流行病学数据的搜集 Collection of Epidemiologic Data 朱保平，医学博士，生物统计学硕士 Bao-Ping Zhu, MD, MS 中国现场流行病学培训项目常驻指导专家 U.S. Resident Expert Advisor to China FETP

现场流行病学数据的类别 Types of epidemiologic data 结果数据 Outcome data 暴露数据 Exposure data 其他相关危险因素的数据 Data on related risk factors

流行病学数据的来源 Sources of epidemiologic data 现成数据 Existing data 需搜集的数据 Data to be collected “The questionnaire should only collect that information that you need for the study”  Keep it Simple will be a common theme of this lecture The research questions lead to the analysis plan, and both play integral parts in designing the questionnaire.

常见现成数据 Examples of existing data 人口统计学数据 Population data 人口普查数据 Census data 普查年间的估计数 Inter-census estimates 生命统计数据 Vital statistics data 出生登记 Birth registry 死亡登记 Death data 结婚、离婚数据 Marriage and divorce 监测数据 Surveillance data 疾病监测 Disease surveillance 健康危险因素监测 Behavioral risk factors surveillance

常见现成数据 Examples of existing data 健康状况和行为数据 Health status and behavioral data 健康和营养调查 Health and nutrition surveillance 行为危险因素调查 Behavioral risk factor surveillance 社会经济学数据 Socioeconomic data 消费者价格指数, 国民生产总值, 失业率 CPI, GNP, unemployment rate 行政管理数据 Administrative data 门诊记录 Outpatient records 住院记录 Hospitalization data 实验室检查记录 Laboratory testing data

流行病学数据的搜集方法 Methods for collecting epidemiologic data 自填式调查表 Self-administered questionnaire 面对面调查 Personal interviews 对生物标本进行生化检测 Biochemical analysis of biological specimens 对环境进行物理、化学分析 Physical or chemical analysis of the environment “The questionnaire should only collect that information that you need for the study”  Keep it Simple will be a common theme of this lecture The research questions lead to the analysis plan, and both play integral parts in designing the questionnaire.

流行病学数据的等级 Scales of Epidemiologic Data 定性数据 Nominal 性别，职业 Gender; occupation; race/ethnicity 定序数据 Ordinal 社会经济状况 Socioeconomic status: low, medium, high 区间数据 Interval 温度 Temperature 出生年份 Year of birth 比率数据 Ratio data 体重 Weight 身高 Height “The questionnaire should only collect that information that you need for the study”  Keep it Simple will be a common theme of this lecture The research questions lead to the analysis plan, and both play integral parts in designing the questionnaire.

优先考虑收集高等级数据 Higher scale data are always preferred 提供更丰富的信息 Richer information 统计学效力更高 Higher statistical power 统计分析方法有更多的选择 More options for statistical analysis 可转换成低等级数据 Higher scale data may be transformed into lower scale data 而低等级数据无法变成高等级数据 Lower scale data may not be transformed into higher scale data “The questionnaire should only collect that information that you need for the study”  Keep it Simple will be a common theme of this lecture The research questions lead to the analysis plan, and both play integral parts in designing the questionnaire.

数据的质量取决于所使用的测量工具 Data is only as good as the measurement instrument used 每次对某特征进行测量，我们都希望得到“真实”情况 We would like to measure the “truth” every time we measure a characteristic 如果达不到这个目标，就意味着有测量误差 Any deviation from this constitutes a measurement error 描述测量误差的两个指标 可靠性 Reliability 可信性 Validity A questionnaire can also be viewed as a measurement instrument (or measurement tool) – used as an indirect method of measurement. e.g. Exposure to a person sick with a respiratory illness: Did an individual visit the ill person? If yes, how long did he/she visit the person?  can generate a quantitative, indirect measure of exposure

可靠性 Reliability 对研究对象进行多次测量，所得到信息的是否相似 Refers to the degree of similarity of information obtained when the measurement is repeated in the same subject 每次测量的结果是否都相同？ Is the same value arrived at every time the measurement is taken/made? Do the values vary a lot on repeated administration To improve reliability we need to address the sources of variation Here instructor could draw a bulls eye on the board. Then draw 3 dots all in the same region, but not close to the bulls eye.

可信性 Validity 测量结果是否反映真实情况？ Is it a true reflection of what you want to measure? 在进行重复测量时，若测量值与真实值相比系统地偏高或偏低，则我们就说可信性较低 A measure lacks validity if an observer or an instrument measures the characteristic in the same individual repeatedly higher or repeatedly lower than the real value 这种系统性误差即“偏差”或“偏倚” This systematic error introduces a bias 一个可靠性好的测量值，其可信性可能仍然较低 A measure can be reliable but not valid This systematic error introduces a bias – have to find out what is the “truth” A measure can be reliable but not valid Here instructor can draw a bulls eye and have the 3 dots all around the bulls eye (“truth”). Then the instructor could draw in 3 dots close together at the bulls eye (reliable and valid).

可信性与可靠性之间的关系 Validity and Reliability 箭靶的比喻 The Target metaphor These 4 target situations show how an instrument (such as a rifle) can either be: Reliable but not valid. Valid but not reliable. Neither reliable nor valid. Both reliable and valid (desired situation). A questionnaire works much the same way. The information it generates should be both reliable and valid. 可靠 但不可信 可信 但不可靠 既不可靠 亦不可信 既可靠 亦可信

测量误差的来源 The sources of error in measurement 测量工具 The instrument 在对同一个研究对象进行测量时，测量工具本 身可能有前后差异 The instrument itself gives different results for the same individual 例如，血压袖带放置位置不同 e.g. blood pressure cuff 测量人员 The observer 不同测量人员之间的差别 Inter-observer variation - differences among observers 同一测量人员不同次之间的差别 Intra-observer variation - differences in the same observer

