PSY2013 Lecture 05 Notes
PSY2013 Research Methods in Psychology
Ch07 Experimental Design I: Single‐Factor Designs
Ch08 Experimental Design II: Factorial Designs
Single-Factor Designs 单因素设计
Learning Objectives 学习目标
Identify and understand the defining features of the four varieties of single-factor designs
识别并理解四种单因素设计的定义特征- Independent groups 独立组设计
- Matched groups 配对组设计
- Nonequivalent groups 不等组设计
- Repeated measures 重复测量设计
Describe two reasons for using more than two levels of an independent variable
描述使用多于两个自变量水平的两个原因Decide when to use a bar graph and when to use a line graph
决定何时使用柱状图,何时使用折线图Describe the goals, methodology, and results of Ebbinghaus memory research
描述艾宾浩斯记忆研究的目标、方法和结果Understand the logic behind the use of three special types of control groups
理解使用三种特殊类型对照组的逻辑- Placebo 安慰剂组
- Wait list 等候名单组
- Yoked 依附组
Understand the ethical issues involved in using certain types of control groups
理解使用某些类型对照组的伦理问题Know when to use an independent samples t-test and when to use a dependent samples t-test
知道何时使用独立样本t检验,何时使用配对样本t检验Understand why a one-way ANOVA is appropriate for single-factor, multilevel studies
理解为何单因素多水平研究应使用单因素方差分析Understand why post hoc statistical analyses typically accompany one-factor ANOVAs
理解为何单因素方差分析通常伴随事后统计分析
Single-Factor Designs 单因素设计
Decision Tree 决策树
Single-Factor—Two Levels 单因素双水平设计
- Between-subjects, single factor designs 被试间单因素设计
Independent Groups Designs 独立组设计
- Manipulated independent variable 操作自变量
- Random assignment to create equivalent groups 随机分配以创建等价组
- Example 例子:
- IV → Note-taking method 记笔记方法
- Laptop note-taking 使用笔记本电脑记笔记
- Handwritten note-taking 手写记笔记
- DV → Performance on memory test 记忆测试表现
- Other concepts 其他概念:
- Conceptual replication 概念复制
- What’s next thinking 接下来是什么
- Ecological validity 生态效度
- IV → Note-taking method 记笔记方法
Matched Groups Designs 配对组设计
- Manipulated independent variable 操作自变量
- Matching to produce equivalent groups 配对以产生等价组
- Example 例子:
- IV → Type of social skills training 社交技能训练类型
- Direct teaching 直接教学
- Play activities 玩耍活动
- Matching variable 配对变量 → Autism Quotient 自闭症商数
- DV → Social Interaction Observation Code 社交互动观察代码
- Other concepts 其他概念:
- Operational definitions 操作定义
- Double-blind procedure 双盲程序
- Inter-rater reliability 评估者间信度
- IV → Type of social skills training 社交技能训练类型
Ex Post Facto Designs 事后设计
- Subject variable as an independent variable 使用受试者变量作为自变量
- Deliberate attempts to select participants to reduce nonequivalence 有意选择参与者以减少不等性
- Example 例子:
- IV → Whether or not traumatic brain injury (TBI) had occurred 是否发生了脑外伤
- Experimental group 实验组 → Had experienced TBI 发生过脑外伤
- Control group 对照组 → No TBI 未发生脑外伤
- DV → Ability to detect insincerity in others 检测他人虚伪的能力
- Other concepts 其他概念:
- Matching 配对
- External validity addressed 外部效度
- IV → Whether or not traumatic brain injury (TBI) had occurred 是否发生了脑外伤
Within-subjects, single factor designs 被试内单因素设计
- Also called repeated measures designs 也称为重复测量设计
- Example 例子:
- IV → Whether or not you share your experience with another person 是否与他人分享你的经历
- Shared 分享
- Unshared 不分享
- DV → Ratings of liking and of flavorfulness of chocolate 对巧克力的喜好和味道评分
- Other concepts 其他概念:
- Confederate 共谋者
- Cover story 掩饰故事
- What’s next thinking 接下来是什么
- IV → Whether or not you share your experience with another person 是否与他人分享你的经历
Single-Factor—More Than Two Levels 单因素多水平设计
- Between-subjects, multilevel designs 被试间多水平设计
Advantages 优势:
- Ability to discover nonlinear effects 发现非线性效应的能力
- Ability to rule out alternative explanations 排除其他解释的能力
Example 例子:
- Bransford and Johnson’s (1972) ‘Laundry Study’ 布兰斯福德和约翰逊(1972)的‘洗衣研究’
- IV → Context 情境
- No context (no topic presented) 无情境(未呈现主题)
