• Save
Planning Data Analysis and Completing the Design Matrix 规划数据分析与完成设计矩阵
Upcoming SlideShare
Loading in...5
×
 

Planning Data Analysis and Completing the Design Matrix 规划数据分析与完成设计矩阵

on

  • 3,132 views

Shanghai International Program for Development Evaluation Training Asia-Pacific Finance and Development Center; 200 Panlong Road-Shanghai, October 15, 2008

Shanghai International Program for Development Evaluation Training Asia-Pacific Finance and Development Center; 200 Panlong Road-Shanghai, October 15, 2008

Statistics

Views

Total Views
3,132
Views on SlideShare
3,129
Embed Views
3

Actions

Likes
3
Downloads
0
Comments
1

1 Embed 3

http://www.slideshare.net 3

Accessibility

Upload Details

Uploaded via as Microsoft PowerPoint

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

Planning Data Analysis and Completing the Design Matrix 规划数据分析与完成设计矩阵 Planning Data Analysis and Completing the Design Matrix 规划数据分析与完成设计矩阵 Presentation Transcript

  • Planning Data Analysis and Completing the Design Matrix 规划数据分析与完成设计矩阵 Shanghai International Program for Development Evaluation Training Asia-Pacific Finance and Development Center 200 Panlong Road-Shanghai, October 15, 2008 Ray C. Rist
    • Ray C. Rist
    • Knowledge & Evaluation Capacity Development Advisor, Independent Evaluation Group of the World Bank
    • President International Development Evaluation Association (IDEAS)
    • rrist@worldbank.org;
    • www.worldbank.org/ieg
    Dadang Solihin Indonesia Delegation Shanghai International Program for Development Evaluation Training
  • Introduction 导言
    • Data Analysis Strategy
    • 数据分析战略
    • Analyzing Qualitative Data
    • 分析定性数据
    • Analyzing Quantitative Data
    • 分析定量数据
    • Linking Quantitative Data and Qualitative Data
    • 将定量数据与定性数据联系起来
    www.dadangsolihin.com
  • Data Collection and Analysis 数据收集与分析 花费的时间 www.dadangsolihin.com Time 时间 Hours Spent Data Collection 数据收集 Data Analysis 数据分析
  • Qualitative Analysis 定性分析
    • Best used when for in-depth understanding of the intervention
    • 在深入理解干预时最适合使用
    • Answers questions like:
    • 回答诸如以下问题:
      • What are some of the difficulties faced by staff?
      • 员工面临的困难有哪些?
      • Why do participants say they dropped out early?
      • 为什么一些参与者说他们很早就退出了?
      • What is the experience like for participants?
      • 参与者的感受是怎样的?
    www.dadangsolihin.com
  • Qualitative Analysis (Cont.)
    • Developing the data analysis strategy is an important part of the planning process.
    • It helps to know the options for data analysis, with the various strengths and weaknesses, as you plan your research.
    • This is very important. In the design matrix, specify so that you know how you will use the information collected.
    • A common mistake is collecting vast amounts of data that are never used. Of course, data analysis is important in the "doing.”
    • This overview will provide you with the big picture issues.
    www.dadangsolihin.com
  • Quantitative Analysis 定量分析
    • Can be used to answer questions like?
    • 可用于回答如下问题?
    • What are the mean scores for the different groups of participants?
    • 不同组的参与者的平均分数?
    • How do participants rate the relevance of the intervention on a scale of one to five?
    • 按照 1-5 分的等级,参与者会如何给干预活动的相关性打分?
    • How much variability is there in the responses to the item?
    • 在各项回答中有多少变动性?
    • Are the differences between the two groups statistically significant?
    • 两组之间的差别是否具有统计显著性?
    www.dadangsolihin.com
  • Qualitative Data 定性数据
    • Description of program, process, and experiences
    • 对计划、程序和经验的说明
    • To understand context of the situation
    • 理解形势背景
    • To understand perceptions
    • 理解感受
    • Research evolves as questions emerge
    • 研究随着问题的出现而发展
    • Flexible design
    • 可行的设计
    www.dadangsolihin.com
  • Qualitative Data Analysis 定性数据分析
    • Used for any non-numerical data collected as part of the evaluation
    • 用于作为评价一部分收集的非数字数据
      • unstructured observations
      • 未组织的考察
      • open-ended interviews
      • 开放式会谈
      • analysis of written documents
      • 书面文件的分析
      • focus groups transcripts
      • 专题组讨论笔录
      • diaries, observations
      • 日志,观察
    • Analysis challenging
    • 分析问题
    • Take care for accuracy (validity concern)
    • 力求准确(对有效性的关注)
    www.dadangsolihin.com