测量误差的来源 The sources of error in measurement 被测量者 The subject 生物学上的不一致性 Biological inconsistencies 记忆偏差 Memory 测量时的环境 The environment 温度 Temperature 湿度 Humidity

问卷 Questionnaires 设计一个好的问卷需要花时间和精力 A good questionnaire takes time and effort to design 不要低估这一点的重要性！Don’t underestimate this !! 问卷中所包含的问题的你确实需要的问题！ The questionnaire should only collect that information that you need for the study 例：人口学，暴露，结局的信息 e.g. Demographics / Exposure / Outcome 设计问卷前先考虑你所研究的问题和分析计划 Think about the research questions and analysis plan 例：是只想知道有无暴露，还是也想知道暴露剂量？ e.g. Are you only interested in exposure (Y/N) or are you also interested in dose ?? “The questionnaire should only collect that information that you need for the study”  Keep it Simple will be a common theme of this lecture The research questions lead to the analysis plan, and both play integral parts in designing the questionnaire.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

确立调查内容或待验假设 State the Research Questions or Hypotheses 提出假设以指导资料的搜集 Generate hypotheses to guide data collection 描述流行病学 Review descriptive studies 分析“异常值”为何异常 Look and learn from outliers 与他人讨论 Talk to people 与当地业务人员座谈讨论 Talk to local officials 去现场考察环境 Go to field and study the environment 1st you need to develop a research question or generate hypotheses to guide you in the process of data collection. Five key steps to do that: Review descriptive epidemiological- person, place and time. Loot at and learn from outliers Talk to patients, families or friends, etc Talk to local officials- what do they know Go to the field and study the environment

State the Research Questions or Hypotheses 确立调查内容或待验假设 State the Research Questions or Hypotheses 调查开始前思考要验证的假设。比如：Think about questions and related outcomes in advance. For examples: 接触病例与发病相关 Contact with a case is associated with illness 疫苗对疾病有保护作用 Vaccine is associated with protection from illness 喝牛奶与疾病有关联 Milk is associated with illness 疾病与喝牛奶的量有关联 Illness was associated with the amount of milk consumed 在这些情况下，需要对哪些暴露因素和结果进行测量？ What exposures and outcomes do we specifically want to measure? First of all, you need to develop research question(s) and any possible outcomes. You can begin this process by interviewing existing cases or reviewing some scientific literatures to get some idea how to conduct an epidemiologic study. The questionnaire will be developed from the exposures and outcomes we want to measure. This questionnaire will serve as our important study tool.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

制订分析大纲（空白表） Outline an analysis plan (empty tables) 用哪一种关联的指标？What measure of association will be used? 哪些关联是我所感兴趣的？Which associations am I interested in? 发病Ill 未发病Not Ill 接触过病例 Contact with case 需要对疾病进行定义或测量 We need to measure/define illness! Illness is defined using the Case Definitions The measure of contact with a case that will lead us to investigation is also in part decided by the case definition. e.g. Cutaneous Anthrax: direct Influenza: droplet The type of contact we are measuring will influence the amount of contacts we will need to interview more people exposed to droplets than direct contact, so if investigating influenza, we will interview more patients that if investigating cutaneous anthrax 未接触病例 No Contact with case 需要对接触进行测量 We need a measure of contact!

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

列出需要收集的变量 Variables to be collected (observed or measured) 被调查者 The participant 基本资料，如姓名，号码Participant (Name, ID #) 地址，年龄，性别，职业Address, age, sex, occupation, etc.... 暴露 The exposure 有无，地点，时间，剂量，持续时间If/where/when/how much/how long 结局 The outcome 症状，体征，检验结果 Symptoms/signs/tests When using a questionnaire on an outbreak investigation, these are the 3 main topics of interest: Participant information, Exposure information, Outcome information We want to know who is the case: name, where they live, where they work, age, sex, etc. what is the exposure: if the case was exposed; where, when, how much, how long, etc. what is the outcome: case develop symptoms; laboratory tests; hospitalization; death, etc.

可以用调查表来采集的信息 Information that may be collected with a questionnaire 个人特征 Individual characteristics 身高，年龄，收入 Height, age, income 环境 Environment 住房情况，家庭成员数 Housing, family size 行为，习惯 Behaviors, practices 饮酒，吸烟 Alcohol or tobacco consumption 判断 Judgements 意见 Opinions 态度 Attitudes 衡量知识的指标 Indicators of knowledge 有关危险因素的知识 Risk factors 对健康生活习惯的认识 Healthy lifestyle Other information that may be collected depending on the questionnaire’s purpose. All may have relevance to an investigation, but some you will more often see on a regular surveillance device, e.g., opinions, attitudes, healthy lifestyle, height, income. Environment, Behaviors/Practices and Risk factors may be important for both survey instruments and investigative questionnaires.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

Methods of obtaining information 获得信息的常用方法 Methods of obtaining information 自填式调查表 Self-administered questionnaire 邮寄调查表 Mail (postal) 网上调查 Web-based 调查员监督下自填 Monitored 调查员填写的调查表 Interviewer-administered questionnaire 电话调查 By telephone 面对面问卷 In person There are 3 main methods of administering questionnaires: Postal = Self-administered Telephone = Interviewer-administered In Person = Interviewer-administered

选择什么样的方法？ What method to choose? 问题的性质 Nature of questions 问的是什么样的问题？What is the research question to be addressed? 方便 Convenience 哪一方法最方便？Which method is most readily available? 花费 Cost 需要How costly is it to collect the information and how affordable is the method? Although the ideal method will almost always be an in-person interview, there are many times when this will not be possible. The biggest reason is cost: paying interviewers for there time and travel can be very expensive. It is not always convenient, either, especially when time is a limiting factor in a study. Some research questions don’t require in-person interviews, many times a phone interview is just as effective.

