วิธีการวิจัยและการศึกษาเชิงทดลอง
International Baccalaureate · IB Diploma · Psychology · SL หัวข้อ 3 8:48 การบรรยายภาษาอังกฤษ · คำบรรยายภาษาอังกฤษ + 中文 ลอยตัวบนภาพ
บท
Transcript
In 2015, 270 researchers tried to repeat 100 published psychology studies.
2015 年，270 名研究者尝试重复 100 项已发表的心理学研究。
Only about a third, 36 percent, found a significant result again.
只有约三分之一，即百分之 36，再次得到显著结果。
Some of the original studies had small samples, flexible analyses or biased designs.
一些原始研究样本小、分析方法随意或设计有偏差。
Research methods are not a list to memorise; they decide which findings you can trust.
研究方法不是要背的清单，它决定了哪些发现值得信任。
You need research methods for three parts of the assessment.
研究方法用于评估的三个部分。
In your class practicals you carry out at least four small studies, one with each method: an experiment, an interview or focus group, an observation, and a questionnaire.
在课堂实践中，你至少要完成四项小研究，每种方法一项：实验、访谈或焦点小组、观察和问卷。
Paper 2 asks about your practicals and an unseen study.
试卷二考查你的课堂实践和一项陌生研究。
The internal assessment is a research proposal: you plan a study for a population of interest, but you do not carry it out.
内部评估是一份研究计划书：你为一个目标人群设计研究，但不实施。
There are six methods.
研究方法有六种。
An experiment changes one variable and measures another, so it can show cause, but it is often artificial.
实验改变一个变量并测量另一个，能显示因果，但常常不自然。
A questionnaire is quick with large samples, but people may not answer honestly.
问卷快速、样本大，但人们未必如实回答。
An interview or focus group gives rich detail, but it is hard to compare.
访谈或焦点小组细节丰富，但难以比较。
An observation records real behaviour, but cannot show cause.
观察记录真实行为，但不能显示因果。
A correlational study shows relationships, but not cause.
相关研究显示关系，但不显示因果。
A case study gives rich detail on rare cases, but may not apply to others.
个案研究对罕见案例提供丰富细节，但可能不适用于其他人。
Quantitative data are numbers; qualitative data are words, pictures or descriptions.
定量数据是数字；定性数据是文字、图片或描述。
Experiments and questionnaires mostly produce quantitative data; interviews mostly produce qualitative data.
实验和问卷主要产生定量数据；访谈主要产生定性数据。
An experiment changes the independent variable and measures the effect on the dependent variable, while keeping other variables the same.
实验改变自变量，测量它对因变量的影响，同时保持其他变量不变。
A variable that changes with the IV and also affects the DV is a confounding variable.
随自变量一起变化、又影响因变量的变量是混淆变量。
Hypotheses must be precise.
假设必须精确。
Research hypothesis: students who learn 20 words in silence will recall more words than students who learn them with music playing.
研究假设：在安静中学习 20 个单词的学生，比在有音乐时学习的学生回忆出更多单词。
Null hypothesis: there will be no difference in recall, or any difference will be due to chance.
零假设：回忆没有差异，或任何差异都出于偶然。
There are three experimental designs.
实验设计有三种。
Independent samples puts different people in each condition: there are no order effects, but the groups may differ, so use random allocation.
独立样本在每个条件中使用不同的人：没有顺序效应，但组间可能不同，所以要随机分配。
Repeated measures has the same people do both conditions: no group differences, but order effects, so use counterbalancing.
重复测量让同一批人完成两个条件：没有组间差异，但有顺序效应，所以要平衡抵消。
Matched pairs pairs similar people, one in each condition: fewer group differences, but it is slow to match.
匹配配对把相似的人配对，各分到一个条件：组间差异较少，但匹配很慢。
A laboratory experiment has high control but may not feel real: low ecological validity.
实验室实验控制度高，但可能不真实，即生态效度低。
A field experiment happens in a real setting.
现场实验在真实环境中进行。
In a natural experiment the researcher does not choose the IV, for example a new school timetable.
