中央大學.資訊管理系 范錚強 mailto: updated

Presentation on theme: "中央大學.資訊管理系 范錚強 mailto: updated"— Presentation transcript:

1 中央大學.資訊管理系 范錚強 mailto: ckfarn@mgt.ncu.edu.tw 2012.03 updated
5 Research Design 中央大學.資訊管理系 范錚強 mailto: updated

2 Learning Objectives Understand . . .
The basic stages of research design. The major descriptors of research design. The major types of research designs. The relationships that exist between variables in research design and the steps for evaluating those relationships. 研究精神：排除「威脅」(threats) 決定適當的研究設計 中央資管：范錚強

3 What Is Research Design?
Blueprint Plan Guide Framework There are many definitions of research design. Research design is the blueprint for fulfilling research objectives and answering questions. Its essentials include 1) an activity and time-based plan, 2) a plan based on the research questions, 3) a guide for selecting sources and types of information, 4) a framework for specifying the relationships among the study’s variables, and 5) a procedural outline for every research activity. 中央資管：范錚強

4 What Tools Are Used in Designing Research?
Exhibit 6-2 provides one project management tool: critical path method (CPM). In a CPM chart: The nodes represent major milestones. The arrows suggest the work needed to get to the milestones. More than one arrow pointing to a node indicates all those tasks must be completed before the milestone has been met. Usually a number is placed along the arrow showing the number of days or weeks required for that task to be completed. The pathway from start to end that takes the longest time to complete is called the critical path. 有些事可以平行作業 中央資管：范錚強

5 What Tools Are Used in Designing Research?
Project Plan in Gantt chart format A Gantt chart ( Exhibit 5-11, MindWriter project) is a common project planning tool that reveals summary tasks, benchmarking milestones, and detailed tasks against a time frame for the overall project. Tasks may be color coded to indicate a particular team member’s responsibilities. Many project-management software packages include Gantt charting. The chart may be used to monitor projects to keep them on time, as well as to alert the client or manager to steps requiring their approval—and what happens to the project’s schedule if approval is not forthcoming when it is needed. 中央資管：范錚強

6 Design in the Research Process
Exhibit 6-1 illustrates design in the research process and highlights the topics covered by the term research design. Subsequent chapters will provide more detailed coverage of the research design topics. 中央資管：范錚強

7 Descriptors of Research Design
Question Crystallization Perceptual Awareness Data Collection Method Descriptors Purpose of Study Experimental Effects Exhibit 6-3 information is presented here in a discussion format. The degree to which the research question has been crystallized Exploratory study Formal study The method of data collection Monitoring Communication Study The power of the researcher to produce effects in the variables under study Experimental Ex post facto The purpose of the study Reporting Descriptive Causal-Explanatory Causal-Predictive The time dimension Cross-sectional Longitudinal The topical scope—breadth and depth—of the study Case Statistical study The research environment Field setting Laboratory research Simulation The participants’ perceptional awareness of the research activity Actual routine Modified routine Research Environment Time Dimension Topical Scope 中央資管：范錚強

8 Degree of Question Crystallization
Exploratory Study Loose structure Expand understanding Provide insight Develop hypotheses The degree to which the research question has been crystallized or structured is the first descriptor of research design. There are two options. Exploratory studies are used when the research question is still fluid or undetermined. The goal of exploration is to develop hypotheses or questions for future research. Formal studies are used when the research question is fully developed and there are hypotheses to be examined. 中央資管：范錚強

9 X Y x y 探索：構念的形成 抽象化 構念化 構念層次 推論：因果關係 因果、共變關係 實質層次 測量 誤差 誤差 測量
中央資管：范錚強 實質層次

10 Degree of Question Crystallization
Exploratory Study Loose structure Expand understanding Provide insight Develop hypotheses Formal Study Precise procedures Begins with hypotheses Answers research questions The degree to which the research question has been crystallized or structured is the first descriptor of research design. There are two options. Exploratory studies are used when the research question is still fluid or undetermined. The goal of exploration is to develop hypotheses or questions for future research. Formal studies are used when the research question is fully developed and there are hypotheses to be examined. 中央資管：范錚強