如何开展问卷调查 Questionnaire administration 面对面调查 In-person interviewing 最常用的一种方法 Most frequently used 优点：应答率较高；有弹性 High response rates; flexibility 缺点：费时；能造成观察者偏倚 Time consuming; potential observer bias 电话调查 Telephone interviews 优点：花费少；快；可在大范围内实施 Inexpensive; rapid; large numbers or area 缺点：无应答较高；没有面对面交流；调查时间仓促；可能造成观察者偏倚 Non-response; no visual cues; rushed; potential observer bias 邮寄调查表 Postal Questionnaires 优点：花费少；快速；可以在大范围内进行 Inexpensive; rapid; large numbers or area 缺点：无应答较高；组织起来复杂 Non-response; complexity Each method has its advantages and disadvantages. In-Person: - The most frequently used and generally yields the highest response rate. - However it is very time consuming and therefore the most costly of the 3 methods. - There is also the potential for observation bias, although a well designed questionnaire should reduce this In-Person interviews is the most expensive method, and takes much more time than phone interviews.. The extra time, however, allows the highest response rate and the most understanding of the more complex questions. It is hard to get participants to open up to sensitive questions during an in-person interview. Telephone: - Still have person interaction, but it is less costly; it also takes much less time to contact many more individuals - Because it is not face-to-face, there is a higher rate of non-response. - It is also harder for the interviewer to gain as much trust and report with the participant – telephone interviews lend themselves to a feeling of hurriedness Telephone interviews will get you information the fastest, and reaches a larger population than in-person. It is also less expensive than in-person interviews, and participants prefer discussing sensitive topics over the phone. Due to the nature of phone interviews, the more complex questions are not always easily grasped without visual clues, and the length of the interview and response rate are low. Postal: - The least expensive method; can reach many people in as long as it takes for the mail to be delivered - However, because it is self-administered, there is a large rate of non-response - Any complex questions cannot be explained Self-administered questionnaires are by far the least expensive. It is also the best way to obtain sensitive information. People feel much more comfortable answering these questions anonymously in the comfort of their own homes. However, self-administered is the slowest way for finding data. Between the mail system and non-response, questionnaires will only very slowly find their way back into the researchers’ hands. It is very difficult to estimate population coverage, and understanding of complex questions; and there is really no way to know how long the interviewee took to complete the questionnaire.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 预调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

问题的类别 Types of Questions 事实性的问题 Factual 人口学，实验室，临床信息 Demographic, laboratory, clinical information 观点与态度 Opinions and Attitudes 与研究有关的信念，想法，前设，价值观，等等 Beliefs, ideas, predispositions, and values related to the topic under study 信息，知识 Information/knowledge 检验的是被调查者知道什么（有没有某种知识） Measures what people know (presence or absence of knowledge) 正确与错误的知识 Correct or incorrect knowledge There are several types of question categories including the following: factual; opinion and attitudes; knowledge and information; behavior and practice; and constructs. Examples of factual questions are demographic, laboratory values, clinical information and so forth. Opinion and attitude questions might address beliefs, ideas, predispositions, and values etc. which could be particularly relevant to the topic under study. Information and knowledge questions might be used to assess either correct or incorrect knowledge. Example of a question assessing knowledge and misconception Choose the right answer(s): HIV/AIDS can be transmitted through Hand shake Shared food Blood transfusion Sexually

问题的类别 Types of Questions 行为 Behavior 对行为进行描述 Description of behavior 人们报告的行为与实际行为是否符合很难判断 Difficult to evaluate how people think they behave vs. what they actually do 量表 – 对被调查者问一系列问题，合起来对某一变量进行描述 Constructs – responses to multiple questions are combined to estimate a particular variable 社会经济状况量表 Socio-economic status (SES) 生活质量量表 Quality of life 改变某一行为效力的量表 Self efficacy to change a behavior Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

剂量-效应关系 Measuring Dose-Response Relationship 洗手频度与诺如病毒感染发病的关系 饭前是否洗手？ □ 每次 □ 经常 □ 偶尔 □ 从不 上厕所后是否洗手？ □ 每次 □ 经常 □ 偶尔 □ 从不 玩耍后是否洗手？ □ 每次 □ 经常 □ 偶尔 □ 从不 洗手时是否用肥皂？ □ 每次 □ 经常 □ 偶尔 □ 从不 平均每天洗手几次？ ___次 喝豆汁与霍乱发病的关系 （调查员用统一的杯子）像这样大小的杯子，你喝了几杯豆汁？ ___杯 麻疹传播在网吧中的传播 从开学到发烧前的一段时间里，你平均每天在网吧上网时间： “乾坤”网吧：每天___小时 “天天”网吧：每天___小时 “网虫”网吧：每天___小时 Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

病例对照研究中的调查表 Measuring Dose-Response Relationship 用尽量相同的调查表 问的方法要一致，如：病例用电话调查，对照也用电话调查 问的次数要一致 Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

协助工具 Aids to recall 统一的容器 照片 帮助回忆日期 这样大小的水杯，喝几杯水？ 不常吃的食物 饮用水的品牌 日历 必要时使用农历 使用重要事件或日期（如春节，结婚）帮助记忆 Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

某村庄甲肝爆发 A hepatitis A outbreak Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

待验假设：11月后的高峰是由于饮用某个品牌桶的装水导致点源爆发 如何询问病例和对照的暴露史？ 需要考虑的问题： 甲肝的潜伏期很长 很难回忆3天以上的饮食史 病例对饮食的回忆比对照要好 Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

询问习惯性的问题 Ask Preference Questions 你平常的饮水类型是： 桶装(饮水机)水： □经常 □有时 □偶尔 □很少 □从不 自来水： □经常 □有时 □偶尔 □很少 □从不 自家井水： □经常 □有时 □偶尔 □很少 □从不 饮水习惯： 如喝桶装水，习惯于喝： □热水 □温水 □凉水 □兑着喝 如喝自来水或井水，习惯于喝：□开水（热或凉） □生水 □生水、开水兑着喝 Behavioral questions may be difficult to evaluate since people may behave differently than they realize. Constructs make use of multiple responses to compute a single outcome. To assess constructs you use several questions and the answers combined will give a score reflecting the construct, suchas SES, quality of life, and self-efficacy to change a behavior.