- Context before (topic presented before reading paragraph) 阅读段落前呈现主题
- Context after (topic presented after reading paragraph) 阅读段落后呈现主题
- DV → Recall of paragraph’s ideas 回忆段落的想法
- IV → Context 情境
- Bransford and Johnson’s (1972) ‘Laundry Study’ 布兰斯福德和约翰逊(1972)的‘洗衣研究’
Multilevel Independent Groups Design 多水平独立组设计
- Example 例子:
- IV → Number of people with children 与儿童在一起的人数
- Alone 独自
- Bystander (with 2 other children who could help) 旁观者(与另外两个可以帮助的儿童在一起)
- Bystander-Unavailable (with 2 other children who could not help) 旁观者不可用(与另外两个不能帮助的儿童在一起)
- DV → Whether or not child (participant) helped teacher 儿童(参与者)是否帮助老师
- Other concepts 其他概念:
- Operational definitions 操作定义
- Confederates 共谋者
- Inter-rater reliability 评估者间信度
- IV → Number of people with children 与儿童在一起的人数
- Example 例子:
Within-subjects, multilevel designs 被试内多水平设计
- Example 例子:
- Multilevel repeated measures 多水平重复测量
- IV → Listening experience 听觉体验
- Listening to Mozart 听莫扎特
- Listening to a rainstorm 听暴雨
- Control – no listening 对照组 - 不听
- DV → Recall of digits 回忆数字
- Other concepts 其他概念:
- Counterbalancing via 3x3 Latin square 通过3x3拉丁方平衡
- Cover story 掩饰故事
- Example 例子:
Analyzing Data from Single-Factor Designs 单因素设计的数据分析
Presenting the Data 数据展示
- Bar Graphs vs. Line Graphs 柱状图 vs. 折线图
- Bar graphs are used for discrete variables. 柱状图用于离散变量。
- Line graphs are used for continuous variables. 折线图用于连续变量。
- Example 例子: Bransford and Johnson’s (1972) data presented in table and graphical forms 布兰斯福德和约翰逊(1972)数据的表格和图形展示
Analyzing Single-Factor, Two-Level Designs 单因素双水平设计的数据分析
t-Test Assumptions t检验的假设
- Interval or ratio scale data 间隔或比率尺度数据
- Data normally distributed 数据近似正态分布
- Homogeneity of variance 方差齐性
t-Test for Independent Samples 独立样本t检验
- Used for independent groups designs and nonequivalent groups designs 用于独立组设计和不等组设计
t-Test for Related Samples 相关样本t检验
- Used for matched groups designs and repeated measures designs 用于配对组设计和重复测量设计
Analyzing Single-Factor, Multilevel Designs 单因素多水平设计的数据分析
- One-Way ANOVA 单因素方差分析
- Multiple t-tests are inappropriate as they increase the chances of Type I error. 多次t检验不合适,因为它们增加了I型错误的机会。
- One-Way ANOVA for Independent Groups 独立组单因素方差分析
- For multilevel independent groups designs and multilevel ex post facto designs 用于多水平独立组设计和多水平事后设计
- One-Way ANOVA for Repeated Measures 重复测量单因素方差分析
- For multilevel matched groups designs and multilevel repeated measures designs 用
于多水平配对组设计和多水平重复测量设计
- Once an overall significant effect is found, post hoc testing is conducted. 一旦发现总体显著效应,就进行事后检验。
Special-Purpose Control Group Designs 特殊用途的对照组设计
Types of Control Groups 对照组类型
Placebo Control Groups 安慰剂对照组
- Participants believe they are receiving treatment but are not. 参与者认为自己正在接受治疗但实际上没有。
- Placebo effect is when subjects improve due to belief in the treatment. 安慰剂效应是指受试者因相信治疗而有所改善。
- Double-blind procedures are often used to prevent bias. 通常使用双盲程序以防止偏见。
Wait List Control Groups 等候名单对照组
- Participants are placed on a wait list and receive treatment after the experimental group. 参与者被列入等候名单,在实验组之后接受治疗。
- Ensures equivalent groups in program effectiveness studies. 确保在项目效果研究中组等价。
Yoked Control Groups 依附对照组
- Each control group subject is matched to an experimental group subject in terms of treatment or conditions. 每个对照组受试者在治疗或条件上与实验组受试者相匹配。
Expanded Notes 扩展笔记:
- Ethics of Placebo Control Groups 安慰剂对照组的伦理
- Placebo-controlled experiments are unethical if an effective therapy is available. 如果有有效疗法,安慰剂对照实验是不道德的。
- Informed consent is required, and the benefits must be weighed against the risks. 需要知情同意,并且必须权衡收益与风险。
Factorial Designs 因素设计
Learning Objectives 学习目标
Describe factorial designs using a standardized notation system (2x2, 3x5, etc.)
