RESEARCH/SCIENCE APPROACHES

Size: px

Start display at page:

Download "RESEARCH/SCIENCE APPROACHES"

晤椅汪
5 years ago
Views:

1 第五章 Chapter 5 总体和样本 Population and Sample 冯先琼 PhD,RN

2 学习目标 :Learning Objectives (1) 掌握抽样的过程 (2) 掌握抽样的原则 (3) 掌握抽样的方法 (4) 熟悉总体样本和误差的基本概念 (5) 了解样本含量的估计方法 2

3 Keywords In This Chapter Population Sampling Target population Accessible population Representative sample Inclusion Criteria Exclusion Criteria Systematic error Random error Nonprobability Sampling Convenient Sampling Quota Sampling Purposive Sampling Network Sampling Probability Sampling Simple Random Sampling Systematic Sampling Stratified Random Sampling Cluster Sampling 3

4 第一节基本概念总体 (Population) 是根据研究目的确定具有相同性质的个体所构成的全体例 : 心脏病人的生活质量 -- 所有心脏病人就是研究的总体研究的总体可以是人病历标本等 4

5 第一节基本概念目标总体 (Target population) 是符合条件的所有个体的集合体, 是研究者所要推论的整个的集合体 The entire set of cases about which the researcher would like to make generalizations 可得总体 (Accessible population) 是研究者根据研究的需要能方便抽取的总体 Available population that meets the criteria 5

6 第一节基本概念样本 (sample) 是指从总体中抽取的部分观察单位抽样的目的是用样本信息来推断总体特征 is a set of elements that make up the population 样本的代表性 (A representative sample) Is one whose main characteristics closely approximate those of the population 6

7 第一节基本概念系统误差 (Systematic error) 是由某些不准确定量的较恒定的因素引起的 : 如来自受试者观察者仪器设备外环境因素随机误差 (Random error) 由于观察单位间存在个体差异等引起的测量误差抽样误差 (Sampling bias) Is the systematic over or underrepresentation of some segment of the population in terms of a characteristic relevant to the research question 通常属于随机误差 7

8 总体与样本 8

9 第一节基本概念纳入和排除标准 (Inclusion and exclusion criteria) Determine the subjects used in a study by defining the criteria used to include or exclude a subject from a study 用于控制外在 / 干扰变量 9

10 Example: inclusion criteria Age 4 to 17 years Diagnosis acute lymphocytic leukemia within 1 month of diagnosis Health status no other existing chronic illness associated with pain, a fulminating disease, or a known cognitive disability Status ability to cope with the burden of research tasks as determined by the primary nurse Language English or Spanish speaking Van Cleve et al. (2004). 10/18/2017

11 第二节抽样过程及方法抽样 (Sampling): 是指从总体中抽取部分观察单位以获得样本的过程抽样的目的 : 提高研究的效率抽样最终希望达到 : 抽取出有代表性的样本 11

12 抽样过程明确总体列出抽样标准选择合适的样本量及抽样方法选择样本 12

13 抽样原则确保样本的可靠性对研究对象的选择有明确的诊断标准纳入标准排除标准选取有代表性的样本应用随机合适的样本量 13

14 抽样方法非概率性抽样 (Nonprobability Sampling): 抽样时没有采取随机化的方法总体中的每一个研究单位被抽取进入样本的概率是不确定的 --- 非概率性抽样! 常见的非概率性抽样方便抽样 (Convenient sampling) 配额抽样 (Quota sampling) 目的抽样 (Purposive sampling) 网络抽样 (Network sampling) 14

15 非概率性抽样方便抽样 - Convenient sampling 用最容易找到的人 / 物作为研究对象优点 : 方便易行缺点 : 所抽取样本的代表性差, 容易出现抽样误差是抽样方法中准确性和代表性最差的一种方法 15

16 非概率性抽样配额抽样 (Quota Sampling) 利用总体内各层的构成比抽取与总体相似的样本 Knowledge about characteristics of the population of interest is used to build representativeness into the sample 16

17 非概率性抽样目的抽样 (Purposive Sampling) 调查者有意识地选择一些有代表性的样本 Researcher selects subjects who are considered to be typical of the population 多用于探索性前瞻性研究 17

18 非概率性抽样网络抽样 (Network Sampling) 又称滚雪球抽样 (Snowball sampling) 利用网络优势和朋友间推荐来发展样本用于寻找特殊总体中的个体 18

19 Nonprobability Sampling Advantages and Disadvantages Lack of representative Sampling bias: threat validity Convenience Economy Practical Easy to carry out 19 10/18/