调查表设计中的常见问题 Common Problems 模棱两可 Ambiguous questions 问题不清楚 Not self-explanatory 一道题中问两个问题 Two questions in one (compound questions) 语言复杂 Complex words 所问的问题无法回忆 Asking things that people won’t remember 假定的问题 Hypothetical questions 可供选择的类别不够 Not enough categories 太长，或有太多的概念 Too long or too many ideas 类别之间有重复 Overlapping categories 开场白有问题 Leading questions 问题的准确度无法界定 Degree of accuracy not specified There are many common problems when designing questions. Most fall within these categories: -Ambiguous questions -Not self-explanatory -Two questions in one -Complex words -Asking things that people won’t remember -Hypothetical questions -Not enough categories -Too long or too many ideas -Overlapping intervals -Leading questions -Degree of accuracy not specified

解决办法 Questionnaire Solutions 使用简单的语言 Simple language 采用封闭式的问题 Closed questions 将问题分成几步来问 Staged questions 将答案分成等级 Scaling 提供图像帮助 Visual clues 预调查 Pilot tests Fortunately there are these solutions to the problems. Use simple language. Here again is the theme of Keep it Simple. Try to avoid complex sentence; avoid technical terms; etc. Closed questions generally simplify the question by having the respondent choose an answer. Staged questions are used to break up multiple thoughts in one question. e.g. Instead of “When you travel by plane do you get airsick?”, ask “Do you ever travel by plane? If yes, do you ever get airsick?” Scaling enables the respondent to quantify an otherwise abstract feeling. Visual clues: can use a card with multiple answers on it for closed questions with many possible choices Pilot tests enable the researchers to change, delete, add new questions for easier understanding.

注意界定资料格式 Be clear about the data format you want! 格式不确定 Unspecified Format 日期 Date: 年 月 日 9 4 年龄Age: 岁year 月month 确定的格式Specified Format 1996 10 12 日期Date: 12/10/06 10/12/06 06/10/12 ? ? 年龄 Age: 9 （是9岁还是9个月？）(What? Years or months) Firstly, it is important to be very clear about the data format you want. As you can see, there are multiple ways for a respondent to answer some simple questions. Without explicit instructions, your data will not be accurate –> valid.

问题的格式 Format of questions 封闭式：答案已预先确定 Closed format  give a ‘fixed’ response 简单、易填 Simple and quick; easy to use in the field 减少了文化程度低的人回答的障碍 Reduces discrimination against those who are less literate 容易进行编码、录入、定量分析 Easy to code, record, and analyze results quantitatively 容易报告结果 Easy to report results 限制了收集数据的种类和量 Limits the variety and amount of data collected There are 2 main question formats in a questionnaire: Closed and Open. Closed is a fixed response: can only tick off boxes. - simple, easy to code, easy to report results; but limits the variety and amount of data collected

问题的格式 Format of questions 开放式: 允许自由发挥 Open-ended format  allows exploration of issues 当需要让被调查者先提供信息，然后再问特定问题时 When there is interest in what information a subject will volunteer before asking a specific response 当进行定量研究，或需要文字描述性数据时When conducting qualitative research; when verbatim responses are needed in a report 当可能答案太多而无法界定时仍可使用 Can be used even if a comprehensive range of alternative choices cannot be compiled 不容易实施和记录 Not easy to administer and record 资料编码、分析和定量分析时比较困难 Difficult to code responses, analyze data, and quantify results Open is generally a single question allowing people to express their views in their own words - Advantages: - useful in hypothesis generating, particularly when there is limited knowledge about the response - Used when pretesting to organize captured responses into categories - When there is interest in what information a subject will volunteer before asking specific responses - when conducting qualitative research - when verbatim responses are needed in a report Disadvantages - Not easy to administer and record - Difficult to code responses, analyze data, and quantify results

开放式问题 Open Question 你吃/喝过那些奶制品？ ________________________ What milk products have you eaten? ________________________ Example of Open question

封闭式问题：是/否 Closed question: Yes / No 你有以下症状吗？Did you have any of the following symptoms ? 是Yes 否No 不清楚 Unsure 腹泻 Diarrhea 发热 Fever 腹痛 Abdominal pain 呕吐 Vomiting 其他 Other 1 2 3 Example of Closed question with Yes/No response

分阶段的问题 Staged Questions 在那个聚会中你是否喝苹果汁，喝了多少？Did you drink the apple juice at the party and how much did you drink? _______________ 或 OR 你有没有喝苹果汁？ Did you drink any apple juice? 有Y 没有N 如有，喝了多少？If “yes” then how much did you drink? 1杯1 cup 2杯2 cups 3杯3 cups 4杯或以上4 cups+ Staged questions: Technically, the first question would fall under a “common problem” question. This question “Did you drink the apple juice at the party and how much did you drink?” combines 2 questions into 1: “Did you drink apple juice?” and “How much did you drink?” The following question is how a question of this nature should be asked: First if you performed the action, and if yes, how much. In this case, “Did you drink?” Yes/No If yes, how much?

