SAMPLING IN MEDICINE
RESEARCH SAMPLING IN MEDICINE
Abstract: Convience sample is probably overused because it is easy to obtain and other strategies are meant to “cover” for its weakness. Students are socialized by information that may be flawed because of the convenience sample. A discussion ensues about new material on improving this situation
INTRODUCTION
Medicine faces numerous problems and must adapt to the most efficient strategies to help or heal a patient. Nearly all divisions of medicines face this issue. This is especially a pharmaceutical problem. Within the myriad of choices is the research used to find the test chemical/biochemical approaches as well of other criteria to be able to complete their problem.
DESCRIPTION
CONVIENCE SAMPLE can be discouraging to most medical personnel who know research methods, but are uncomfortable with the shaky foundation of the outcomes. That which brings academic malaise is the convience sample.
OPPORTUNITY SAMPLE is another name for the above. Without a hard count, it would appear that many studies are drawn from individuals and groups that are not random. The stimuli have already happened or some other events make the sample easy to find and use. This would also include a small group in which the stimuli are gathered at the present time. Past or present it is easy to accomplish. In other words, it saves funding, labor, and time.
RANDOM SAMPLE usually means that rather than research hundreds or thousands of people, the academic borrows from a small sample that ideally and mathematically reflects the larger group. The results are ultimately improved. This is the legitimate strategy.
BIASED SAMPLES
They will now be discussed. The following hopefully are understandable and brief .If research is generally varied, it is called a convenience sample. If another sample of folks who have a specific problem, then it is called a purposive sample. Many years ago, the author would suffer from nausea. With anxiety, pyloric muscle would constrict and the author would feel spasmodic and uncomfortable. As this is being written, the med appears to have moved off the market. It caused a great deal of fatigue. The second was excellent, but it is no longer available. The third is just fine. There are 79 others to help alone or with other meds.
CROSSECTIONAL SAMPLE
At any rate, there are other strategies that appear to try to enhance validity and reliability of a sample. Crossectional sample indicates that there is a “within” comparison of an already sampled mass in which the stimuli has already occurred. So “A” and “B” are compared. They differ because they say so on the questionnaire or to the interviewer. There is more structure. It is purposive, and it is not random of the entire population of respondents that have nausea.
LONGITUDINAL SAMPLE
This means that a sample is drawn independently over and over. It may be reported on a continuous graph. However it is not continuous and not random. It is series of individual samples from independent studies that are strung together.
PANEL SAMPLE
This uses the same sample over and over in terms of time. One may see that some respondents may drop out and need to be replaced. It may also mean that friends or acquaintances know the one is in a panel study and may inadvertently try to influence those who are in the sample. Thus it is close to random, but not perfect randomness.
ADJUSTED SAMPLE
This means numbers of a comparison of two groups that differ in terms of population. So one convenience group is from a large city and the other is from a small town. The numbers with nausea may differ, because of the raw numbers, but because the size of the samples come from two entirely different population bases in terms of size. Thus both numbers of a convenience sample use the adjusted sample by multiplying each group by the same constant. The comparison are fairer although they are of convenience to the researcher and not random.
META-ANALYSIS SAMPLE
It is a literature review of which individual numbers and parameters from different studies are added together and a statistical test is used. So a gathering of data is relatively convenient and not very accurate.
PRELIMINARY SAMPLES
The subject is incredibly difficult to find and is even harder to gather larger samples. This is not convenient but it is legitimate. It is part of the steps toward validity. Simple statistical test are used. So the risk of small numbers is acceptable here. They are collapsed into a larger non-random whole. At a specific time, eye balling the information suggest drawing a random sample and testing the larger data.
OFF LABEL SAMPLES
The medication reduces skin eruptions or acne. This is serendipitous as it also reduces nausea so new samples are purposive and acceptable. This is convenient because some of the major work has already been accomplished. However it is very helpful.
SYSTEMATIC SAMPLES
This draws from the target population of every tenth person. Although it is consistent and the number can be larger or smaller. Regardless it is not random.
CLUSTER SAMPLES
They mean that the population is widely dispersed and geographically distance. If it is not properly gathered and analyzed and it is not random.
STRATIFIED SAMPLES
This means that three samples independent of each other and differ in size are compared with each other. The samples are not necessarily random. They are weighted to control for size and a statistical test. Once the samples are summated the equation is complete. Again, it is not random.
(Small samples when they are drawn randomly generate a 6 percent margin of error. That means 6 up and 6 down. When one medicine is vastly superior to the placebo, the margin of error of this size can be tolerated. The “N” is about 210. However, when the difference between the two is very close, five thousand or more (5,000) is necessary. However if the sample is not random even if all these precautions are taken this does not make it random)
LONG TERM SAMPLING
This is the sum of numerous uses through purchases by a large bulk from hospitals or retail purchases and is reanalyzed. Once on the market, the pill is not acceptable. It is taken off the market. The sample is now similar to perfect sample, but off market generally indicates that the sample is not random. Perhaps the researcher does discover some latent information so it is kept.
RANDOMIZED SAMPLES
What is the target population? Are we making a non-random into a random sample? Numerous strategies are discussed, but what about the manipulation of the larger whole?The reader may want to read the Wikipedia article or some other source.
(This is beyond both both author or this article. However, what is the importance of randomness using dummy variables?)
DISCUSSION
If we stop here, then the behavioral methodologies or social sciences are a metaphor of a boat with leaks that are covered with a patch(s) that imply randomness. Some are closer to random, but they are not.
The author could finish with the Ioannidis online article how studies published in the top echelon of Google Rank; nearly 50% could not be replicated. Simple crosstabs along with random numbers may suffice. Not only does chi-squared accurately assess homogeneity but can also cover heterogeneous numbers. In this selected area, physical and behavioral (social sciences) come together. Further, although less powerful, chi-squared can do a lot of other assessments.
However, there is hope.
Please see:
Babalola, S,Nawanzu, S. Serpa, S.(2021)The Current Phase of Social Science Research:A Thematic Overview of the Literature. Cogent Social Sciences, Vol. 7