Download presentation
1
Sampling Designs and Techniques
R. Eric Heidel, PhD University of Tennessee Graduate School of Medicine © 2014 Pearson Education, Inc.
2
The Purpose of Sampling
It should represent the total population so that the data collected will be as accurate as if taken from the entire population. The population of interest is designated, and then a sample is derived. Sampling provides a savings in time and money. The sample may achieve a greater response rate owing to greater cooperation than might occur in the full population survey. Researchers can keep a low profile by using a sample. In the case of interviewing, a sample reduces the number of interviews and interviewees. The benefits are realized only if the sample is drawn with precision. © 2014 Pearson Education, Inc.
3
The Sampling Frame The sampling frame is a list of all the persons (objects) from whom the sample is to be drawn. The sample cannot be more accurate than the sampling frame from which it is selected. It lists every person in the population but does so only once so as not to increase someone's likelihood of being chosen. As the size of the study increases, the construction of the sampling frame becomes more formidable. If people are excluded randomly from the total population, then little harm will be done. © 2014 Pearson Education, Inc.
4
The Sampling Unit The sampling unit could be: an individual
an intact group such as a classroom or school an organization such as the local medical society a geographical region such as a city or county The sampling unit is particularly important for data analysis. © 2014 Pearson Education, Inc.
5
The Sampling Unit © 2014 Pearson Education, Inc.
6
Sampling Techniques Two types of sampling techniques:
Probability samples Those wherein the probability of selection of each respondent, address, or even object is known Nonprobability samples Reflect an unknown probability of selection © 2014 Pearson Education, Inc.
7
Probability Sampling Common thread in all of the probability sampling techniques is the random selection of participants or objects. Because selection is random, each participant or object has an equal chance of being chosen and a known probability of being chosen. © 2014 Pearson Education, Inc.
8
Probability Sampling (cont'd)
Types of probability sampling follow: Simple random sampling Systematic sampling Stratified random sampling Cluster or area sampling Multistage cluster sampling © 2014 Pearson Education, Inc.
9
Nonprobability Sampling
The probability that a person will be chosen is not known, with the result that a claim for representativeness of the population cannot be made. Sampling error is unknown. The researcher's ability to generalize findings beyond the actual sample is greatly limited. It has an advantage over probability sampling in that it is less expensive, less complicated, and lends itself to spontaneity. © 2014 Pearson Education, Inc.
10
Nonprobability Sampling (cont'd)
Types of nonprobability sampling follow: Convenience sampling Quota sampling Dimensional sampling Purposive sampling Snowball sampling Mixed sampling designs © 2014 Pearson Education, Inc.
11
Sample Size Correct sample size is dependent on both the nature of the population and the purpose of the study. An ideal study would have a sample large enough to represent the population so generalization may occur yet be small enough to save time and money, as well as to reduce the complexity of data analysis. © 2014 Pearson Education, Inc.
12
Considerations in Sample Size
It is a misconception that a sample is a small carbon copy of the original population, identical in every way. One can never be certain of representativeness unless the entire population is used. The larger the sample, the greater the likelihood of representativeness. This is especially true if the population is quite heterogeneous on the given variable; the greater the heterogeneity, the greater the necessity for a larger sample. © 2014 Pearson Education, Inc.
13
Considerations in Sample Size (cont'd)
Sampling Error Sometimes called level of precision, it is the degree to which the sample means of repeatedly drawn random samples differ from one another and from the population mean. The larger the sample size, the smaller the sampling error. It is because the larger the sample, the closer it is to the study population in all measured characteristics. © 2014 Pearson Education, Inc.
14
Sample Size for Studies with Hypotheses: Analytical Studies
When the research effort has a null and alternative hypothesis, such factors as alpha (α), beta (β), effect size, power (1 – β), and directionality should be taken into account. Alpha level When testing the null hypothesis to determine if it is true, the researcher sets a cutoff point called the alpha level to make the decision. This alpha level is the guide for rejecting or not rejecting the null hypothesis. If the researcher rejects the null hypothesis when in fact it is true, a Type I error occurs. The probability is called alpha, or the level of significance. © 2014 Pearson Education, Inc.
15
Sample Size for Studies with Hypotheses: Analytical Studies (cont'd)
If the researcher does not reject the null hypothesis when in fact it is false, a Type II error occurs. The term beta level refers to the chance of making a Type II error. Type I and Type II errors require a balancing act by the researcher. As the researcher decreases the possibility of making a Type I error, the possibility of making a Type II error automatically increases. © 2014 Pearson Education, Inc.
16
Sample Size for Studies with Hypotheses: Analytical Studies (cont'd)
A related issue when considering sample size for analytical studies is called power. Power is the probability of correctly rejecting the null hypothesis. Effect size is another factor in determining sample size. Effect size goes beyond significance testing in an attempt to quantify the size of the association between the sample and the population or any two groups under investigation. It may be determined by prior research or pilot study. There are five steps in selecting a sample size. © 2014 Pearson Education, Inc.
17
Sample Size for Studies without Hypotheses: Surveys and Descriptive Studies
Descriptive studies and surveys may have research questions but generally do not compare groups or have specific outcome variables. These studies look at information through descriptive eyes such as proportions and means. In surveys, the researchers frequently look for precision by establishing confidence levels. This confidence level, also called probability level, is usually set at the 95% or 99% level. © 2014 Pearson Education, Inc.
18
Sample Size for Studies without Hypotheses: Surveys and Descriptive Studies (cont'd)
At the 95% level, the researcher has confidence that there is a 95% chance that the sample is distributed in the same way as the population. There are 11 steps using this information and more to determine sample size for health surveys. © 2014 Pearson Education, Inc.
19
Additional Sample Size Considerations
Two key elements that play a role in sample size are design, specifically stratified random sampling, and weighting where necessary. Confidence level indicates the probability that the sample proportion will reflect the population proportion with a specific degree of accuracy. A stratified random sampling is often disproportionate in that a greater proportion is sampled in one stratum than another. Two major reasons for this are: differences in population size differences in homogeneity among strata © 2014 Pearson Education, Inc.
20
Additional Sample Size Considerations (cont'd)
One way around this dilemma is to use weighted sampling. With this procedure the additional problem of combining subsamples (strata) into one overall sample for the purpose of data analysis can be overcome. Weights are assigned to each of the strata. There are nine considerations in determining sample size. © 2014 Pearson Education, Inc.
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.