Other Types of Structured Questions 其他形式的结构性问题 Other Types of Structured Questions 命名性答案 Nominal response: 病人的性别 What is the patient’s gender? 1= 男male 2=女 female 定序性答案 Ordinal response (useful in ranking): 请把你对下列食品的爱吃程度进行排序，1=最爱，5=最不爱 Please rank the following food items in degrees of how much you would like to eat with 1 being the highest and 5 being the lowest? __ 鸡肉 Chicken __ 牛肉 Beef __ 马肉 Horse __ 鱼肉 Fish __ 羊肉 Sheep Nominal response: The response is a Nominal variable. Here it is gender. Another example would be eye color, religion, ethnicity, etc. Ordinal response: Response requires ranking

用等级来衡量 Using Scales 请描述你疼痛的程度 How would you characterize the pain that you had? 轻微Mild 中度 Moderate 重度Severe 或 OR 请将你的疼痛程度按0-10分级 How would you characterize the pain that you had on a scale of 0 to 10? 1 2 3 4 5 6 7 8 9 10 Scales – quantifies the abstract: can use checkboxes or a sliding scale. Checkboxes are easier for coding and analysis 无痛 No pain 极剧烈 Very severe pain

常见问题与陷阱 Problems and Pitfalls 避免“一题二问”；否则你无法知道回答者回答的是那件事 Avoid questions that ask two things at once - you won’t know which ‘bit’ people are answering: 例如：你有脖子痛和腹泻的症状吗？有/没有 e.g. Have you ever had neck pain and diarrhea? Yes/No? 模棱两可 Ambiguity..... 例如：你上厕所次数很多吗？ e.g. Do you go to the toilet a lot? Two things at once: This example asks about 2 symptoms. Can ask the class how to fix the question: Simple way to fix the problem is to simply to split the question up: Have you ever had neck pain? Have you ever had diarrhea? Ambiguity: “A lot” is a phrase that can vary from person to person. Ask the class how to fix it: Need to quantify “a lot”: e.g. How often do you go to the toilet? 3 times per day, 5 times per day, 10 times per day, >10 times per day?

常见问题与陷阱 Problems and Pitfalls 避免使用“行话”Avoid jargon 你参加过艾防知行活动吗？ “艾防知行”=艾滋病防治知识与行为干预 你打过流感减毒活疫苗吗？ 许多人不知道什么是流感减毒苗，即使知道，可能也无法知道所打的是哪种疫苗 避免各组间的重叠 Avoid options that are not mutually exclusive when they need to be 你的年龄段 What age are you? 16-20  20-25  25-30  35-40  Jargon = “doctor” talk. Don’t use technical words for the anatomy, medical procedures, acronyms, etc. Not mutually exclusive – ask class to fix the age ranges: e.g. 16-20; 21-25; 26-30; 31-35; 36-40

常见问题与陷阱 Problems and Pitfalls 避免暗示性问题 Avoid leading questions 你认为是旅馆里的食物让你得病的吗？Do you think that the food in the hotel made you sick? 你是不是觉得旅馆里的服务员看起来很不卫生？Did the hotel staff seem unhygienic to you? 你觉得医院里的工作人员看起来是不是很累？ Do you agree that the hospital staff were close to exhaustion? 避免问卷过长 Avoid making the questionnaire too long 避免错别字 Typographical / spelling errors 学校的卫身环境如何 有没有提供洗手夜 发病前用了那些抗升素 有没有打过易苗 Leading questions: Basically, don’t put ideas into the participants’ heads. To correct the above questions: Where did you eat last night? List 4 locations with one being the hotel Were the bathrooms in the hotel always supplied with soap and tissues? How long do you think is the average shift of the hotel staff? 4 hours, 5 hours, 6 hours, 7 hours Too long: Again the key to a good questionnaire is to keep the questions Short and Simple Spelling errors: Always review your own questions, and have at least 2 other people review them also.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

问题的顺序 Sequence of Questions 将问题按照题目或回答方式来分组 Group questions by topic or response options 问题的顺序很重要 Question order is important: 由一般到特别 General to particular 由易到难 Easy to difficult 由具体到抽象 Factual to abstract 从封闭式的问题开始问 Start with closed format questions 从与本次调查有关、但敏感度小的问题开始问 Start with non-threatening questions relevant to the main subject 不要把最重要的问题放在最后问 Don’t put the most important item last Topics: Demographic; Food; Location; Contacts Response options: Yes/No; Checkboxes; Open General to particular – start with questions that apply to everyone Easy to difficult – “What is your name?” vs. “How many servings of dairy did you have in the past 2 weeks?” Factual to abstract – “When did you check into the hotel?” vs. “How long do you think is the average shift of the hotel staff?” Closed format Non-threatening questions relevant to main subject – if food poisoning: “When did you first develop symptoms?” Don’t put the most important item last! By the end the subject will be tired and not as fresh to answer as completely as he may have at the beginning of the interview.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

调查表的布局 Layout: General 对邮寄调查尤其重要 Particularly important with postal questionnaires 设计美观 Attractive 语言及填写说明要清楚 Clear language and instructions 易读易填 Easy to follow 跳项要非常清楚地标明 Well illustrated jumps/skips 尽量设计得短而易 Keep it short and simple An easy-to-read layout is important for both a self-administrator and the telephone or in-person interviewer. The paper should be attractive. This may sound a bit odd, but a basic white piece of paper with regular 10-point font will not grab the attention of a participant – especially for postal questionnaires. Have the title of the study in bold print at the top of the page. If possible, use colored paper. Mix up the fonts between bold, CAPS, italics, underline. The questionnaire needs to have explicit instructions in clear language. Sentences should be kept short. If there are jumps or skips, e.g. What is your gender? If male, skip to Question #5, make sure anyone reading the questionnaire understands the structure.

调查表的布局 Layout: General 一些实用窍门 Practical tips 一些让调查员容易执行的窍门 Make it easy for survey team members in the field 要有足够的行距及空白区 Keep form well-spaced and open 留出空白区让调查员记录资料 Make spaces for recording data obvious 问题不要窜页 Each question needs to fit wholly on a page 一些方便录入的窍门 Make it easy for keypunching 不要太挤太密 Place spaces for recording data 填写资料的位置在视觉上要尽量一致 Record data in consistent visual framework “不详”项要用统一的代码，如“9”或“-” Use common symbol for unknown: "9", "Unk", "-", etc. Note to translator: “Unk” = Unknown (How could that be abbreviated in Russian? Georgian? Etc.)

