Week 6 Lecture 1 of 1 BM4903 PDF
Document Details
Uploaded by LucidHafnium
Nanyang Technological University
Tags
Summary
This is a lecture on sampling methods, including probability and non-probability sampling. It covers topics such as population and sample, types of sampling procedures, research error, and sample size, and is a part of BM4903.
Full Transcript
BM4903 Week 6 Lecture 1 of 1 SAMPLING 2 1. Population & Sample 2. Types of Sampling Procedures Learning outcomes 3. Research Error 4. Sample Size 1. Population & Sample Population: Group or class of sub...
BM4903 Week 6 Lecture 1 of 1 SAMPLING 2 1. Population & Sample 2. Types of Sampling Procedures Learning outcomes 3. Research Error 4. Sample Size 1. Population & Sample Population: Group or class of subjects, variables, concepts or phenomena. The Goal of scientific research is to describe the nature of a population Process of examining every member of a population is called a Census Not possible to conduct census every time – due to time and resource constraints Sample is a subset of the population that is representative of the entire population. Population vs. Sample Sample representation is directly affected by three factors: sampling procedure, sample size and participation rate. If a sample is sufficiently large and is representative of the population, the results from a study using the sample can be generalized to the population. Population Sample size : This is the group of people chosen for a survey from a larger population. The larger your sample in comparison to the overall population, the Sample more accurate your answers will become. Sampling Frame: the list of population elements from which a sample will be A subset of the drawn. The list could consist of geographic areas, institutions, individuals, or other units. population 1. Population & Sample 2. Types of Sampling Procedures Overview 3. Research Error 4. Sample Size 2. TYPES OF SAMPLING : PROBABILITY, NON-PROBABILITY PROBABILITY SAMPLING A sample that is selected according to mathematical guidelines A sample in which each target population element’s chance for selection is known. One can statistically assess level of sampling error & allows researchers to calculate the amount of sampling error present in a research study. Inferences can be made about the population, and not just the sample. Thus, results are generalizable from the sample to the population ❖ SIMPLE RANDOM ❖ SYSTEMATIC ❖ STRATIFIED ❖ CLUSTER (AREA) 6 2. TYPES OF SAMPLING : PROBABILITY, NON-PROBABILITY NON-PROBABILITY SAMPLING A sample that relies on personal judgment in the element selection process Neither sampling error nor the margin of sampling error can be estimated or calculated Inferences are limited to the sample Thus, results are not generalizable from the sample to the population but can still be used for insights or developing hypotheses ❖ CONVENIENCE ❖ JUDGEMENT & SNOWBALL ❖ QUOTA 7 TYPES OF SAMPLING PROBABILITY NON-PROBABILITY SIMPLE RANDOM CONVENIENCE SYTEMATIC JUDGEMENT STRATIFIED SNOWBALL CLUSTER QUOTA 8 SAMPLING Probability 9 SIMPLE RANDOM: Each subject, element, event, or unit in the population has an equal chance of being selected. “Random” here is in the everyday sense. Example: the Sample drawn by a computer or from a 2. TYPES OF physical list using a random number table Every possible sample has an equally likely chance of being selected SAMPLING : Each individual is numbered and then a certain number of them is randomly selected PROBABILITY – SIMPLE RANDOM https://youtu.be/yx5KZi5QArQ SYSTEMATIC: Sample in which every kth element (k = sampling interval) in the population is systematically selected for the sample pool after a random start 2. TYPES OF Example: 250 NYP student’s attitudes toward the SAMPLING : newly implemented tuition hike need investigation Known Population: 5000 students published in the campus directory PROBABILITY Approach: – k = 5000/250 = 20 or 1 out of every 20 SYSTEMATIC students on campus will be surveyed. – Randomly select the first name then count down 20 names. – Select that person to be surveyed and then count down 20 names again. – Select that person and so on until you get 250 names. https://youtu.be/QFoisfSZs8I STRATIFIED: An approach to obtain adequate representation of a subsample 2. TYPES OF Characteristics of the subsample (strata or segment) may include almost any variable SAMPLING : – E.g. age, gender, religion, income level Used when researchers wish to ensure that a specific subsample is adequately represented. Strata are based on real proportions in the PROBABILITY - population (Proportionate stratified sampling): – e.g. 30% of the population is adults ages 18-24, then 30% of the total sample will be STRATIFIED subjects in this age group Disproportionate stratified sampling may be used when researchers wish to over-represent a particular stratum. STRATIFIED: The population is divided into mutually exclusive and exhaustive subsets A simple