使用标准记号系统描述因素设计(2x2,3x5等)Place data accurately into a factorial matrix and calculate row and column means
准确地将数据放入因素矩阵并计算行和列的平均值Understand what is meant by a main effect and know how to determine if one exists
理解主效应的含义并知道如何确定是否存在Understand what is meant by an interaction effect and know how to determine if one exists
理解交互作用的含义并知道如何确定是否存在Know how to interpret interactions and know that the presence of an interaction sometimes lessens or eliminates the relevance of a main effect
知道如何解释交互作用,并且知道交互作用的存在有时会减弱或消除主效应的相关性Describe the research design of Jenkins and Dallenbach’s (1924) famous study on sleep and memory and understand why their results could be considered an interaction
描述Jenkins和Dallenbach(1924)关于睡眠和记忆的著名研究设计,并理解为何他们的结果可以被认为是交互作用
Essentials of Factorial Designs 因素设计的基础知识
Overview 概述
- Factorial Design 因素设计
- More than one independent variable (IV). 多个自变量。
- Factors 因素: Refers to the independent variables. 指自变量。
- Notation System 记号系统:
- Example 例子: 2 × 3 factorial (two IVs, one with 2 levels, one with 3 levels = 6 total conditions). 2 × 3 因素设计(两个自变量,一个有2个水平,一个有3个水平 = 总共6个条件)
- Example 例子: 2 × 4 × 4 factorial (three IVs, one with 2 levels, two with 4 levels = 32 total conditions). 2 × 4 × 4 因素设计(三个自变量,一个有2个水平,两个有4个水平 = 总共32个条件)
Main Effects and Interactions 主效应与交互作用
Main Effects 主效应
- The overall effect of an IV, averaging over the levels of the other IV(s). 自变量的总体效果,平均其他自变量的水平。
- Example 例子: Imagery vs. rote learning, presentation rate (2-sec vs. 4-sec). 想象 vs. 死记硬背,展示速度(2秒 vs. 4秒)。
- Calculation 计算:
- Compare row means 比较行平均值
- Compare column means 比较列平均值
- Hypothetical Data 示例数据:
- Row mean #1 (imagery) = 20 行均值#1(想象)= 20
- Row mean #2 (rote) = 15 行均值#2(死记硬背)= 15
- Column mean #1 (2-sec) = 14.5 列均值#1(2秒)= 14.5
- Column mean #2 (4-sec) = 20.5 列均值#2(4秒)= 20.5
Interactions 交互作用
- The effect of one IV depends on the level of another IV. 一个自变量的效果取决于另一个自变量的水平。
- Example 例子: Context-dependent memory (study conditions: silent vs. noisy, test conditions: silent vs. noisy). 情境依赖记忆(学习条件:安静 vs. 嘈杂,测试条件:安静 vs. 嘈杂)。
- Identifying Interactions 识别交互作用:
- If the lines in a graph are parallel, there is no interaction. 如果图中的线是平行的,则没有交互作用。
- If the lines are nonparallel, an interaction likely exists. 如果线不平行,则可能存在交互作用。
Varieties of Factorial Designs 因素设计的种类
Types 类型
Mixed Factorial Designs 混合因素设计
- At least one IV is a between-subjects factor and at least one IV is a within-subjects factor. 至少一个自变量是被试间因素,至少一个自变量是被试内因素。
- Example 例子:
- 2x3 Mixed Factorial Design 2x3 混合因素设计
- IV1: Salience of instructions (mortality vs. exam) 指示的显著性(死亡 vs. 考试)(between-subjects 被试间)
- IV2: Leadership style of political candidates (charismatic, task-oriented, relationship-oriented) 政治候选人的领导风格(有魅力的,任务导向的,关系导向的)(within-subjects 被试内)
- DV: Evaluation of candidates 候选人的评价
- Results 结果: Main effect of leadership style and an interaction. 领导风格的主效应和交互作用。