布局举例 Layout example 儿童甲型肝炎调查表 一般信息 1. 儿童姓名：_________ 家长姓名：_________ 联系电话：__________________ 2. 年龄 岁 性别： □ 男 □ 女 3. 家中其他12岁以下儿童: 1) __岁, □男 □女；2个月内是否患甲型肝炎：□ 是 □ 否 2) __岁, □男 □女；2个月内是否患甲型肝炎：□ 是 □ 否 3) __岁, □男 □女；2个月内是否患甲型肝炎：□ 是 □ 否 4. 职业: □ 散居儿童 □ 托幼儿童 □ 小学生 学校或幼儿园名称：□ 西光录小学 □ 东光录小学 □ 育苗幼儿园 □ 好孩子幼儿园 卫生习惯 6. 吃水果时是否洗后再吃？ □ 总是 □ 经常 □ 偶尔 □ 从不 吃水果时是否削皮吃？ □ 总是 □ 经常 □ 偶尔 □ 从不 7. 饭前是否洗手？ □ 总是 □ 经常 □ 偶尔 □ 从不 上厕所后是否洗手？ □ 总是 □ 经常 □ 偶尔 □ 从不 玩耍后是否洗手？ □ 总是 □ 经常 □ 偶尔 □ 从不 8. 洗手时是否使用流动水？ □ 总是 □ 经常 □ 偶尔 □ 从不 洗手时是否用肥皂？ □ 总是 □ 经常 □ 偶尔 □ 从不

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

翻译 Translation 最好用被调查者的语言来调查 The best way to interview people is to address them in their own language 研究这可以用他/她自己的语言来设计调查表Researcher makes questionnaire in his language 然后翻译成当地语言 Translation into local language 让另一个人回译以确定调查表的准确性 Back-translation (check) 调查员应该完全熟悉被调查者的语言 Interviewers should be fluent in the language of the interviewees The researcher will make the questionnaire in his own language so that he knows and approves the content. When complete, the questionnaire will be translated into the local language, and then will be back-translated as a check of the translation. Interviewers should be fluent in the language of the interviewees. It is a good idea to test a translated questionnaire on a sample audience to be sure the locals hear the questions the way the researchers intended.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

培训调查员 Training Interviewers 熟悉调查表 Need to understand the questionnaire 名词含义 Terminology 跳跃 Skips 定义 Definitions 要求调查员只读调查表，不要作解释 Only read the questions, not interpret them for the respondent 熟悉如何处理知情权的问题 Master the consent process 用一种平静、非指责性的方式来询问，以避免调查员偏倚 Remove interviewer bias by asking everyone questions in calm, non-judgemental fashion The interviewers need to learn the questionnaire backward and forward. Interviewers should understand and be able to define all the terminology, and know all the skip patterns. It is their job to earn the trust of the respondent. Only when the respondent feels trusted will they give consent for the interview. The interviewer finally needs to be able to ask the same questions of everyone in the same non-judgmental fashion so that any interviewer bias can be removed.

培训调查员 Training Interviewers 在需要完成的每个环节上都要进行培训 Interviewers should be trained in all tasks required to conduct the interview 寻找被调查者的家庭所在 Find the household to interview 友好地接近被调查者，并取得被调查者的同意 Approach people pleasantly and acquire consent 尽可能为被调查者保密 Assure confidentiality and anonymity (when possible) 如何按调查表询问，如何回答问题 Administer questionnaire, and respond to questions 调查完毕以后，对调查表重新检查一遍 Review filled out questionnaire forms Training the interviewers: These are some of the tasks, and skills the interviewer is required to learn before actually collecting data in a survey. Find the household to interview Pleasantly approach people and acquire consent Assure people of confidentiality and anonymity Administer questionnaire, and respond to questions Review filled out questionnaire forms for errors and inconsistency in the responses. Recognition of inconsistencies early and while in the field gives the interviewer the opportunity to go back to the participants and correct inconsistent responses.

调查表设计的十个步骤 Steps in questionnaire design 确立调查内容或待验假设 State the research question(s)/hypotheses 拟定分析计划 Outline an analysis plan (“empty” tables) 列出需要要测量的变量 List the variables to be measured 确定调查如何进行 Decide how the questionnaire will be administered 拟定要问的问题 Formulate the questions 问哪些问题？Decide on the questions 如何表述？ Phrase the questions 确定问题的顺序 Determine the sequence of questions 设计调查表的布局 Plan the layout and design of the questionnaire 必要时把调查表翻译成当地语言 Translate if necessary 培训调查员 Train interviewers 试调查 Pilot test There are 10 steps in designing a questionnaire. 1st step: State the research question

Why pretest questionnaires? 为何要试调查？ Why pretest questionnaires? 使调查表更明白易懂 Ease of understanding 消除模棱两可的问题 Avoid ambiguity 发现其他未考虑到的暴露因素 Identify other exposures 澄清可供选择的答案 Clarify response categories 消除非必要的问题 Eliminate unnecessary questions 了解那些敏感的问题 Understand sensitive issues 估计所需时间 Estimate time needed 发现后勤上可能存在的问题 Uncover logistical problems The last, but crucial step of creating a questionnaire is the Pretest, or Pilot test. The pretest is a massive quality control. It is done so that any kinks in the questions can be worked out: e.g. too many questions; response categories don’t match the question asked; questions are ambiguous; question is too complex; open-ended would be better as a closed question, or vice-versa; etc. The pretest is a longer process that includes people on the research staff, general test subjects, and a target audience of the population that the questionnaire is intended for.

试调查 Pretest 第一步在调查员中进行试调查，发现问题后对调查表进行修改 Administer questionnaire to other survey staff and correct questionnaires 认真测试每个问题 Go over each question 问题读起来是否拗口？Ensure questions are easily readable 跳跃是否清楚合理？Be sure skip patterns clear The first step if for a review process. The entire staff will go one by one through the questions to verify they are correct, the questions are readable, the skip patterns are clear, it is audience appropriate, etc.