random sample of elements is chosen 2. TYPES OF independently from each group/subset This method is most appropriate when the population within each subset is homogeneous but heterogeneous between subsets/among the entire SAMPLING : sample with respect to key variables. The population is separated into non-overlapping groups called strata and then a proportional simple random sample is obtained from each group. PROBABILITY – The individuals within each group should be similar in some way. STRATIFIED (CONTD) https://youtu.be/sYRUYJYOpG0 CLUSTER: Like stratified sampling: the population is divided into mutually exclusive and exhaustive subsets Unlike stratified sampling: a simple random sample of sub-sets (i.e. clusters) is chosen 2. TYPES OF Most appropriate when subsets are heterogeneous within but homogeneous SAMPLING : between subsets/among the entire sample with respect to key variables “Area Sampling” – A form of cluster sampling that uses census PROBABILITY - tracks or city blocks as sampling units CLUSTER Block 56 Block 9 Block 103 https://youtu.be/QOxXy-I6ogs Block 408 CLUSTER: A cluster sample is obtained by selecting all individuals within a randomly selected 2. TYPES OF collection or group of individuals. SAMPLING : PROBABILITY - CLUSTER https://youtu.be/QOxXy-I6ogs SAMPLING Non-Probability 16 CONVENIENCE SAMPLE: Population elements are sampled 2. TYPES OF simply because they are at the right place at the right time SAMPLING : Collection of readily accessible subjects for study Also called “Accidental” Sample NON- – Examples PROBABILITY Television news “question of the day” polls CONVENIENCE Street/Shopping Mall intercepts 17 JUDGEMENT SAMPLE : Population elements are handpicked because they are expected to 2. TYPES OF serve the specific research purpose – Example – panelists who are SAMPLING : knowledgeable about the issue being researched are selected rather than selecting random respondents. NON- SNOWBALL SAMPLE : The Initial sample is chosen by a probability technique (e.g., PROBABILITY – systematic sampling) then the population elements are asked for referrals of others they JUDGEMENT & know who might be interested in participation SNOWBALL – Example – A demand study for a new product where initial respondents know people with a high interest level within the product category QUOTA SAMPLE: Sample chosen so that the proportion of sample elements with certain characteristics is about the same as 2. TYPES OF the proportion of the elements with the characteristics in the target population. Example: Investigate 100 NYP students’ attitudes toward a newly implemented tuition fee hike SAMPLING : imposed by the government Known Population Parameters: Class (33% 1st year, 33% 2nd year, 33% 3rd year) and Gender (50% Female, 50% Male) Approach: NON- Select 10 students – comprising 5 males, 5 females. – The split between the years is equal (with the PROBABILITY extra person in one of the three years). These 10 students will then interview 10 friends each QUOTA (in the same proportions) for a total of 100 responses The resulting sample looks like the overall population on key aspects, it may not accurately reflect other aspects of the population Video illustration of all 3 Non probability methods https://youtu.be/4VO_xHHD180 1. Population & Sample 2. Types of Sampling Procedures Overview 3. Research Error 4. Sample Size Since a sample does not provide the exact data that would come from an entire population, ERROR must be addressed when interpreting results The smaller the sample size, the higher the error Sample Sampling/ Non-sampling Errors Research = Results Truth + (Non-coverage, Respondent, Administrative….) 3. RESEARCH ERROR a. SAMPLING; b. NON-SAMPLING b. NON-SAMPLING ERROR: Error created by every other aspect of a research study e.g. data analysis, the influence of the research situation itself or errors from unknown sources that can never be identified/controlled/ eliminated a. SAMPLING ERROR : Error related to selecting a sample from a population Measurement Error 3. RESEARCH ERROR a. SAMPLING ERROR : RANDOM/SYSTEMATIC I. RANDOM SAMPLING ERROR: relates to problems where measurements and analyses vary inconsistently from one study to another (results lean in one direction in one study and in the opposite direction in another study). Caused by unknown and unpredictable variables may be virtually impossible to detect and correct. a. SAMPLING ERROR : AKA standard error. The pure difference between results obtained from a sample and results that would have been obtained had information been gathered from or about every member of the population (census) Sampling error is decreased by increasing sample size Can be estimated (assuming probability sample) Usually less troublesome than other kinds of error II. SYSTEMATIC SAMPLING ERROR: consistently produces incorrect (invalid) results in the same direction, or same context , is therefore predictable. May be able to identify the cause of systematic errors and eliminate their influence. 