SEQUENTIAL SAMPLING
Author : Jayashree Jadhav S (2048125)
Sequential
sampling is a non-probability sampling technique wherein the researcher picks a
single or a group of subjects in a given time interval, conducts his study, analyses
the results then picks another group of subjects if needed and so on.
This sampling
technique gives the
researcher limitless chances of fine tuning his research methods and gaining a
vital insight into the study that he is currently pursuing
In sequential sampling, a
sequence of one or more samples is taken from a group. Once the group has been
sampled, a hypothesis test is performed to see if you can reach a
conclusion. If you can’t, the whole procedure is repeated. A characteristic
feature of sequential sampling is that the
sample size is not set in advance, because you don’t
know at the outset how many times you’ll be repeating the process.
Difference of Sequential Sampling from All Other Sampling
Techniques
If we are to consider all the other
sampling techniques in research, we will all come to a conclusion that the experiment and the
data analysis will either boil down to accepting the null hypothesis or disproving the null hypothesis
while accepting the alternative hypothesis.
In sequential sampling technique, there exists another step, a third
option. The researcher can accept the null hypothesis, accept his alternative
hypothesis, or select another pool of subjects and conduct the experiment once
again. This entails that the researcher can obtain limitless number of subjects
before finally making a decision whether to accept his null or alternative
hypothesis.
This
method is designed for two clear choices. For example:
·
Is the
heat in a system above or below a critical level? Heat is measured in one part
of the system to see if it has reached the critical level. If the heat is close
to the critical level, but not over it, resample and repeat the
calculations.
·
Should I
spray pesticide or not? Pests could be counted on a plant. If there are a large
number of pests, spray pesticide. If there are a small number of pests, do not
spray pesticide. If there are a middling number of pests, sample another plant.
Sequential
samples can either be:
·
Item-by-item: one sample at a time.
·
Group: sample sizes of two or more.
In order
to use this method, you must be able to sample
serially. If you have to choose all of your sample items at the
same time, you should choose another sampling method (like simple random sampling or
a non-probability sampling method).
Three Outcomes
With traditional sampling methods, a hypothesis test has one of two possible results: you either reject the null hypothesis, or you do not. With sequential sampling, you have three possibilities:
1.
Reject
the null hypothesis (end the experiment),
2.
Do not
reject the null hypothesis (end the experiment),
3.
Fail to
draw any conclusion (draw another sample and repeat the test).
Time-Sequential Sampling
In this
variant, sometimes called time-sequential classification, you use time as your
sampling frame instead of a physical population to sample from. For example,
you might choose a sample member at 24-hour intervals.
Advantages of Sequential Sampling
1. Although it sounds like the process could go on and on forever, sequential sampling usually ends up with smaller samples than traditional (set size) sampling.
The researcher has a limitless option when it comes to sample size and sampling schedule. The sample size can be relatively small of excessively large depending on the decision making of the researcher. Sampling schedule is also completely dependent to the researcher since a second group of samples can only be obtained after conducting the experiment to the initial group of samples.
As mentioned above, this sampling technique enables the researcher to fine-tune his research methods and results analysis. Due to the repetitive nature of this sampling method, minor changes and adjustments can be done during the initial parts of the study to correct and hone the research method.
Disadvantages of Sequential Sampling
1. However, the mathematics needed to analyse data for sequential sampling is much more complex and the procedure is generally more time consuming (and can be more expensive) than fixed-size sampling.
2. This sampling method is hardly representative of the entire population. Its only hope of approaching representativeness is when the researcher chose to use a very large sample size significant enough to represent a big fraction of the entire population.
3. The sampling technique is also hardly randomized. This contributes to the very little degree representativeness of the sampling technique.
The package Sequential is designed for
continuous and group sequential analysis, where statistical hypothesis testing
is conducted repeatedly on accumulating data that gradually increases the
sample size. This is different from standard statistical analysis, where a single
analysis is performed using a fixed sample size. It is possible to analyze
either Poisson type data or binomial 0/1 type data. For binomial data, it is
possible to incorporate an off-set term to account for variable matching
ratios. For Poisson data, the critical value is based on a Wald-type upper
boundary, which is flat on the scale of the log-likelihood ratio, and on a
predetermined maximum sample size. Alternatively, it is also possible to apply
a user defined alpha spending function in order to specify non-flat signaling
thresholds. For group sequential analyses, there are functions for
pre-specified group sizes and for the situation when the group sizes are not
known a priori. It is also possible to perform mixed continuous/group
sequential analysis, where, for example, there is at first a big batch of data
that arrives in one group, followed by continuous sequential analysis. All
results are exact, based on iterative numerical calculations, rather than
asymptotic theory or computer simulations. In the Sequential package, there are
functions to calculate critical values, statistical power, expected time to
signal, and expected sample size at the end of the sequential analyses whether
the null hypothesis was rejected or not. For example, for any desired power,
relative risk and alpha level, the package can calculate the required upper
limit on the sample size (maximum length of surveillance), the critical value
needed, and the corresponding expected time to signal when the null hypothesis
is rejected.
Comments
Post a Comment