NUT602 Sampling Methods PDF Fall 2024

NUT602: Research Methods in Fall Nutrition and Food Science 2024 Sampling Methods Sampling Parameter Statistic Sampling Sampling ▪ Target population: people the study is focused on, i.e., group to whom you want to apply your results ▪ E.g., postmenopausal women...

NUT602: Research Methods in Fall Nutrition and Food Science 2024 Sampling Methods Sampling Parameter Statistic Sampling Sampling ▪ Target population: people the study is focused on, i.e., group to whom you want to apply your results ▪ E.g., postmenopausal women with breast cancer ▪ Accessible or source population: subset of the target population to whom you can get access → should accurately reflect target population ▪ E.g., members of postmenopausal breast cancer registry ▪ Sampling frame: list of members of the accessible population from which the sample is drawn ▪ List of names of women with postmenopausal breast cancer from the registries ▪ Sample: people from the accessible population asked to participate (some members may be eliminated if they do not meet the eligibility criteria for the study/survey) Sampling ▪ Application of the sampling methodology depends on ▪ Available resources ▪ Decisions of the investigators regarding how the sample will best represent the reference population Types of Sampling Methods (4.1) - YouTube  Sampling: Unbiased vs biased ▪ Sampling methodology or sample design ▪ Representative selection of sample members from the reference population → data that accurately reflect the population ▪ Sample members can be individuals, households, schools… (sampling unit) ▪ Selection criteria (inclusion/exclusion criteria)  Random sample Types of Samples Probability Samples Systematic Each member of the sample population has a known non- zero probability of being Stratified selected sample CHANCE More rigorous statistical analysis Multi-stage cluster sample Probability vs. and generalization Non-probability Convenience sample Purposive Non-probability Samples sample Members are selected from the population in some nonrandom manner Quota sample HUMAN JUDGEMENT More suited to qualitative or Snowball inductive research sample  Probability Sampling ▪ Theoretically the probability of a member of the population being included is the same for all members of the population ▪ Simple random sampling ▪ Requires an available comprehensive sampling frame ▪ Involves randomly selecting individuals on a numbered list using SPSS, an online random sample generator, or a table of random numbers (cumbersome when seeking a larger sample) Generate a Simple Random Sample from a Random Number Table - YouTube Probability Sampling ▪ Systematic sampling ▪ Involves selecting every nth individual according to a random starting point and a fixed periodic interval ▪ For example, every 3rd patient who arrive at a clinic after the 5th patient ▪ Starting point should be random and Interval (every nth person) should not correspond with any repeated pattern in the sequence of the reference population ▪ There might be repeated patterns that are difficult to identify and may be associated with specific characteristics of the sample Probability Sampling ▪ Stratified random sampling 1. Dividing the accessible population into specific groups (strata), based on a factor that could influence the variable being measured, such as age, gender, or urban/rural location, or socioeconomic status 2. Using simple or systematic random sampling for each stratum ▪ Random sampling usually, but not always, proportionally: more respondents are drawn from the larger strata ▪ Appropriate when there may be indications that dietary exposures or health outcomes vary between strata (to make sure that the sample accurately represents the different groups in the population) Probability Sampling ▪ Multistage cluster random sampling 1. Dividing the population into clusters (such as geographic clusters or nursing homes) 2. Randomly picking some of the clusters 3. Randomly sampling within each of those clusters ▪ Clusters are more or less alike, each resembling the overall population ▪ Useful when the population is widely distributed geographically and occurs in natural clusters (sampling more practical or affordable) ▪ Popular design for cross-sectional studies ▪ Realistic ▪ Ensures the representativeness of a sample deriving from a large geographical area Probability Sampling ▪ Multistage cluster random sampling ▪ Cross-sectional nutrition studies aim to evaluate the dietary habits of large populations such as a country ▪ → Unrealistic to select a sample of 1000 participants from a country using a simple random sample design ▪ Non-existence of a comprehensive sampling frame ▪ Even if a frame is available, the randomly selected members might live far from each other → impractical in terms of time and resources to recruit them ▪ Divide the country into regions ▪ Select a random sample of regions (first-stage) ▪ In each region, a list of towns should be drawn and a random sample of them can be selected (second-stage ▪ In each town, a sample frame may be available to make a random selection of sample members (third-stage) ▪ Even if the sample frame is incomplete, further sampling stages can be used with geographical or other related criteria ▪ → Random (representative), feasible recruitment Probability sampling methods with examples Nonprobability Sampling ▪ Respondents are selected by convenience or judgment → less likely to produce a representative sample ▪ Convenience sampling ▪ When members are chosen because they are easy to reach and the researcher often has a comfort level asking them to participate: Weakest form but frequently used ▪ Quota sampling ▪ Convenience sampling with quotas (limits) put on the number of people in the sample of a particular gender, age, race, or other characteristic → like stratified random sampling, yet the respondents are not being drawn randomly ▪ Often quotas are based on proportions found within the population ▪ Purposive or judgment sampling ▪ Convenience sample, but respondents are chosen as they are a good representative of the population ▪ Network (snowball) sampling and Respondent driven sampling ▪ Used when locating respondents with the needed characteristics is difficult ▪ Researcher finds a few respondents who are appropriate for the study (purposive sampling) and ask to be directed to further potential respondents ▪ Used more often in qualitative studies Non- probability sampling methods with examples Take-home message 16