在自然实验中，研究者不选择自变量，例如新的学校课表。
In a quasi-experiment the IV is a fixed trait such as age or gender, so participants cannot be randomly allocated.
在准实验中，自变量是年龄或性别等固定特征，所以参与者不能被随机分配。
Random sampling gives everyone an equal chance, but needs a full list of the population.
随机抽样让每个人都有同等机会，但需要完整的总体名单。
Opportunity sampling uses whoever is available, so it may not represent the population.
方便抽样使用能找到的人，可能不代表总体。
Self-selected sampling uses volunteers, who may be different.
自愿抽样使用志愿者，他们可能与众不同。
Purposive sampling chooses people for a feature, and the choice may be biased.
目的抽样按某一特征选人，选择可能有偏差。
Snowball sampling lets participants recruit others, so people know each other and are alike.
滚雪球抽样让参与者招募其他人，于是大家彼此认识、彼此相似。
Purposive and snowball sampling are common for interviews.
目的抽样和滚雪球抽样常用于访谈。
A questionnaire can use closed questions, such as yes or no, or a Likert scale from 1, strongly disagree, to 5, strongly agree; or it can use open questions.
问卷可以用封闭式问题，例如是或否，或者从 1“非常不同意”到 5“非常同意”的李克特量表；也可以用开放式问题。
Closed answers are easy to count; open answers give reasons.
封闭式答案容易统计，开放式答案能给出理由。
People may give the answer that sounds good: social desirability bias.
人们可能给出听起来好的答案：社会期望偏差。
Anonymous questionnaires reduce it.
匿名问卷可以减少这种偏差。
An interview can be structured, with fixed questions; semi-structured, with an interview guide and follow-up questions; or unstructured.
访谈可以是结构化的，问题固定；半结构化的，有访谈提纲和追问；或非结构化的。
A focus group is an interview with about 6 to 10 people at once, where answers spark each other's ideas, but a strong speaker can lead the group.
焦点小组是同时访谈约 6 到 10 人，彼此的回答能激发想法，但强势的发言者可能主导小组。
Interviews are recorded, transcribed and analysed by finding themes.
访谈要录音、转录，并通过寻找主题来分析。
An observation can be naturalistic or controlled, overt or covert, and participant or non-participant.
观察可以是自然的或受控的，公开的或隐蔽的，参与式的或非参与式的。
Observers use a behaviour checklist with clear categories, for example hits, shares a toy, helps.
观察者使用有清晰类别的行为清单，例如“打人”“分享玩具”“帮助”。
Two observers record the same sessions, and high agreement shows inter-rater reliability.
两名观察者记录同一时段，高度一致说明评分者间信度高。
Participant bias means participants change their behaviour.
参与者偏差指参与者改变了自己的行为。
With demand characteristics they guess the aim; the fix is to hide the aim and debrief later.
需求特征指他们猜出了研究目的；解决办法是隐藏目的，事后说明。
Researcher bias means the researcher's expectations affect the results; the fix is a double-blind design or a second coder.
研究者偏差指研究者的预期影响了结果；解决办法是双盲设计或第二名编码者。
Generalisability means results apply to the population, and a large random sample helps.
可推广性指结果适用于总体，大而随机的样本有帮助。
Qualitative findings are judged differently.
定性研究的评判方式不同。
Credibility asks whether the findings reflect what participants meant; the fix is triangulation, using several methods or researchers.
可信度问发现是否反映了参与者的本意；解决办法是三角验证，即使用多种方法或多名研究者。
Transferability asks whether the findings apply to other settings; the fix is to describe the setting in detail.
可迁移性问发现是否适用于其他情境；解决办法是详细描述研究情境。
Every study must follow ethical rules.
每项研究都必须遵守伦理规则。
Informed consent: people know what they agree to, and parents consent for children under 16.
知情同意：人们知道自己同意了什么，16 岁以下儿童由父母同意。
The right to withdraw at any time.
随时退出的权利。
No deception unless it is necessary and harmless.
除非必要且无害，否则不得欺骗。
Debriefing: telling participants the true aim afterwards.
事后说明：事后告诉参与者真实目的。
Protection from harm.