11 X Y x y 理論驗證 操作化 操作化 構念層次 因果關係：實徵支持 學說：因果關係 因果、共變關係 統計結論 實質層次 測量 誤差 誤差
操弄、測量 y 測量 誤差 操作化 實質層次 因果、共變關係 統計結論 中央資管：范錚強

12 Approaches for Exploratory Investigations
Participant observation Film, photographs Projective techniques Psychological testing Case studies Ethnography 民俗誌 Expert interviews Document analysis Proxemics and Kinesics 肢體語言學 The objectives of exploration may be accomplished with qualitative and quantitative techniques, but exploration relies more heavily on qualitative techniques. Qualitative techniques are non-quantitative data collection used to increase understanding of a topic. Qualitative refers to the meaning, definition, analogy, model, or metaphor characterizing something, while quantitative assumes the meaning and refers to a measure of it. There are many approaches useful for exploratory investigations of management questions. Several such approaches are listed in the slide. These techniques are expanded upon in Chapter 8. 中央資管：范錚強

13 Desired Outcomes of Exploratory Studies
Established range and scope of possible management decisions Established major dimensions of research task Defined a set of subsidiary questions that can guide research design An exploratory study is finished when the researchers have achieved the objectives listed in this and the next slide. 尋找規則 Law 和初步檢視可能的原因 中央資管：范錚強

14 Desired Outcomes of Exploratory Studies (cont.)
Developed hypotheses about possible causes of management dilemma Learned which hypotheses can be safely ignored Concluded additional research is not needed or not feasible 中央資管：范錚強

15 Commonly Used Exploratory Techniques
Secondary Data Analysis Experience Surveys Focus Groups While there are several types of exploratory techniques possible these are the three techniques with the widest applications for business researchers. Secondary data analysis is also called a literature search. Within secondary data exploration, researchers should start first with an organization’s own data archives. The second source of secondary data is published documents prepared by authors outside the sponsor organization. Experience surveys are semistructured or unstructured interviews with experts on a topic or a dimension of a topic. Focus groups are discussions on a topic involving a small group of participants led by a trained moderator. 中央資管：范錚強

16 Face-to-face interaction—one of the best ways to learn from participants.
This ad for Delve, a marketing research firm, points out the importance of primary data. 中央資管：范錚強

17 Experience Surveys What is being done?
What has been tried in the past with or without success? How have things changed? Who is involved in the decisions? What problem areas can be seen? Whom can we count on to assist or participate in the research? Experience surveys are sometimes called expert interviews or key informant surveys. Even though the term survey is in the name, it is not a closed-ended, structured survey. Rather, experience surveys are interviews designed to extract as much information as possible from the expert’s knowledge. Broad questions guide the discussion. Several questions that could be used in an experience survey are listed in the slide. Some examples of groups who might be identified for an experience survey include potential car buyers, dealer sales representatives, advertising columnists, and automotive industry analysts. 中央資管：范錚強

18 Focus Groups Group discussion 6-10 participants Moderator-led
90 minutes-2 hours Focus groups are widely used in business research. They are led by a trained moderator and typically include 6-10 participants. Mini-focus groups with just 3 people are increasingly common. The facilitator uses group dynamics principles to focus or guide the group in an exchange of ideas, feelings, and experiences on a specific topic. Focus groups can take place in a variety of settings, but many take place in a focus group room equipped with one-way window and recording devices. 中央資管：范錚強