  • Making Good Notes 很好地记录
    • Capture as much information as possible
    • 捕捉尽可能多的信息
    • Pay close attention to language
    • 密切关注语言
    • Write down observations
    • 记录下观察
    • Capture your immediate thoughts
    • 捕捉你的直接思路
    • Leave time to write up notes immediately
    • 留出时间来立即做详细书面记录
    www.dadangsolihin.com
  • Making Good Notes (Cont.)
    • When you collect qualitative data, you want to capture as much information as possible. It is important to accurately capture your observations; good notes are essential.
    • This means paying close attention to language: what people say and how they say it.
    • Try not to interpret what they say as you write your notes. Write down anything thing you observed, any body language, or anything that happened while you were collecting data (for example, many interruptions during the interview).
    • You may also want to capture your immediate thoughts, reactions, and interpretations. Keep them in a separate section of your notes.
    www.dadangsolihin.com
  • Making Good Notes (Cont.)
    • As mentioned in Module 9, it is extremely important to leave time soon after an interview, observation, or focus group to review your preliminary notes and make additions, and write up your notes so they will make sense when you look at them later on.
    • It is surprising how difficult it is to make sense of notes taken in an interview, focus group, or observation session – even from just a day earlier.
    • Even if you have tape-recorded the session, a small amount of time invested in a preliminary write-up while it is fresh in your mind will save hours and hours of listening to or watching tapes or poring over transcripts later on.
    www.dadangsolihin.com
  • Triangulation 三元法
    • Can use three or more sources of information to verify and substantiate your data
    • 可以使用三个或更多的信息来源,以检验和证明你的数据
    • Examples:
    • 例子:
      • interviews, focus groups, questionnaires
      • 会谈,专题组讨论,调查问卷
      • questionnaires, available data, expert panels
      • 调查问卷,可利用的数据,专家小组
      • observations, program records, interviews
      • 考察,计划记录,会谈
      • interviews, diaries, available data
      • 会谈,日志,可利用的数据
    www.dadangsolihin.com
  • Triangulation (Cont.)
    • As you probably recall, triangulation is the use of three or more theories, sources or types of information, or types of analysis to verify and substantiate an assessment.
    www.dadangsolihin.com
  • Early Steps in Qualitative Analysis (1 of 3) 定性分析的早期步骤(第 1 页,共 3 页)
    • While collecting data:
    • 在收集数据的同时:
      • keep good records
      • 很好地坚持记录
      • write up interview, impressions, notes from focus groups
      • 详细书面记录下专题组讨论的会谈、印象和笔记
      • make constant comparisons as you progress
      • 在进展的同时不断进行比较
      • meet with team regularly to compare notes and make adjustments
      • 定期与小组会面,比较笔记并进行调整
    www.dadangsolihin.com
  • Early Steps in Qualitative Analysis (2 of 3) 定性分析的早期步骤(第 2 页,共 3 页)
    • Write contact summary report
    • 撰写联系的总结报告
      • one page summary after each major interview or focus group
      • 在每次重大会谈或专题组讨论之后写一页的概述
      • main issues
      • 主要问题
      • major information obtained
      • 获得的主要信息
      • what was the most interesting, illuminating, or important?
      • 最令人注意的、最有启发意义的或最重要的是什么?
      • what new questions need to be explored?
      • 需要探索什么新问题?
    www.dadangsolihin.com
  • Early Steps in Qualitative Analysis (3 of 3) 定性分析的早期步骤(第 3 页,共 3 页)
    • Use tools to help you
    • 利用工具帮助你
      • create a subjectivity file with your own reactions during the study, including your feelings, hunches, and reactions
      • 建立一份关于你自己在研究期间反应的主观性文件,包括你的感受、预感和反应
      • file your ideas that emerge as you proceed
      • 把你在进展过程中出现的想法归档
      • keep a file of quotations from the data collection
      • 坚持对来自数据收集的引用进行归档
    www.dadangsolihin.com
  • Maintain an Iterative Dialogue 坚持反复对话
    • Share information early and often with key informants
    • 早早地共享信息,并且经常是与关键的信息提供者
    • Have others review early drafts with the intention of eliciting information, questions, other ways of interpreting data
    • 让其他人审查初稿,旨在得出信息,问题,解释信息的其它方法
    www.dadangsolihin.com
  • Reading and Coding Data 数据阅读和编码
    • Read all of the data carefully
    • 仔细阅读所有数据
    • Come up with names or labels for topics