免受伤害。
And confidentiality: no names in the report.
以及保密：报告中不出现姓名。
Responsibility, one of the six concepts, asks you to apply these rules to a particular study.
六个概念之一的“责任”要求你把这些规则用于具体研究。
Paper 2 Section A asks four questions about one class practical: describe what you did, for 4 marks; explain a concept in relation to it, for 4 marks; compare it with another method, for 6 marks; and design a study with a different method on the same topic, for 6 marks.
试卷二 A 部分针对一项课堂实践提出四个问题：描述你做了什么，4 分；结合它解释一个概念，4 分；与另一种方法比较，6 分；用另一种方法就同一话题设计研究，6 分。
Describe how you used an interview in your class practical.
描述你在课堂实践中如何使用访谈。
Our aim was to find out how IB students experience stress before mock exams.
我们的目的是了解国际文凭学生在模拟考前如何体验压力。
We used purposive sampling to choose four DP2 students who had just taken mocks, and each gave informed consent.
我们用目的抽样选了四名刚考完模拟考的 DP2 学生，每人都给出了知情同意。
We ran semi-structured interviews of about 15 minutes with six open questions, and recorded them.
我们进行了约 15 分钟的半结构化访谈，有六个开放式问题，并录了音。
We transcribed and coded the recordings, and two themes appeared: lack of sleep, and comparing scores with friends.
我们转录并编码了录音，出现了两个主题：睡眠不足，以及和朋友比较分数。
Section B, worth 15 marks, gives you an unseen study.
B 部分 15 分，给你一项陌生研究。
You discuss it with two or more of four concepts: bias, causality, measurement and responsibility.
你要用偏差、因果关系、测量和责任四个概念中的两个或以上来讨论它。
For each concept: point to a detail in the study, explain its effect on the conclusion, and suggest a better design.
对每个概念：指出研究中的一个细节，解释它对结论的影响，再提出更好的设计。
The internal assessment is a research proposal worth 24 marks.
内部评估是一份 24 分的研究计划书。
You choose a topic and a population of interest, for example international students in their first year in China, and plan a study with one of the four class-practical methods.
你选择一个话题和一个目标人群，例如在中国读大一的国际学生，并用四种课堂实践方法之一来设计研究。
It has four parts: introduction, research methodology, data collection, and discussion.
它有四个部分：引言、研究方法、数据收集和讨论。
You argue for every choice, but you do not collect data.
你要论证每一个选择，但不收集数据。
Here is an example plan.
这是一个计划范例。
How do first-year international students in Ningbo experience making friends?
宁波的大一国际学生如何体验交朋友？
Method: semi-structured interviews, because the question is about experience and needs rich data.
方法：半结构化访谈，因为问题关乎体验，需要丰富的数据。
Sample: purposive sampling of eight students, through the university's international office.
样本：通过大学国际处，用目的抽样选八名学生。
Ethics: informed consent, a right to withdraw, no names, and information about counselling services.
伦理：知情同意、退出权、不写姓名，并提供心理咨询服务的信息。
Discussion: the researcher's own views could shape the questions, so a second researcher codes the transcripts; and eight students from one university may not transfer to other cities.
讨论：研究者自己的观点可能影响问题，所以由第二名研究者为转录稿编码；来自同一所大学的八名学生的结果未必能迁移到其他城市。
Check yourself.
自我检查。
Can you give a strength and a limit of each of the six methods?
你能说出六种方法各自的一个优点和一个局限吗？
Can you define IV, DV and confounding variables, write research and null hypotheses, and fix each design's problem?
你能定义自变量、因变量和混淆变量，写出研究假设和零假设，并解决每种设计的问题吗？
Can you name five sampling techniques with their problems?
你能说出五种抽样方法及其问题吗？
Can you explain social desirability, focus groups and inter-rater reliability; participant and researcher bias; credibility and triangulation; and the ethical rules?
你能解释社会期望偏差、焦点小组和评分者间信度；参与者偏差和研究者偏差；可信度和三角验证；以及伦理规则吗？
And can you describe Paper 2 and the four parts of the research proposal?
你能描述试卷二和研究计划书的四个部分吗？