19 Data Collection Method
Monitoring: 觀察、資料蒐集 行為 Communication: 詢問 態度、看法、瞭解程度 Method of data collection distinguishes between monitoring and communication processes. Monitoring processes are studies in which the researcher inspects the activities of a participant or the nature of some material without eliciting responses from the participant. Examples of monitoring include traffic counts, library searches, and counting cars in a parking lot. In a communication study, the researcher questions the participants and then collects their responses by personal or impersonal means.The collected data may result from 1) telephone or interview conversations, 2) self-administered or self-reported instruments sent through the mail, dropped-off in convenient locations, or transmitted electronically, 3) instruments presented before and/or after a treatment or stimulus condition in an experiment. 中央資管：范錚強

20 The Time Dimension Cross-sectional 單一時間斷面 無法確認因果關係方向 Longitudinal 長時間
投入資源可能太多 Cross-sectional studies are studies conducted only once. They seek to reveal a snapshot at one point in time. Longitudinal studies include repeated measures over an extended period of time. Therefore, longitudinal studies can track changes over time. Despite this advantage, longitudinal studies are expensive and time-intensive. 中央資管：范錚強

21 The Topical Scope Statistical Study Breadth Population inferences
Quantitative Generalizable findings Case Study Depth Detail Qualitative Multiple sources of information The topical scope refers to the breadth (what properties will be measured) and depth (at what level will the properties be measured) of the study in question. A statistical study is designed for breadth rather than depth. It attempts to capture a population’s characteristics by making inferences from a sample’s characteristics and then testing resulting hypotheses. A case study places more emphasis on full contextual analysis of a few events or conditions and their interrelations. Case studies rely on qualitative data and emphasize the use of results for insight into problem-solving, evaluation, and strategy. While case studies are not considered “scientific,” they do play an important role in challenging theory, providing new hypotheses, and offering new ideas on constructs. 陷阱：整合採取兩個方法的短處 統計分析缺乏廣度 個案分析缺乏深度 中央資管：范錚強

22 The Research Environment
Field conditions Lab conditions Designs also differ as to whether they occur under actual environmental conditions. Field conditions mean that the research occurs in the actual environmental conditions where the dependent variable occurs. Under laboratory conditions, the studies occur under conditions that do not simulate actual environmental conditions. In a simulation, the study environment seeks to replicate the natural environment in a controlled situation. For instance, a lab set up as a kitchen would serve as a simulation of a consumer’s own kitchen. Simulations 中央資管：范錚強

23 Purpose of the Study Reporting Descriptive Casual -Explanatory
Causal -Predictive 中央資管：范錚強

24 Descriptive Studies Who? How much? What? When? Where? 中央資管：范錚強
The purpose of the study asks whether the research is concerned with describing the population’s characteristics or with trying to explain the relationships among variables. Descriptive studies discover the answers to the questions who, what, when, where, or how much. 中央資管：范錚強

25 Descriptive Studies Descriptions of population characteristics
Estimates of frequency of characteristics Discovery of associations among variables In contrast to exploratory studies, more formalized studies are typically structured with clearly stated hypotheses or investigative questions. Formal studies serve a variety of research objectives such as those listed in the slide. The third objective, discovery of variable associations, is sometimes labeled a correlational study, which is a subset of descriptive studies. Correlation is the relationship by which two or more variables change together, such that systematic changes in one accompany systematic changes in the other. 樣本代表性極端重要 中央資管：范錚強

26 Experimental Effects Ex Post Facto Study
After-the-fact report on what happened to the measured variable 事後回溯 Experiment Study involving the manipulation or control of one or more variables to determine the effect on another variable 可操弄、控制 Causal studies are differentiated by their ability to control and manipulate variables. Causal studies may be experiments or ex post facto studies. Experiments are studies involving the manipulation of one or more variables to determine the effect on another variable. For example, direct marketers can use split tests on mailings to test which mailing resulted in the highest response rate. Ex post facto designs are evaluations made after-the-fact based on measured variables. 中央資管：范錚強