- 2x3 Mixed Factorial Design 2x3 混合因素设计
P × E Designs P × E 设计
- Factorials with both subject and manipulated variables. 既有受试者变量又有操控变量的因素设计。
- P = Person Factor 受试者因素
- E = Environmental Factor 环境因素
- Example 例子:
- 2x2 P×E Factorial Design 2x2 P × E 因素设计
- IV1: Gender of subject (men vs. women) 受试者性别(男 vs. 女)
- IV2: Whether subject took math test with same-sex or opposite-sex group 受试者是否在同性或异性组中参加数学测试
- DV: Math scores 数学成绩
- Results 结果: Women performed poorly only when in a room with men. 女性只有在与男性在同一个房间时表现差。
- 2x2 P×E Factorial Design 2x2 P × E 因素设计
Analyzing Factorial Designs 因素设计的数据分析
Factorial ANOVAs 因素方差分析
- Origins of Factorial Designs 因素设计的起源
- Fisher’s agricultural research 费雪的农业研究
- Looking for interactions between such factors as fertilizer type and wheat strain 寻找肥料类型和小麦品种等因素之间的交互作用
- Invented ANOVA 发明了方差分析
The Ethical Researcher 道德研究者
On Being a Competent and Ethical Researcher 成为一名有能力和道德的研究者
- Guidelines 指导方针
- Be organized, congenial, and welcoming 组织良好、和蔼可亲、热情欢迎
- Take consent seriously 认真对待同意书
- Write out instructions 写出指示
- Practice first, with friends 先与朋友练习
- Be alert to distress 注意受试者的痛苦
- Be careful and thorough in debriefing 小心且彻底地进行事后解释
- Remind departing subjects about contact information 提醒离开的受试者关于联系方式
- Thank
them for their time 感谢他们的时间
Summary 总结
Single-factor designs have one independent variable (factor), which can be either a between-subjects or within-subjects factor
单因素设计有一个自变量(因素),可以是被试间因素或被试内因素。Independent groups and ex post facto designs are between-subjects designs and repeated measures are within-subjects designs
独立组和事后设计是被试间设计,重复测量是被试内设计。Single-factor designs with more than two levels can demonstrate nonlinear effects and rule out alternative explanations
具有多于两个水平的单因素设计可以展示非线性效应并排除其他解释。Data from single-factor designs can be presented in table or graphical form (bar graph or line graph)
单因素设计的数据可以以表格或图形形式展示(柱状图或折线图)。Data from single-factor designs can be analyzed with t-tests for two-level designs, or with one-way ANOVAs for multilevel designs
单因素设计的数据可以用t检验分析双水平设计,或用单因素方差分析分析多水平设计。Special-purpose control groups may be used to compare the effect of experimental treatments to no-treatment controls
特殊用途的对照组可以用来比较实验处理与无处理对照的效果。Factorial designs allow us to evaluate the effects of multiple IVs on the DV or DVs
因素设计允许我们评估多个自变量对因变量或因变量的影响。There are different types of factorial designs, depending on how you manipulate your IVs
根据自变量的操控方式,有不同类型的因素设计。Main effects of each IV and interactions among IVs are the results from factorial designs
每个自变量的主效应和自变量之间的交互作用是因素设计的结果。Factorial ANOVAs are the statistical tests used for analyzing data from factorial designs
因素方差分析是用于分析因素设计数据的统计检验。With the experimental design tools at your disposal, remember to be an ethical researcher
使用手头的实验设计工具时,记住要成为一个有道德的研究者。
PSY2013 Lecture 05 Notes