试调查 Pretest 一定要在被调查的人群中试用调查表！Always pretest on target population 试调查时监管员要在场观察 Supervisors should observe pretest 对资料搜集的全部过程进行测试 Pretest entire data collection procedures 征求被测试者的看法 Ask respondents about interview questions 征求调查员的看法 Ask survey workers about 对具体问题的看法 Individual questions 对整个资料搜集过程的看法 Entire process of data collection The target audience always needs to be involved in the pretest! The trained interviewers should be giving the questionnaire with supervisors present to observe. If there are any data collection procedures as part of the questionnaire, that too should be tested, e.g. blood samples. Once the target audience has completed the interview process, the supervisors should talk to them about the questions and ask them if they have any suggestions: if there were any questions they didn’t feel comfortable answering, if any were too complex, etc. The supervisors should also talk to their interviewers and other survey workers for any insight they may have.

试调查要回答的问题 Pretest answers these questions 被调查者愿意回答这些问题吗？Are people willing to answer questions? 被调查者允许调查者采集标本吗？ Are people willing to allow specimen collection? 有没有难以回答的问题？Are any questions difficult to answer? 有没有很敏感的问题？Are any questions sensitive? 被调查者明白这些问题吗？Do people understand every question? 调查员能看懂调查说明吗？Can interviewers follow instructions on data collection form? 调查表有无其他难点？Are there other difficulties with the data collection form? 有没有漏掉重要问题？Are there any answers that should be included on data collection form? 一份调查表要多长时间才能完成？How long does data collection take? 如果一个家庭里要调查几个人，按什么顺序？What is best organization of data collection steps within household? Read above list.

调查表的优点 Advantages of questionnaires 可以相对经济、有效地对大量被调查者进行调查（尤其是用邮寄法时）Can reach a large number of people relatively easily and economically (especially postal questionnaires) 可以得到定量数据 Provide quantifiable answers 分析起来比较容易 Relatively easy to analyze Advantages of questionnaires: Can reach a large number of people relatively easily and economically (especially postal questionnaires) Provide quantifiable answers easy to code and analyze Relatively easy to analyze faster analysis yields a faster answer to the question – especially important in an outbreak investigation situation

调查表的缺点 Disadvantages of questionnaires 对问题了解的深度可能不够 May provide only limited insight into the problem: 回答时可以回答的答案有限 Limited response allowed by questions 也许没有问到点子上 You may not have asked the right questions 可能会答非所问 Varying response 被调查者可能会误解你的问题而答非所问 They may not understand your question and may interpret it differently 邮寄调查表应答率常较低 Postal questionnaires  low response rate 如资料错漏，很难弥补；因此应设法一次成功 Hard to chase after missing data – need to get it correct the first time Disadvantages of questionnaires: May provide only limited insight into the problem: Limited response allowed by questions You may not have asked the right questions e.g. Andorra, norovirus and ice-cubes Varying response They may not understand your question and may interpret it differently Postal questionnaires  low response rate Hard to chase after missing data – need to get it correct the first time

其他有关问题 Other Issues 注意借用以前使用过的调查表 Look at previously used questionnaires 使用标准的、已验证过的指标 Standard validated measurements 伦理问题 Ethics 观察与反应 Observations vs. Responses 问题→实际的回答→记录的回答 Question/actual response/recorded response 资料的编码与分析 Coding for analysis Other ways to aid in the creation of a questionnaire is to look at existing questionnaires that ask a similar question.

这个问题合适吗？如何改进？ 我们的调查发现，在麦当劳就餐与这次沙门氏菌的爆发有关。你同意吗？ □是 □否 我们的调查发现，在麦当劳就餐与这次沙门氏菌的爆发有关。你同意吗？ □是 □否 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个如何？ 请列出你在腹泻前一周内用过餐的饭馆： ____________ The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题合适吗？如何改进？ 你好，我是市疾控中心的，调查这次沙门氏菌的爆发与哪些餐馆有关。请问你在发病前一周有没有在麦当劳或其他餐馆用餐？ □是 □否 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个如何？ 你好，我是市疾控中心的，调查这次沙门氏菌的爆发与哪些餐馆有关。请问你在发病前一周有没有在以下饭馆用餐？ 麦当劳： □有 □没有 □记不清了 肯德基： □有 □没有 □记不清了 而新快餐店： □有 □没有 □记不清了 部维生饭店： □有 □没有 □记不清了 泰展烤鸭： □有 □没有 □记不清了 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题合适吗？如何改进？ 你的感官器官，如眼，耳，鼻子有没有受到影响？ □是 □否 □是 □否 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个如何？ 请问你以下感觉器官有没有受到影响？ 视力： □有 □没有 听力： □有 □没有 嗅觉： □有 □没有 视力： □有 □没有 听力： □有 □没有 嗅觉： □有 □没有 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题合适吗？如何改进？ 在发病前一个月你有没有嫖过娼？ □是 □否 在发病前一个月你有没有嫖过娼？ □是 □否 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个如何？ 请问你在发病前一个月内有没有花钱与其他女人发生性关系？ □有 □没有 请问你在发病前一个月内有没有花钱与其他女人发生性关系？ □有 □没有 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个如何？ 过去10年内，你多长时间吃一次土豆？ □从未吃过 □每周吃一次 □每年吃一次 □每月吃一次 □从未吃过 □每周吃一次 □每年吃一次 □每月吃一次 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题如何？ 你有没有感染过艾滋病并接受治疗？□ 有 □ 没有 你有没有感染过艾滋病并接受治疗？□ 有 □ 没有 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题如何？ 请问你的年龄是多少？ □ 12岁以下 □ 25-35 □ 12-19 □ 35-45 □ 19-25 □ 45岁以上 □ 12岁以下 □ 25-35 □ 12-19 □ 35-45 □ 19-25 □ 45岁以上 The following are some examples of poorly designed questions and the ways to improve them. Instructor: These are optional slides to use when you think there is some more time for examples. You might want to print out these examples and give them to the participants.