27 Non-Fishing-Club Member
Ex Post Facto Design Fishing Club Member Non-Fishing-Club Member Age High Absentee Low Absentee Under 30 years 36 6 30 48 30 to 45 4 35 117 45 and over 5 115 This shows employment absenteeism results by age of head of household and club membership. This is an example of results that could come from an ex post facto study. Instead of manipulating variables or controlling exposure to an experimental variable to judge absenteeism, we study subjects who have been exposed to the independent factor and those who have not. 中央資管：范錚強

28 Causation and Experimental Design
Control/ Matching Random Assignment To be convincing, inferences from experimental designs must meet two other requirements, in addition to those covered on the last slide. The first is control. Control means that all factors but the IV are held constant and are not confounded with another variable that is not part of the study. The second requirement is that each person in the study must have an equal chance for exposure to each level of the independent variable. This is accomplished through random assignment of subjects to groups. Random assignment uses a randomized list of participants for assigning participants to experimental and control groups. The control group is the group of participants that is measured but not exposed to the independent variable being studied. Control can also be accomplished using matching. Matching is an equalizing process for assigning participants to experimental and control groups. In this case, we ensure that all groups are essentially equal with respect to the variable of influence. 中央資管：范錚強

29 Mills Method of Agreement
Exhibit 6-4 中央資管：范錚強

30 Mills Method of Difference
Exhibit 6-5 中央資管：范錚強

31 Causal Studies Symmetrical Z  X Z  Y Reciprocal X  Y Y  X
Asymmetrical X  Y People without scientific training may think that a correlation is causation. However, just because two things change together does not imply a cause-and-effect relationship. The essential element of causation is that some external factor produces a change in the dependent variable: A produces B. Empirically, we never demonstrate causality with certainty because we do not prove causal linkages deductively. Empirical research conclusions are based on inferences or inductive conclusions. These conclusions are probability statements, based on what we observe and measure and what we conclude is likely to happen. There are three possible relationships that can occur between two variables. These are named in the slide. A symmetrical relationship is one in which two variables vary together but we assume the changes in neither variables are due to changes in the other. A reciprocal relationship exists when two variables mutually influence or reinforce each other. With asymmetrical relationships, we postulate that changes in one variable (independent variable) are responsible for changes in another (dependent variable). Exhibit 8-3 describes the four types of asymmetrical relationships and is provided on the next slide. 中央資管：范錚強

32 Understanding Casual Relationships
Property Behavior Response Exhibit 8-3 details…relationship but students need to understand the foundations: A stimulus is an event or force (e.g., drop in temperature, crash of stock market, product recall, or explosion in factory). A response is a decision or reaction. A property is an enduring characteristic of a subject that does not depend on circumstances for its activation (e.g., age, gender, family status, religious affiliation, ethnic group, or physical condition). A disposition is a tendency to respond in a certain way under certain circumstances (e.g., attitudes, opinions, habits, values, and drives). A behavior is an action (e.g., consumption habits, work performance, interpersonal acts, and other kinds of performance). Disposition Stimulus 中央資管：范錚強

33 Asymmetrical Casual Relationships
Stimulus-Response Property- Disposition Property- Behavior Exhibit 8-3 details the four types of asymmetrical causal relationships. A stimulus-response relationship refers to an event or change that results in a response from some object. An example is a change in commission structure that leads to a higher level of agent sales. A property-disposition relationship refers to an existing property that causes a disposition. An example is the relationship between age and attitudes about saving. A disposition-behavior relationship refers to a disposition that causes a specific behavior. An example is the relationship between moral values and shoplifting. A property-behavior relationship refers to an existing property that causes a specific behavior. An example is the relationship between age and sports participation. Disposition-Behavior 中央資管：范錚強

34 Fishbein & Ajzen 的 TRA Aijen and Fishbein: TRA 哪個是哪個？
Theory of Reasoned Actions 慎思行動理論 Beliefattituteintentionbehavior 哪個是哪個？ Stimulus: 刺激 Property: 性質（人口變數、個人特質等） Disposition: 態度、選擇、傾向 Behavior: 行為 中央資管：范錚強