    • 给主题命名或贴标签
    • Describe the topics
    • 描述主题
    • Using codes, classify all of the data
    • 使用编码给所有数据分类
    www.dadangsolihin.com
  • Coding 编码
    • Manually coding 人工编码
      • use colored pens, pencils, or papers 使用彩色水笔、铅笔或纸张
      • cut with scissors to manually sort 用剪刀剪裁进行人工分类
      • use a number coding system 使用数字编码系统
    • Computer program coding 计算机程序编码
      • word processors 文字处理器
      • presentation software 演讲软件
      • databases, spreadsheets, or qualitative data analysis software 数据库、电子表格或者定性数据分析软件
    www.dadangsolihin.com
  • Drawing-out Themes and Patterns 得出主题与模式
    • As you review, begin to make notes
    • 在你评价的同时,开始做笔记
    • Goal is to summarize what you have seen or heard:
    • 目标是总结你看到的或听到的
      • common words
      • 共同词语
      • phrases
      • 短语
      • themes
      • 主题
      • patterns
      • 模式
    www.dadangsolihin.com
  • Drawing-out Themes and Patterns (Cont.)
    • Also identify where they are so you can find them again if you need to verify
    • 同样确认它们的位置,这样如果你需要进行验证,能够再次找到它们
    • May want to use a spreadsheet
    • 可能需要使用电子表格
    www.dadangsolihin.com
  • Drawing-out Themes and Patterns (Cont.)
    • As you review the material, you will begin to make notes. It helps to read a few to get a sense of what is there and develop a general framework for analyzing the rest of the data.
    • However, you may discover other themes as you go along and may have to re-read earlier material.
    • Some people find it helpful to use a spreadsheet that identifies the common themes and where they are located in their notes.
    • Other people use note cards to sort through the qualitative data.
    www.dadangsolihin.com
  • Content Analysis 内容分析
    • Identify certain words or concepts in text or speech
    • 确认文本或讲话稿中的某个词语或概念
    • Conceptual analysis 概念分析 :
      • look at word frequencies 关注词语频率
    • Relational analysis: 关联分析
      • look at word frequencies 关注词语频率
      • explore relationships among concepts
      • 探索各种概念之间的关系
    www.dadangsolihin.com
  • Computer Help for Qualitative Data Analysis 计算机有助于定性数据分析
    • Software packages to help you organize data
    • 软件包帮助你组织数据
    • Search, organize, categorize, and annotate textual and visual data
    • 搜索、组织、分类和注释文本和视觉数据
    • Help you visualize the relationships among data
    • 帮助你将数据之间的关系直观化
    www.dadangsolihin.com
  • Computer Help for Qualitative Data Analysis (Cont.)
    • These are text-oriented database managers, word processors, or automatic-indexing software.
    • They are specifically developed for working with text applications.
    • When you enter transcripts into a word processor, they can be organized, indexed, and coded.
    • Qualitative analysis software can be a powerful tool for organizing vast amounts of data produced through focus groups or individual interviews.
    • These software programs are called Computer-Assisted Qualitative Data Analysis Software (CAQDAS) — also sometimes simply called Qualitative Data Analysis Software (QDAS or QDA software.)
    • Programs of this type help you visualize the relationships between and among data and/or theoretical constructs to support theory-building.
    www.dadangsolihin.com
  • Examples of QDA 定性数据分析的例子
    • N6 from QSR( 以前称作 NUD*IST)
    • Ethnograph
    • Qualpro
    • Hyperqual
    • Anthropax
    • Atlas-ti
    www.dadangsolihin.com
  • Controlling for Bias 控制误差
    • We tend to see what we want to see and may miss things that do not conform to our expectations
    • 我们倾向于看到我们希望看到的,可能会错失那些不符合我们期望值的事物
    • Use well trained recorders
    • 很好地使用受过培训的记录员
    • Evaluators review documents and code them in themes
    • 评价者评价文件并把它们按主题编码
    www.dadangsolihin.com
  • Concluding Thoughts on Qualitative Data 定性数据的结论性思考
    • Qualitative data collection is not the easy option
    • 定性数据收集不是一个简单的选择方案
      • labor intensive and time consuming
      • 劳动密集型的和耗时的
      • reliability among coders, using a coding scheme is essential
      • 编码员之间的可靠性,必须使用编码方案
    • Can reveal some of the most valuable information
    • 能够揭示出一些最宝贵的信息
    www.dadangsolihin.com
  • Concluding Thoughts on Qualitative Data (Cont.)
    • An excellent resource with step-by-step guides for how to do systematic qualitative data analysis (descriptive, causal, and other) is Miles and Huberman’s book, Qualitative Data Analysis: An Expanded Sourcebook.