这个问题如何？ 食品名称 炸鸡块 □ 有 □ 没有 □ 不知道 炒鸡蛋 □ 有 □ 没有 □ 不知道 蛤蜊汤 □ 有 □ 没有 □ 不知道 炸鸡块 □ 有 □ 没有 □ 不知道 炒鸡蛋 □ 有 □ 没有 □ 不知道 蛤蜊汤 □ 有 □ 没有 □ 不知道 How would you improve this question? Issues: Directions are unclear due to use of an incomplete sentence. Need to be more specific and clear. Ask, “Have you eaten the following food items in the previous week?”

这个问题如何？ 你多长时间喝一次井水？ □ 从不喝 □ 几乎从不 □ 极少喝 □ 偶尔喝 □ 有时喝 □ 常常喝 □ 从不喝 □ 几乎从不 □ 极少喝 □ 偶尔喝 □ 有时喝 □ 常常喝 □ 几乎每周喝 □ 每周喝 □ 每周喝多次 □ 几乎每天喝 □ 每天喝 □ 每天喝几次 How would you improve this question? Issues: Too many response options. Respondents may distinguish between response options in different ways.

这个问题如何？ 你每天喝多少水? _________ How would you improve this question? Issues: Desired response format is unclear. Respondents may answer in various formats: glasses, ounces, liters, etc. Respondents that answer in terms of glasses may have different sized glasses in mind. Better to state the desired response format in the question.

这个问题如何？ （禽流感病毒感染调查中的问题） 你是否认识感染过禽流感病毒的人并与之有过接触？ □ 有 □ 没有 □ 不知道 你是否认识感染过禽流感病毒的人并与之有过接触？ □ 有 □ 没有 □ 不知道 An example of 2 questions in one. Avoid using such questions.

这个问题如何？ （梅毒调查中的问题） 你有没有停止嫖妓？ □ 有 □ 没有 你有没有停止嫖妓？ □ 有 □ 没有 This is an example of a loaded question. It implies something about the person being asked which if not correct leaves him/her without an accurate option for answering the question. In addition to being loaded this sensitive topic may offend some of the people. Note: this example stimulates discussion, yet might offend some people.

这个问题如何？ （霍乱调查中的问题） 你是否像大多数爱清洁的人那样，饭前洗手？ □是 □ 否 □是 □ 否 This is a leading question. The respondent is being led to respond yes.

这个问题如何？ 你有没有接触过很多病人？ □ 有 □ 没有 □ 不知道 This is an ambiguous question. □ 有 □ 没有 □ 不知道 This is an ambiguous question. The degree of accuracy is not specified. Leads to miscommunication.

这个问题如何？ （甲肝调查中的问题） 在开始生病前45天内你有没有 This is an example of a loaded question. It implies something about the person being asked which if not correct leaves him/her without an accurate option for answering the question. In addition to being loaded this sensitive topic may offend some of the people. Note: this example stimulates discussion, yet might offend some people.

THANK YOU

Extra Slides At the end of the session, participants will be able to List objectives of a questionnaire Describe advantages and disadvantages of questionnaire List steps in questionnaire design Describe different types of questions and when to use them Identify the characteristics of effective questionnaire Define pilot testing Describe ways to avoid common problems Explain sensitive issues related to translating questionnaire

Learning Objectives List objectives of questionnaire Describe advantages and disadvantages of questionnaire List steps in questionnaire design Describe different types of questions and when to use them Identify characteristics of effective questionnaire Define pilot testing Describe ways to avoid common problems Explain sensitive issues related to translating questionnaire At the end of the session, participants will be able to List objectives of a questionnaire Describe advantages and disadvantages of questionnaire List steps in questionnaire design Describe different types of questions and when to use them Identify the characteristics of effective questionnaire Define pilot testing Describe ways to avoid common problems Explain sensitive issues related to translating questionnaire

A Questionnaire is a Way to Measure It is a list of questions (answered by a respondent) to generate an indirect measure of the variables under investigation. Questions regarding: - exposure - outcome - both Design and administration of a questionnaire is critical part of study. Poor design or administration can decrease the reliability and the validity of a measure. What is a questionnaire? And why do we need use them? A Questionnaire is a way to measure It is a list of questions (answered by a respondent) to generate an indirect measure of the variables under investigation. Questions regarding: Exposure Outcome both Note: Questions seek to generate information on different Variables such as exposure of interest, outcome or disease or interest, or both. Design and administration of a questionnaire is critical part of study. Poor design or administration can decrease the reliability and the validity of a measure.

变量的类别 Types of Variables 定性或分类 变量 Qualitative or categorical 定量测量 Quantitative measurement It is important to remember the types of variables when designing the questions. Categorical variables cannot be measured numerically. Nominal: Nationality – Georgian, Turkish, Russian Ordinal: Response to treatment – improve, no change, worsen Quantitative variables can be measured numerically. Discrete: Count data – 1,2,3,4,…, n; has to be whole numbers Continuous: Data can be described differently depending on how good your measure is, i.e. height can be 2.1 or 2.13435456 meters depending on the measure 命名变量 （无顺序，如国籍） Nominal (not ordered) e.g. nationality 有序变量 （例：对治疗的反 应等级） Ordinal (ordered) e.g. response to treatment 离散型数据 （例：住院人数） Discrete (count data) e.g. number of admissions 连续变量 （实际测量数值） （例如，身高） Continuous (real-valued) e.g. height

Objectives of a questionnaire To ensure that the information obtained is as accurate and as relevant to the study as possible To maximize the proportion of subjects answering the questionnaire – as high a response rate as possible To get a high response rate: make the questionnaire relevant; ensure participants wee its potential and usefulness e.g. “we are trying to find out why you and your family became ill after attending a wedding”