35 Exhibit 6-6 Types of Asymmetrical Causal Relationships
Relationship Type Nature of Relationship Examples Stimulus-response An event or change results in a response from some object. A change in work rules leads to a higher level of worker output. A change in government economic policy restricts corporate financial decisions. A price increase results in fewer unit sales. Property-disposition An existing property causes a disposition. Age and attitudes about saving. Gender attitudes toward social issues. Social class and opinions about taxation. Disposition-behavior A disposition causes a specific behavior. Opinions about a brand and its purchase. Job satisfaction and work output. Moral values and tax cheating. Property-behavior An existing property causes a specific behavior. Stage of the family life cycle and purchases of furniture. Social class and family savings patterns. Age and sports participation. 中央資管：范錚強

36 Evidence of Causality Covariation between A and B Time order of events
No other possible causes of B When testing causal hypotheses, we seek three types of evidence: Covariation between A and B. Do we find that A and B occur together in the way hypothesized? When A does not occur, is there also an absence of B? When there is more or less of A, does one also find more or less of B? 2. Time order of events moving in the hypothesized direction. Does A occur before B? 3. No other possible causes of B. Can one determine that C, D, and E do not covary with B in a way that suggests possible causal connections? 中央資管：范錚強

37 Participants’ Perceptional Awareness
No deviation perceived Deviations perceived as unrelated The usefulness of a research design is reduced when people in a disguised study perceive that research is being conducted. Participants’ perceptions can influence the outcomes of research. This was first discovered in the 1920s when researchers at the Hawthorne plant of the Western Electric Company found that participants reacted favorably to receiving attention. There are three levels of perception to consider and these are highlighted in the slide. Mystery shopping sometimes provides an example of the third level of perception. Mystery shopping involves individuals who pose as customers and visit retail or service organizations to observe and measure specific behaviors or circumstances. If a retail sales associate knows that she is being observed and evaluated, she is likely to modify her performance. Deviations perceived as researcher-induced 中央資管：范錚強

38 X Y x y 研究中的效度Validity理念 構念層次 外部效度 構念效度 構念效度 內部效度 統計結論效度 操作層次 替代效果 操作化
因果關係 統計結論效度 操弄、測量 共變關係 測量 誤差 操作層次 誤差 中央資管：范錚強

39 統計結論效度 Statistical Conclusion Validity
Q:研究中的因變數和自變數是否相關？ (共變 co-variation) 研究設計的敏感度是否足夠顯示資料間的共變 若敏感度構，是否有足夠的証據支持「因」和「果」之間有共變？ 如果有充分証據顯示其間的共變，其強度為何 中央資管：范錚強

40 對統計結論效度的威脅 分析缺乏統計力 違背統計測試的假設 在資料中找尋「規則」 多重比對、資料的特例 衡量方法的信度 實驗操弄的信度
樣本大小、信心水準的影響 違背統計測試的假設 母體分布、均質性、自變數之間的獨立性 在資料中找尋「規則」 多重比對、資料的特例 衡量方法的信度 實驗操弄的信度 實驗情境中的隨機偏誤 其他外在的、未被控制的影響因素 回答問卷者的系統性誤差 中央資管：范錚強

41 內部效度 Internal Validity
Q:若A和B共變，其間有何因果關係？ A A->B A??B 替代的 因果關係 B B->A A->C->B C A A + A - + + B - B - B 第三變項 的干擾 C C C - + + 內部效度的精神在於排除第三變項對A-B關係的替代性解釋 中央資管：范錚強