    www.dadangsolihin.com
  • Quantitative Data: Statistics 定量数据:统计
    • Quantitative data are analyzed with statistics
    • 利用统计分析定量数据
      • descriptive statistics: used with census or non-random sample data
      • 描述统计:与统计数字或非随机样本数据一起使用
      • inferential statistics: used with random sample data
      • 推论统计:与随机样本数据一起使用
    www.dadangsolihin.com
  • Quantitative Data: Statistics (Cont.)
    • Typically, descriptive statistics are used to summarize data collected from a sample about a qualitative or qualitative variable, while inferential statistics involve making a prediction about a range of population values for a quantitative or qualitative variable, based on information for that variable from the random sample.
    • Part of the prediction includes a reliability statement which tells the probability that the true population value lies within the range of values.
    www.dadangsolihin.com
  • Descriptive Statistics 描述统计
    • Describes the frequency and/or percentage distribution of a single variable
    • 说明一个单独变量的频率和 / 或百分比分布
    • Tells how many and what percent
    • 说明有多少以及占多大百分比
    • Example:
    • 例子:
      • 33% of the respondents are male and 67% are female (table on next slide)
      • 33% 的被调查人是男性, 67% 是女性(表格见下一张幻灯片)
    www.dadangsolihin.com
  • Example of Descriptive Statistics in a Table 表格描述统计的例子 Write up: Of the 300 people in this program, 67% are women and 33% are men. 书面记录:在参与该计划的 300 人中, 67% 为女性, 33% 为男性。 www.dadangsolihin.com Male 男性 Female 女性 Total 总计 Number 人数 Percent 百分比 Number 人数 Percent 百分比 Number 人数 100 33% 200 67% 300 How many men and women are in the program? Table 11.5: Distribution of Respondents by Gender 该计划里有多少男性和女性? 表 11.5 :受调查者的性别分布 Source: Fabricated Data
  • Distributions 分布
    • Measures of central tendency
    • 集中趋势测量
      • how similar are the data?
      • 数据在多大程度上类似?
      • example: How similar are the ages of this group of people?
      • 例如:这一群人的年龄在多大程度上类似?
    • Measures of dispersion
    • 离中趋势测量
      • how dissimilar are the data?
      • 数据在多大程度上不一样?
      • example: How much variation in the ages?
      • 例如:年龄差异有多大?
    www.dadangsolihin.com
  • Measures of Central Tendency 集中趋势测量
    • The 3-M ’ s
    • 3M
      • mode: most frequent response
      • 众数:出现最频繁的响应值
      • median: mid-point of the distribution
      • 中位数:分布的中点
      • mean: arithmetic average
      • 平均数:算术平均数
    • Which to use depends on the type of data you have
    • 使用哪一个取决于你的数据类型
      • nominal, ordinal, interval/ratio
      • 名目,序数,间隔 / 比率
    www.dadangsolihin.com
  • Nominal Data 名目数据
    • Data of names or categories
    • 名称或类别数据
    • Examples:
    • 例如
      • gender (male, female)
      • 性别(男性,女性)
      • religion (Buddhist, Christian, Jewish, Muslim)
      • 宗教(佛教,基督教,犹太教,穆斯林)
      • country of origin (Burma, China, Ethiopia, Peru)
      • 原产国(缅甸,中国,埃塞俄比亚,秘鲁)
    • Use mode as a measure of central tendency
    • 使用 众数 作为集中趋势的度量
    www.dadangsolihin.com
  • Ordinal Data 序数数据
    • Data that has an order to it but the “ distance ” between consecutive responses is not necessarily the same
    • 有一定顺序但是连续响应值之间的 “ 距离 ” 不一定相同的数据
    • Lacks a zero point
    • 缺少零点
    • Examples:
    • 例子
      • opinion scales that go from “ most important ” to “ least important ” or “ strongly agree ” to “ strongly disagree ”
      • 从 “ 最重要 ” 到 “ 最不重要 ” 或者从 “ 强烈同意 ” 到 “ 强烈不同意 ” 的评价尺度
    • Use mode or median as a measure of central tendency
    • 使用 众数 或 中位数 作为集中趋势的度量
    www.dadangsolihin.com
  • Interval/Ratio Data 间隔 / 比率数据
    • Data of real numbers, numbers with a zero point and can be divided and compared into other ratio numbers
    • 实数数据,数字有零点并且可以被分割或比较成其它比率数字
    • Examples:
    • 例子:
      • age, income, weight, height
      • 年龄,收入,体重,身高
    www.dadangsolihin.com
  • Interval/Ratio Data ( Cont.)
    • Use mode , median , or mean as a measure of central tendency — the choice depends on the distribution
    • 使用 众数 , 中位数 或 平均数 作为集中趋势的度量 —— 选择取决于分布
      • for normal data, mean is best
      • 对于标准数据来说, 平均数 最适合
      • for data with few high – or - few low scores, median is best
      • 对于只有少数很高或少数很低值的数据来说, 中位数 最适合
    www.dadangsolihin.com
  • Calculating 计算
    • Mode: the one with the most
    • 众数:出现最频繁的值
    • Median: place in order then count down to half way
