Topics Covered in this Session

• Sampling
• Sampling Techniques
• Sample Sizes

Sampling

Definition - sampling is selecting a group (subgroup) from a much larger population that is similar in its trait (i.e. gender, ethnicity, age, income, etc.) distribution of the larger population. Findings made from studying the group can then be generalized to the larger population.

Depending on the methodology being used in a study, sampling a population may or may not be necessary. Studies that limit themselves to describing activity for a specific population do not have to use sampling techniques but can accept whatever sample is available. This is frequently the case with historical, ethnographic, action and evaluation research.

However, studies that attempt to infer that the findings from a sample can be extended to a larger population need to establish that the sample is representative of the larger population. Representation here refers to the basic trait distribution of a population. For example, if a general population is made up of 51% females and 49% males, a sample (regardless of size) that is made up of 58% females and 42% males is not representative on this trait.

Sampling Techniques

Various sampling techniques can be used depending on the type of research to be conducted. The two major types of techniques are probability sampling and nonprobability sampling.

• Probability Sampling - any sampling procedure that specifies the probability that each member of a population has of being selected. Probability sampling techniques include:
  
  
  • Random Sampling - a group drawn from the population, with every member of the population having an equal chance of being selected. This is the most common and highly recommended technique.
  
  
  
  • Stratified Sampling - a group selected from a population that reflects accurately certain segments of a population. In this type of sample, certain segments or traits are identified as important to the research and the sample selected is controlled to insure that those traits are accurately represented.
  
  
  
  • Cluster Sampling - is used when certain groupings important to the research are already established. This is frequently the case when studying teaching techniques in classroom settings. Rather than the sample of students being taught, the classes (groupings) of students need to represent the larger population (i.e. all fourth grade classes).
  
  
  
• Nonprobability Sampling - any sampling procedure that cannot specify the probability that each member of a population has of being selected. Nonprobability sampling is used when probability sampling is not feasible. Nonprobability sampling techniques include: o Convenience Sampling - a group of participants in a study are selected that happen to be available. In educational research, convenient sampling is used frequently by teachers who use their own classes for their research. Findings from such research generally are limited to the population studied and not extended to larger populations.


• Judgmental Sampling - a researcher uses his or her judgment to select a population that reflects an important aspect of the research (i.e. all female valedictorians from the class of 1959 at the Bronx High School Science). Judgmental sampling is generally more appropriate for qualitative research than quantitative research.

Sample Sizes

There is an extensive literature base on sample sizes. However, overall agreement on sample sizes is elusive. The basic assumption that the larger the sample the more representative of the population is not always the case. This is frequently true when the sample size is improperly selected or controlled. Many researchers hold that the accuracy of a sample is more important than the size. On the other hand, when using certain statistical procedures such correlation coefficients or analysis of variance, minimal sample sizes are recommended for the statistical significance to be valid. A broad rule of thumb is that samples smaller than 30 subjects are not likely to reflect the trait distributions that exist in a population.