42 對內部效度的最主要威脅 發生歷史事件 樣本成熟 測量工具改變 統計迴歸 選擇偏誤 樣本流失 因果關係的混淆
受測樣本在前測和後測之間的成長，例如，熟悉測試工具 測量工具改變 統計迴歸 例如：將樣本依前測得分分組，高分組退步，低分組進步 選擇偏誤 樣本流失 因果關係的混淆 在相關性研究中，經常發生 中央資管：范錚強

43 構念效度 Construct Validity
構念的操作化是否良好 操作化：操弄和測量 必須確保 自變數的確反映了相關的構念 自變數不會隨著相關、但不同的變數做改變 依變數應該反映了相關的構念 依變數不應該受無關變數的影響 針對同一個構念的不同衡量結果應該收斂 針對相關但不同構念的衡量結果應該發散 中央資管：范錚強

44 對構念效度的威脅 操作化前對構念搞不清楚 單一操作化的偏誤 單一方法的偏誤 樣本對研究假說的猜測 構念的定義、操作化定義、它確切的意義
實驗中，需要多項操弄水準、劑量 單一方法的偏誤 需要測量多個「外顯」的變數 樣本對研究假說的猜測 中央資管：范錚強

45 對構念效度的威脅2 樣本對測量感到緊張 樣本對研究有特殊期望 構念混淆、模糊，構念的程度差異 不同操弄之間的相互影響
操弄和測量之間的相互影響 中央資管：范錚強

46 信度 Reliability 測量工具的穩定度、可靠度 測量工具的內部一致性 重測信度、折半信度、複本信度
項目間的信度(e.g. Cronbach a) 評分員間的信度 中央資管：范錚強

47 外部效度 External Validity
Q: 研究結果能否推論到整個目標母體？ 母體—我們在說哪一個？ 目標母體 針對目標母體所作的正式描述 在實徵研究中，實際達成的樣本所對應的母體 另外一個和外部效度相關但不同的觀念： 可類化性Generalizability 研究結果可推廣到其他母體的程度 中央資管：范錚強

48 外部效度 External Validity
目標母體 正式描述 實際達成的母體 規劃樣本 實際達成的樣本 中央資管：范錚強

49 對外部效度的威脅 研究情境的影響 選擇和操弄間的相互影響 研究情境和操弄間的相互影響 歷史事件和操弄間的相互影響 在某企業中進行研究
自我選擇可能會排除某一類的樣本 例如：新藥的測試，自願者求生意願較強 研究情境和操弄間的相互影響 歷史事件和操弄間的相互影響 過程長的研究 中央資管：范錚強

50 透過抽樣增進外部效度 全面隨機抽樣 分層隨機抽樣 選取典型樣本 求取代表性、統計力強 經常昂貴、實驗室實驗下不可行 刻意求取樣本的異質性
每一個母體中的子群體都被納入 選取典型樣本 統計力弱、在學術研究中不利 中央資管：范錚強

51 透過重複研究增進外部效度 不同的重複可能、各有優劣 比較好的重複研究 同時重複、連續重複 獨立研究人員、相同研究人員
真實重複、假設重複（大型研究中的不同部份） 比較好的重複研究 多個小型的研究 多個情境 多個研究者 中央資管：范錚強

52 研究效度間的取捨 學術性研究 實務性研究、應用研究 在進行研究設計時，研究者必須清醒的對各種效度的考量，做全盤的取捨
內部效度構念效度統計結論效度外部效度 實務性研究、應用研究 內部效度外部效度依變數的構念效度統計結論效度自變數的構念效度 在進行研究設計時，研究者必須清醒的對各種效度的考量，做全盤的取捨 例如：在研究管理者決策訂定的研究時，應該用學生為樣本（高可行性、高內部效度、低外部效度）；還是在真實情境用真實的經理人員（低可行性、低內部效度、高外部效度） 中央資管：范錚強

53 研究方法的選擇 強方法 Strong methods 弱方法 Weak methods 能明確地呈現內部校度 實驗室研究
Ex post facto, 自然觀察，訪談… 中央資管：范錚強