    • 中位数:按顺序放置,然后数到中间
    • Mean: (most people think of it as the average)
    • 平均数:(大多数人认为它是平均值)
    www.dadangsolihin.com
  • Calculating (Cont.)
    • The median is defined as the value of the variable where half of the observations are above the value and half are below the value. In even numbered data sets, there will not be a single case, which divides the sample in half. In those situations, the median is defined as the average of the two middle cases (some of the two middle cases, divided by 2.)
    • The mean is defined as the sum of the observations of the variable (X) divided by then number of observations (n):
    • mean =  (X i )  n
    • The mean is what most people think of as the average of the variable.
    www.dadangsolihin.com
  • Example Data 数据例子 Table 11.7: Sample Data 表 11.7 : 样本数据 www.dadangsolihin.com Country 国家 % Urban 城市 Bolivia 玻利维亚 65 Algeria 阿尔及利亚 60 Central Africa Republic 中非共和国 41 Georgia 格鲁吉亚 61 Panama 巴拿马 58 Turkey 土耳其 75 Source: Fabricated Data, 资料来源:制作好的数据
  • Example Calculations for % Urban Data 计算城市百分比数据的例子
    • Mode: no mode, all have only one
    • 众数:没有众数,所有都只出现过一次
    • Median: total entries is 6, with data in order two middle scores are (61 and 60) ÷ 2 = 60.5
    • 中位数:共有 6 条,按顺序放置数据,中间两个值是( 61 和 60 ) ÷ 2 = 60.5
    • Mean:
    • 平均数:
      • (65+60+41+61+58+75) ÷6 = 60
    www.dadangsolihin.com
  • Example Calculations for % Urban Data 计算城市百分比数据的例子
    • Notice that the mean is greatly affected by extreme values in the sample, while the median is not. Suppose the urban percentage for Turkey had been 87 (instead of 75.) The mean would increase to 62, but the median would be unaffected.
    • For this reason, the median is the preferred summary measure for variables that are skewed to extremely high or low values (median income usually gives a clearer picture of the center of the income distribution than the mean, because income is widely skewed for many countries).
    • On the other hand, it can be argued that the median wastes information (in not being affected by extremes.)
    • For that reason, the sample mean might be considered a better predictor of the center of the population, than the sample median.
    www.dadangsolihin.com
  • Measures of Dispersion 离中趋势测量
    • Range
    • 值域
      • difference between the highest and lowest value
      • 最高值与最低值之间的差
      • simple to calculate, but not very valuable
      • 计算简单,但是不太有价值
    • Standard deviation
    • 标准差
      • measure of the spread of the scores around the mean
      • 数值围绕平均数分布情况的度量
      • superior measure, it allows every case to have an impact on its value
      • 优秀的度量,它让每个案例都对它的值产生影响
    www.dadangsolihin.com
  • Measures of Dispersion (Cont.)
    • The standard deviation is a measure of the spread of the scores around the mean.
    • The more the scores differ from the mean, the larger the standard deviation will be.
    • If everyone scored 75 on a test, the standard deviation would be 0.
    • If everyone scores between 70-80 (mean 75), the standard deviation would be smaller then if everyone scored between 40-90 (mean 75). Put another way:
    • Small standard deviation = not much dispersion.
    • Large standard deviation = lots of dispersion.
    www.dadangsolihin.com
  • Example Calculation for Range 计算值域的例子
    • Range: high score – low score = range
    • 值域:最高值 — 最低值 = 值域
      • range = 75 – 41
      • 值域 =75 — 41
      • range = 34
      • 值域 =34
    www.dadangsolihin.com
  • Normal Curve (Bell) 正常曲线(钟形) 频率 www.dadangsolihin.com y x 0 Frequency Value 值
  • Standard Deviation 标准差 www.dadangsolihin.com y x 0 Mean 平均数 One standard deviation from the mean 与平均数相隔一个标准差 68% 95% Two standard deviations from the mean 与平均数相隔两个标准差 98% Three standard deviations from the mean 与平均数相隔三个标准差
  • Calculating Standard Deviation 计算标准差
    • Calculating is time consuming
    • 计算非常耗时
    • Can use statistical programs:
    • 可以使用统计程序:
      • SPSS
      • SPSS
      • Excel or other spreadsheet program
      • Excel 或其它电子表格程序
    www.dadangsolihin.com