20 概率性抽样 Probability Sampling 利用随机化方法抽取样本 Each element of the population has an equal and independent chance of being included in the sample 随机抽样常见的随机抽样法单纯随机抽样 (Simple random sampling) 系统抽样 (Systematic sampling) 分层随机抽样 (Stratified random sampling) 整群抽样 (Cluster sampling) 20

21 概率性抽样单纯随机抽样 Simple random sampling Researcher first defines the population (a set) Lists all of the units of the population (a sampling frame) Selects a sample of units (a subset) from which the sample will be chosen 常用 : 抽签法随机数字表电脑随机数字 21

22 概率性抽样系统抽样 Systematic sampling 又称等距抽样 Systematic Sampling involves the selection of every kth case from a list Eg: every 5 th, every 22 th.. 抽样具体方法 : Determine the size of the population: 5000 Determine the size of the desired sample: 50 Calculate the sampling interval: 5000/50=100 Select the first case randomly: 72 Then 172, 272,

23 概率性抽样分层随机抽样 Stratified random sampling Similar to quota sampling but has random selection. 研究者先根据总体某种特征分成若干层, 然后每一层内按照比例随机抽取一定数量的个体 23

24 概率性抽样整群抽样 Cluster sampling 将总体中所有的个体按某种属性分成若干个群体, 再从所有的群体中随机抽选一部分群体构成样本多级抽样 Multistage sampling 24

25 Probability Sampling Advantages and Disadvantages Obtaining representative samples Inconvenience Complexity Impractical 抽样误差的大小 : 分层抽样 < 系统抽样 < 单纯随机抽样 < 整群抽样 25 10/18/

26 第三节样本含量的估计与样本含量相关的一些参数 : 1. 检验水准 ( ) 本次研究允许第一类错误发生概率的大小 ( 假阳性率 ).05 与.01 若取.01, 则样本量要求更多! 2. 检验效能 (Power of test) 把握度在特定的水准下, 若总体间确实存在差异, 该项研究能发现此差异的概率.80 or above 样本含量越大, 检验效能越高 3. 总体的标准差 : 越大, 代表其中个体的变异度越大, 所需样本越多 4. 容许误差 : 预计样本统计量和相应参数的最大相应误差控制在什么范围在其他条件确定的情况下, 容许误差越小, 样本含量越大, 反之 26

27 第三节样本含量的估计样本含量的估计方法 : 经验法 : 实验研究 : 每组至少 10 例 ( 个人认为这太少!) 调查研究 : 至少 100 例描述性研究 : 样本应占总体的 20-30% 查表法计算法电脑软件 :Power analysis: 这是国外用得最多的方法! 27

28 Web-based Sample Estimates Rollin Brant's collection of sample size calculators Effect Size Calculators Effect Size Calculator Excel 28 10/18/

29 第三节样本含量的估计样本含量估计的注意事项多组设计时, 一般要求各组间含量相等多种样本含量估计方法相结合考虑样本的丢失情况 / 样本的回收率考虑统计方法学上的要求 : 如检验水准单 / 双侧检验等考虑研究方法学的不同 : 质性研究通常要求样本量较小, 量性研究则要求较大 29

30 Sampling Threats Internal validity Selection effect bias because of subject self-selection or researcher bias in selecting Attrition and mortality loss of subjects from the study. If the subjects who remain in the study are different from those who drop out, the results can be affected 30

31 Sampling Threats External validity Generalizability to other populations or settings Bias in sample representativeness and generalizability of findings are important issues that have generated national concern Women excluded as subjects African Americans, Asians, Hispanics, and any other cultural groups excluded 31

32 Practice: Identify the sampling strategy Using a table of random numbers, the sample of 50 subjects was selected from a list of all mothers giving birth in the county during the first 6 months of the year. The sample was selected from residents of eight nursing homes in Arkansas and consisted of cognitively impaired people with no physical impairments or other psychiatric illness. To study the educational opportunities for nurses in various ethic groups, a list of all nurses in the state of California was sorted by ethnicity. The sample consisted of 10% of the nurses in each ethic group, selected according to a table of random numbers. 32

RESEARCH/SCIENCE APPROACHES

RESEARCH/SCIENCE APPROACHES 第五章 Chapter 5 研究对象的确定冯先琼 PhD,RN 学习目标 :Learning Objectives (1) 掌握抽样的过程 (2) 掌握抽样的原则 (3) 掌握抽样的方法 (4) 熟悉总体样本误差等基本概念 (5) 了解样本含量的估计方法 2 Population Sampling Target population Accessible population Representative