quantitative research methods. This course is designed to provide an introduction to some of the most commonly used quantitative techniques, including logit/probit models, count models, event history models, and pooled cross-section techniques.

This is a course about the research process. My explicit goal is to help you understand the relationship between theory, data and statistical methods. In that sense this is not a course in statistical theory; we will not spend a lot of time deriving likelihood functions, etc., although we will be explicit about the assumptions underlying the statistical procedures. Instead, this is a course in how to use statistical techniques to answer research questions. We will spend considerable time thinking about how theoretical insights can be translated into testable propositions, and how those propositions are best tested. We will do this through discussions of published research from leading journals in organizational research, and by having you work with data and estimate models. My primary goal in these skills is to help you increase your comfort with using statistical methods to ask and answer research questions, and to develop critical skills in evaluating others' research, such that you might apply those skills to your own.

Format and Requirements

The structure of the course involves (roughly) alternating lectures on the principles associated with a particular method in one week, followed the next week by the application of those models. We will pursue two types of application:

  • First, we will discuss working papers (found on the Internet and elsewhere) that use the particular methods in question, with an eye toward assessing whether the data and methods are appropriate for the research question.

  • Second, I will ask students to work with data to estimate models (in STATA®), write up an interpretation of the results, and then (occasionally) present the results in class.

Some words about the data assignments are in order. Most importantly, these assignments are deliberately open-ended. I will not specify what the dependent variable should be, what you are trying to explain, etc. Part of the goal of these exercises is to give you practice with working with data, estimating models and making sense of the results. In the language of 15.347, I want you to formulate a conceptual model of the data as well as a statistical model. The major constraint is that you apply the method discussed during that week's class (e.g., count models). I will make appropriate datasets available for this purpose, but you are also allowed to use your own data sources.

At the beginning of class in each "Application" week, I ask that you hand in a brief summary of the results of your analysis. This summary should be a maximum of two pages of text, plus any tables and figures, and should discuss the results in light of the conceptual model you have developed. These assignments may be done in groups of up to two people. I will ask two or three groups each week to present the results of their work, so that we may discuss them as a class.

The grading is broken down as follows:

ACTIVITIES PERCENTAGES
Class participation 30%
Discussion leadership for one of the assigned working papers 20%
The five short assignments that involve working with data 50%


Software

If you have not worked with STATA® before, I encourage you to do things during the first few weeks of class:

  1. Get yourself set up (e.g., in the computer lab) so that you know how to get it going and

  2. At least look through some of the manuals. It can be particularly helpful to look at "Getting Started with STATA®."

There are also resources available on the Internet, e.g., the STATA® Starter Kit provided through UCLA .

Texts

Aldrich, John H., and Forrest D. Nelson. Linear Probability, Logit and Probit Models. Newbury Park, CA: Sage, 1984. ISBN: 0803921330.

Singer, Judith D., and John B. Willett. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press, 2003. ISBN: 0195152964.


 
MIT Home
Massachusetts Institute of Technology Terms of Use Privacy