  • Guidelines for Analyzing Quantitative Survey Results 分析定量调查结果的指导原则 www.dadangsolihin.com 1 Choose a standard way to analyze the data and apply it consistently 选择一个分析数据的标准方法并始终如一地应用它 2 Do not combine the middle category with each side of the scale 不要把尺度各边与中间类别混合起来 3 Do not report an agree or disagree category without also reporting the strongly agree agree or strongly disagree category 不要在不报告强烈同意、同意或强烈不同意类别的情况下报告同意或不同意类别 4 Analyze and report percentages (or numbers) 分析并报告百分比(或数字) 5 Provide the number of respondents for an anchor 提供被调查人数量,用作支撑点 6 If there is little difference in the data, raise the benchmark 如果数据几乎没有差异,提出基准点 7 Like any art or skill, it gets easier with training and practice 和任何艺术或技能一样,通过培训和练习,它会越来越容易
    • Non-open-ended survey results can be reported in terms of percent answering (52% women, 48% men). Sometimes, the questions ask for specific counts (“Were you employed in the past week?” or “How many goats do you own?” These would also be reported in terms of percents, absolute numbers, or both.
    • Other times, people are asked to give opinions along a scale. For example, one may ask whether the respondents have been able to apply what they have learned, and are given a scale ranging from “Not at all” to “A lot.” This is a five-point scale. When analyzing this type of data, you would want to establish a decision rule: will you focus on the percent who answered at the extreme ends of the scale, or will you focus on those who answered on either side of the middle category? Some guidelines might be helpful but there are no firm rules here
    Guidelines for Analyzing Quantitative Survey Results (Cont.) www.dadangsolihin.com
  • Common Descriptive Statistics 共同描述统计
    • Frequencies
    • 频率
    • Percent
    • 百分比
    • Mean
    • 平均数
    • Median
    • 中间数
    • Mode
    • 众数
    • Money
    • 金钱
    • Percent
    • 百分比
    • Ratio
    • 比率
    • Comparisons
    • 比较
    www.dadangsolihin.com
  • Describing Two Variables at the Same Time 同时描述两个变量
    • Two variables at once
    • 一次两个变量
    • Example: What percent were boys and what percent were girls in hands-on and traditional classes?
    • 例如:在实践班和传统班里,男孩的百分比是多少,女孩的百分比是多少?
    www.dadangsolihin.com
  • Describing Two Variables at the Same Time 同时描述两个变量
    • Sometimes you want to describe two variables at the same time.
    • For example, suppose you want to describe the composition of the hands-on and lecture classes.
    • For each class, you want to know what percent were boys and what percent were girls.
    • Analysis of the data shows that the hands-on classes comprises 55% boys and 45% girls, while the traditional lecture classes comprises 55% girls and 45% boys
    www.dadangsolihin.com
  • Example Two Variables at the Same Time 同时描述两个变量的例子 Source: Fabricated Data: 2004 Survey 资料来源:已制作好的数据: 2004 年调查 www.dadangsolihin.com Hands-on 实践 Hands-on 实践 Traditional 传统 Traditional 传统 Boys 男孩 28 55% 34 45% Girls 女孩 22 45% 41 55% Total 125 总数 125 N=50 100% N=75 100%
  • Two Variables with Crosstabs 交叉表的两个变量
    • Cross tabulation (crosstab)
    • 交叉表
      • presented as a contingency table in a matrix format
      • 在矩阵格式里作为列联表显示
      • displays two or more variables simultaneously
      • 同时显示两个或更多个变量
      • each cell shows number of respondents
      • 各个单元显示出被调查人的数量
    www.dadangsolihin.com
  • Two Variables with Crosstabs 交叉表的两个变量
    • A cross tabulation (often abbreviated as crosstab ) displays the joint distribution of two or more variables Crosstabs are usually presented as a contingency table in a matrix format.
    • Whereas a frequency distribution provides the distribution of one variable, a contingency table describes the distribution of two or more variables simultaneously.
    • Each cell shows the number of respondents that gave a specific combination of responses, that is, each cell contains a single cross tabulation.
    • Crosstabs are used when you are working with nominal and ordinal data, or when you have categorized interval/ratio data.
    www.dadangsolihin.com
  • Example Crosstabs 交叉表的例子 N=125 Source: Fabricated Data N=125 资料来源:虚构数据 www.dadangsolihin.com Hands-on 实践 Traditional 传统 Total % 总计 % Boys 男孩 45% 55% 100% Girls 女孩 35% 65% 100%
  • Variables 变量
    • Independent
    • 自变量
      • Variable which you believe explains a change in the dependent variable
      • 你认为说明因变量变化的变量
      • Program evaluation: the program
      • 程序评价:程序
    • Dependent
    • 因变量
      • Variable you want to explain
      • 你希望说明的变量
      • Program evaluation: the outcomes
      • 程序评价:结果
    www.dadangsolihin.com
