Lecture 2 IDP
Lecture 2 IDP
By:
Anas Hussein
RSDD
Sampling
In clinical research, we usually study a sample of individuals
who are assumed to be representative of a wider group, to
whom (with a good research design and appropriate sampling)
the research might apply. This wider group is known as the
target population.
Advantages of Sampling
1- Saves time, effort and money, while gives the required data.
2- Allows fulfillment of work with limited resources (manpower,
equipment and financing).
3- Decrease possible errors
Classification of sampling
Sampling methods in clinical research can be classified into two major categories:
Random/Probability sampling methods: all subjects in the target population
have equal chances to be selected in the sample
In this simplified example, just like in snowball sampling, you start with one person, and as
they provide you with contacts, your network grows. It's a way of tapping into existing social
connections to reach a larger group, even when you don't have a complete list of potential
participants initially.
Judgmental sampling
Systematic Sampling: Start with the third customer and then select every 10th customer
thereafter. So, you would survey customers 3, 13, 23, 33, and so on.
The key characteristic of systematic random sampling is the use of a fixed interval after
selecting a random starting point. This method provides a balance between the simplicity
of simple random sampling and the structure of stratified sampling.
Systematic random sampling is efficient and can provide a representative sample when
there is a regular pattern in the population. However, it assumes that the population is
randomly ordered or that any systematic patterns do not coincide with the sampling
interval.
Cluster sample
Some studies are carried out among clusters (confined groups) and not the
general group, of the population. Important Clusters are schoolchildren,
industrial workers, camp members, and households. After identifying the
clusters, certain clusters are chosen using simple random sampling while the
others remain unrepresented in a study; only individuals within selected
clusters are surveyed
Imagine you are a researcher studying the preferences of TV shows in a town with multiple
neighborhoods. Instead of surveying every household, you decide to use a cluster sampling
approach.
Identify Clusters: Divide the town into clusters based on neighborhoods. Each
neighborhood is a cluster.
Randomly Select Clusters: Use a random method to select a few neighborhoods. Let's say
you randomly choose three neighborhoods out of ten in the town.
Survey All within Selected Clusters: Instead of surveying every household in those three
neighborhoods, you survey all households within those selected neighborhoods.
So, in this simplified example, you're not individually selecting households; you're treating
entire neighborhoods as your sampling units. This makes the survey more manageable
while still providing insights into the overall preferences of the town.
Remember, the key idea with cluster sampling is to randomly
select groups (clusters) and then survey all members within
those selected groups, rather than surveying individuals
randomly across the entire population.
Multi-stage random sample
It is from of random sampling that can be applied in national or
widespread studies, where the field of work is arranged into
levels or stages. The first level is the governorates, then the
cities and towns, then districts and villages and lastly families
and individuals. A random sample is successively drawn from
each stage.
Imagine you want to understand the opinions of students in a big university about
their campus experience. Instead of surveying every student, you decide to use a multi-
stage random sampling approach.
Stage 1 - Dividing into Clusters: The university is divided into clusters based on
departments. Each department is a cluster.
Randomly Select Clusters: Use a random method to choose a few departments. Let's
say you randomly select three departments out of ten.
Stage 2 - Subdivide the Clusters: Within each selected department, further divide it
into classes or courses.
Randomly Select Sub-Clusters: Use a random method to choose specific classes
within each of the three selected departments.
Stage 3 - Survey at the Smallest Level: Finally, within each selected class, you
conduct your survey on individual students.
So, instead of surveying all students across the entire university, you've taken a step-
by-step approach. First, you randomly chose departments, then within those
departments, you randomly chose specific classes, and finally, you surveyed students
within those selected classes.
This multi-stage process helps make the study more manageable while still ensuring
that you gather insights from different departments and classes, making the sample
representative of the entire university
Sample size
A common question posed to a biostatistician from a medical researcher is “How
many subjects do I need to obtain a significant result for my study?”.
Factors that Affect the Sample Size of a Research Study
There is no fixed size of sample, it varies according to certain determining factors:
1. Available resources and facilities: time, personnel, equipment and money.
2. Prevalence of the disease or problem: the lower the prevalence, the larger the
size and vice versa.
3. Variability of the group: the many the subdivisions and subgroups, the
larger the sample size, so as to get a reasonable number of cases in each of
the many categories.
1.Available Resources:
Think about the time, people, money, and equipment you
have for your study. Your sample size should be something
you can manage with the resources you have. If you have a
lot of resources, you might be able to study more people.
2.Variability in Characteristics:
Imagine you're studying something like education levels. If
people in your group are all similar in their education, you
might not need as many people in your study. But if there's a
big range in education levels, you might need a larger group
to get a good picture of what's going on.
3.Prevalence of the Disease or Problem:
If what you're studying is not very common (like a rare
disease), you might need a bigger group to make sure you get
enough cases to study. If it's something a lot of people have,
you might get away with studying fewer people.
4.Variability of the Group:
If you're looking at different kinds of people or subgroups
(like age groups or income levels), the more categories you
have, the more people you might need to study. This helps
you make sure you're not missing important information
from any specific group.
In simple terms, the size of your group depends on how
much you have to work with, how much the things you're
studying vary, how common or rare they are, and how many
different types of people you're looking at. It's about finding
a balance so you can get good results without using too many
resources.
Learn how to determine sample size
To choose the correct sample size, you need to consider a few different factors that
affect your research then you’ll be able to use a sample size formula:
Population Size:
How many people are you studying in total? If it's a big group, you might need a larger
sample size to get accurate results.
Margin of Error:
Think of this as the "wiggle room" in your results. If you're doing a survey and say "68% of
people like chocolate with a margin of error of +/- 5%," it means the actual number might be
anywhere from 63% to 73%.
Confidence Level:
This is about how sure you want to be that your results are right. If you say you're 95%
confident, it means you're pretty sure your results are accurate, but there's still a small chance
they could be off.
Standard Deviation:
This is like a measure of how spread out the numbers might be. If you're not sure what it is, a
safe guess (before you do your study) is to assume there's a bit of variation, like "plus or minus
5%.“
Now, the sample size formula takes all these things into account and helps you figure out how
many people you should study to get reliable results. It's like a math recipe that considers the
size of the group, how much room for error you're okay with, how confident you want to be,
and how much you think the numbers might vary.
Methods of data collection
Data can be collected from either available sources or from field studies.
A- Available sources:
1- The Census
2- Records of the Health office which include: births, deaths, compulsory immunizations
and infectious diseases
3- Case Records from hospitals, outpatient clinics and other medical centers have records
for registration of their activities (clinical, surgical, laboratory and emergency) which
provide valuable data.
B. Research work and field studies: Sometimes the required data for planning and
evaluation of health programs are not available, or unreliable.The data can be obtained
through research work and survey studies using suitable questionnaires.