3. RESEARCH ERROR a. SAMPLING ERROR : SYSTEMATIC – NON- COVERAGE/SAMPLING FRAME ERROR SYSTEMATIC SAMPLING ERROR : NON-COVERAGE OR SAMPLING FRAME ERROR This error arises because of a failure to include some units or entire sections, of the defined target population in the sampling frame Non-coverage error is basically a sampling frame problem Can be reduced, although not necessarily eliminated, by recognizing its existence and working to improve the sampling frame (e.g. use other sources) 3. RESEARCH ERROR b. NON-SAMPLING : RESPONSE/NON-RESPONSE b. NON-SAMPLING ERROR : RESPONSE Occurs when an individual provides a response to an item, but the response is inaccurate for some reason Can be responding either consciously or subconsciously Possible causes of response errors include Does the respondent understand the question? Does the respondent know the answer to the question? Is the respondent willing to provide the true answer to the question? Is the wording of the question or the situation in which it is asked likely to bias the response? (Not to be confused with response errors from observations) 3. RESEARCH ERROR b. NON-SAMPLING : RESPONSE/NON-RESPONSE b. NON-SAMPLING ERROR : NON-RESPONSE Non-sampling error that represents a failure to obtain information from some elements of the population that were selected and designated for the sample Two types of non-response errors: 1. Refusals: respondents refuse to participate 2. Not-at-home This is a potential problem that only occurs when those who do respond are systematically different in some important way from those who don’t respond Methods of handling: 1. Have interviewers make advance appointments 2. Call back at another time, preferably at a different time of day 3. Attempt to contact using another approach 4. Attempt to convince the respondent of the importance of his/her participation 5. Keep the survey as short as possible 6. Guarantee confidentiality or anonymity 7. Train interviewers well and match their characteristics to those of the subject pool 8. Personalize the recruiting message/script where possible 9. Use incentive 10. Send follow-up/reminder surveys 3. RESEARCH ERROR b. NON-SAMPLING : ADMINISTRATIVE/OFFICE b. NON-SAMPLING ERROR : ADMINISTRATIVE/OFFICE Non-sampling errors that arise in the editing, coding, or analysis phases of research Usually created by researchers themselves Most office errors can be reduced, if not eliminated, by exercising proper controls 1. Population & Sample 2. Types of Sampling Procedures Overview 3. Research Error 4. Sample Size 4. SAMPLE SIZE - Factors to consider The size of the sample required for a study depends on at least one or more of the following factors : Project Purpose & Type Some research studies are not designed to generalize the results to the population but rather to investigate variable relationships or collect exploratory data to design questionnaires OR measuring instruments. Non-probability sampling is appropriate in these situations. Budget/Cost vs. Value Sample should produce the greatest value for the least investment If the cost of a probability sample is too high vs. the quality and type of data to be collected, a non-probability sample would be preferred Time constraints In the presence of time constraints, a non-probability sample is preferred Amount of acceptable error In preliminary studies or pilot studies, where error control is not a prime concern, a non- probability sample is usually adequate. Other factors: Previous research in the area 4. SAMPLE SIZE - Factors to consider Research designed as a preliminary search for general indications does not usually require a large sample. Projects intended to answer significant questions (e.g. large investment decisions or involve human lives) require high levels of precision and require therefore large samples. Generally speaking, the larger the sample, the better. However a large unrepresentative sample is as meaningless as a small unrepresentative sample. Researchers should not consider just large numbers alone. Sample quality is always more important in sample selection than mere size 4. SAMPLE SIZES, Industry Case For academic use & illustration purposes only Source: AVIA Industry Report 2020-21 For academic use & 4. SAMPLE SIZES, Industry Case, Source: AVIA Industry Report 2020/21 illustration purposes only A sample is subset of a population that is representative of the population 2 broad types of Sampling Methods 1. Population & Sample PROBABILITY : – SIMPLE RANDOM, SYSTEMTATIC, STRATIFIED, CLUSTER 2. Types of Overview : NON-PROBABILITY : Sampling – CONVENIENCE, JUDGEMENT, SNOWBALL, Procedures QUOTA RECAP 3. Research Error 2 broad types of Error in research SAMPLING : – RANDOM, SYSTEMATIC NON-SAMPLING : – RESPONDENT, ADMINISTRATIVE 4. Sample Size Large Sample Sizes reduce error but Project Purpose, Quality and Representation are also important BM4903 Week 6 Lecture 1 of 1