  • Example: Comparison of Means 例子:平均数的比较 - dependent variable: annual income - 因变量:年收入 -independent variable: gender - 自变量:性别 www.dadangsolihin.com Mean Income 平均收入 Women 女性 27,800 SA Rand 27,800 南非南特 Men 男性 32,400 SA Rand 32,400 南非南特
  • Measure of Relationship 关系的度量
    • How strongly variables are related, reported differently
    • 变量之间的关系有多强,不同的报告
    • Measures of association
    • 相联度量
      • range from zero to 1
      • 值域从 0 到 1
    • Measures of correlation
    • 相关性度量
      • range from – 1 to +1
      • 值域从 -1 到 +1
    www.dadangsolihin.com
  • Interpretation of Correlation 对相关性的解释
    • Measures of correlation:
    • 相关性的度量
      • perfect relationship: 1 or – 1
      • 完美关系: 1 或 -1
        • closer to 1 or – 1: strong relationship
        • 更接近于 1 或 -1 :强有力的关系
        • .5: moderate/strong (maybe as good as it gets)
        • .5 :适中 / 强烈(可能更高)
      • closer to zero: no relationship
      • 更接近 0 :没有关系
        • .2 - slight/weak relationship
        • .2 —— 轻微 / 微弱的关系
    www.dadangsolihin.com
  • Direct Relationship 直接相关
    • Plus sign +
    • 加号 +
      • both variables change in the same direction
      • 两个变量都向着同一个方向变化
      • example:
      • 例子
        • as driving speed increases, death rate goes up
        • 随着车速的增加,死亡率也上升
    www.dadangsolihin.com
  • Inverse Relationship 逆相关
    • Minus sign —
    • 减号 —
      • both variable change but in the opposite direction
      • 两个变量向着相反的方向变化
      • example:
      • 例子:
        • as age increases, health status decreases
        • 随着年龄的增加,健康状态下降
    www.dadangsolihin.com
  • Inferential Statistics 推论统计
    • Used to analyze data from randomly selected samples
    • 用来分析随机选择的样本数据
    • Risk of error because your sample may be different from the population as a whole
    • 因为你的样本可能与总体人群不同,存在着误差风险
    • To make an inference, you first need to estimate the probability of that error
    • 为得出推论,你首先需要估计这一误差的概率
    www.dadangsolihin.com
  • Statistical Significance Tests 统计显著性检测
    • Tools to estimate how likely the results are in error
    • 估计结果出现误差可能性的工具
    • Called tests of statistical significance
    • 称为统计显著性检测
      • to estimate how likely it is that you have gotten the results you see in you analysis by chance alone
      • 估计你纯属 偶然 得出你的分析结果的可能性
    www.dadangsolihin.com
  • Statistical Significance 统计显著性
    • Benchmark of .5%
    • .5% 为基准点
      • .05 Alpha level or P value
      • .05 Alpha 水平或 P 值
    • It means we are 95% certain that our sample results are not due to chance
    • 它意味着我们 95% 地肯定我们的样本结果不是偶然得出的
      • Or
      • 或者
    • The results are statistically significant at the .05 level
    • 结果大于 .05 则具备统计显著性
    • Most reports do not go beyond .5
    • 大多数报告不超过 .5
    www.dadangsolihin.com
  • Statistical Significance (Cont.)
    • Statistical tests come in 100+ varieties. You may have heard of some of the more common statistical tests, such as Chi Square and the t-test.
    • The good news is that all the different statistical tests are interpreted using the same guidelines. Evaluators typically set the benchmark for statistical significance at the .05 level. This is sometimes called the alpha level or the p value.
    • That is, we set the benchmark so that we are at least 95% certain that the sample results are not the result of random chance. If we want to raise the bar, we would set the level at .01 to be 99% certain that the sample results are not due to chance alone.
    www.dadangsolihin.com
  • Chi Square and t-Test 卡方和 t 检验
    • One of the most popular statistics
    • 最流行的统计法之一
      • easy to calculate and interpret
      • 易于计算和解释
    • Used to compare two sets of nominal data (i.e marital status and religious affiliation)
    • 用来比较两组名目数据(即,婚姻状况和宗教关系)
    • Used to compare two ordinal variables or a combination of nominal and ordinal variables
    • 用来比较两个序数变量或名目变量与序数变量的结合
    • Used to determine if one group of numerical scores is statistically higher or lower than another group of scores
    • 用来决定一组数值是否在统计学上比另一组数值更高或更低
      • two means
      • 两个平均数
      • post-test only, two group randomized design
      • 仅进行事后检验,两组随机设计
    www.dadangsolihin.com Chi Square 卡方 t-Test T 检验
  • Chi Square and t-Test 卡方和 t 检验
    • The chi square statistic is the sum of the contributions from each of the individual cells in a contingency table. Every cell in the table contributes something to the overall chi square statistic. If a given cell differs markedly from the expected frequency, then the contribution of that cell to the overall chi square is large. If a cell is close to the expected frequency for that cell, then the contribution of that cell to the overall chi square is low. A large chi square statistic indicates that somewhere in the table, the observed frequencies differ markedly from the expected frequencies. It does not tell which cell (or cells) are causing the high chi-square, only that they are there. When a chi-square is high, you must visually examine the table to determine which cell(s) are responsible. [1]
    • The statistic is computed using this formula: where O are the observed cell counts and E are the expected cell counts.
    • [1] StatPac the Statistics Calculator. http://www.statpac.com/statistics-calculator/counts.htm
    www.dadangsolihin.com
  • Hypothesis Testing 假设检验
    • Research hypothesis is your best guess as to the relationship between variables
    • 研究假设 是你关于变量关系的最佳预测
      • Example: there is a difference between the per capita incomes of men and women in South Africa
      • 例子:南非男性和女性之间在人均收入上存在差异
    • Null hypothesis is always a statement that “there is no difference” or “no impact” between our variables
    • 零假设 永远是变量之间“不存在差异”或“无影响”的陈述
      • Example: there is no difference between the per capita incomes of mean and women in South Africa
      • 例子:南非男性和女性之间在人均收入上不存在差异
    www.dadangsolihin.com
  • Testing for Significant Difference 显著性差异检验
    • What is the probability of getting a R4,600 difference if we assume there is no difference in the population from which this sample was drawn?
    • 如果我们假设抽样人群不存在任何差异,得出域值 4600 差数的概率是多少?
    • If the probability is “ small ” for getting a R4,650 difference, then we reject the null hypothesis
    • 如果得出域值 4650 差数的概率 “ 小 ” ,则我们放弃零假设。
      • Small is defined at less than .05
      • 小被定义为少于 .05
    • If the probability is ” high ” then reject the null hypothesis
    • 如果概率 “ 高 ” ,则放弃零假设。
    www.dadangsolihin.com
  • Remember: 记住
    • A significant test is nothing more than an estimate of the probability of getting the results by chance if there really is no difference in the population
    • 显著性检验无非是如果人群之中实际上不存在差异,对偶然得出结果的概率进行的估计。
    www.dadangsolihin.com
  • Linking Qualitative Data and Quantitative Data 将定性数据与定量数据联系起来
    • Should qualitative and quantitative data and associated methods be linked during study design?
    • 在研究设计中是否应该将定性和定量数据以及相关方法联系起来?
      • How?
      • 怎么做?
      • Why?
      • 为什么?
    www.dadangsolihin.com
  • Qualitative-Quantitative Linkages 定性 - 定量联系
    • Confirmation or corroboration – triangulation
    • 证实或确证 —— 三元法
    • Richer detail
    • 更丰富的细节
    • Initiate new lines of thinking
    • 开启新的思路
    • Expand the scope
    • 扩大范围
    www.dadangsolihin.com
  • Thank you! 谢谢!
    • Beside working as Assistant Professor at Graduate School of Asia-Pacific Studies, Waseda University, Tokyo, Japan, he also active as Associate Professor at University of Darma Persada, Jakarta, Indonesia.
    • He got various training around the globe, included Shanghai International Program for Development Evaluation Training (2008) , Public Officials Capacity Building Training Program for Government Innovation, Seoul –Korea (2007), Advanced International Training Programme of Information Technology Management, at Karlstad City, Sweden (2005); the Training Seminar on Land Use and Management, Taiwan (2004); Developing Multimedia Applications for Managers, Kuala Lumpur, Malaysia (2003); Applied Policy Development Training, Vancouver, Canada (2002); Local Government Administration Training Course, Hiroshima, Japan (2001); and Regional Development and Planning Training Course, Sapporo, Japan (1999). He published more than five books regarding local autonomous.
    • You can reach Dadang Solihin by email at [email_address] or by his mobile at +62812 932 2202
    Dadang Solihin currently is Director for Regional D evelopment Performance Evaluation at Indonesian National Development Planning Agency (Bappenas). He holds MA degree in Economics from University of Colorado, USA. His previous post is Director for System and Reporting of Development Performance Evaluation at Bappenas. Dadang Solihin’s Profile www.dadangsolihin.com