UNIT II

Introduction to
Inferential Statistics
INTRODUCTION
Unit II of this module introduces you to the inferential type of statistics. Inferential statistics
makes use of sample data called confidence interval estimation. In inferential statistics we
make inferences about the population from which the sample is drawn and this includes
confidence interval analysis and hypothesis testing.
This part of your study, which we call Unit II, is composed of seven modules, namely
Module 8 – Introduction to Inferential Statistics
Module 9 – Random Sampling Designs
Module 10 – The Logic of Statistical Tests of Significance
Module 11 – Difference Between Means Test
Module 12 – Analysis of Variance
Module 13 – Applications of Interferential Statistics

OBJECTIVES
After completing the Unit II Modules, you will be able to:
1. see the difference between descriptive and inferential statistics
2. choose which type of statistics go with your data
3. apply the inferential statistics to your research data
4. appreciate the importance of inferential statistics in your daily activities, projects and
programs.
As in Unit I, I would like to encourage you to keep on reading the Modules. There are reference
books given at the end of the total module, take time to read them. It will work to your great
advantage. Take your time and enjoy statistics!
 8
Introduction to
Inferential Statistics

INTRODUCTION
This module introduces you to inferential statistics. Statistics as a science is comprised of two
divisions: describing and inferring. You will learn from this module the components and uses of
inferential statistics as compared from the components and uses of descriptive statistics.

This module carries mainly definitions and concepts pertinent to inferential statistics. The
mathematical portions will come in the succeeding modules. Examples of what an inference as
all about will facilitate your understanding of the application of inferential statistics to the
business profession.

So, take your time, grasp the concept laid out in every module. It will be helpful if you keep on
reading the modules – do the exercises and activities to sharpen your logical skills and data
interpretation. You will find this information very useful for your research projects such as your
Master’s thesis. So good luck, and carry on your interest in statistics.

OBJECTIVES
At the end of this module study, you will be able to:
1. differentiate inferential statistics from descriptive statistics
2. describe the terms:
2.1 population
2.2 sample
2.3 parameter
2.4 statistic
2.5 hypothesis testing

8.1 THE CONCEPTS OF DESCRIBING AND INFERRING

Statistics makes use of quantitative data. Data are from within and outside of us, for example,
our bodies, parts of our body, events, number of patients admitted in the hospital, etc. in other
words, statistics investigates a collection of events, procedures and anything around us.

We have learned from Unit I that descriptive statistics simply describes the characteristics of a
sample or a population. However, statistics is not limited to this function only. Through
inferential statistics, we can make inferences about a population through a sample.

Let us look at these objectives in a little more detail.

8.1.1 DESCRIPTIVE STATISITCS

Descriptive statistics enable us to describe the features or characteristics of a sample or a

population. For example, with an older people group, we might be interested to describe the
older people who are still actively and gainfully working after retirement in terms of age,
earning, social life, or their attitudes toward old age. Such features often referred to as the
descriptive or baseline characteristics can be analyzed by using descriptive statistics.
Descriptive statistics may include any or all of the following objectives:
 To obtain a broad overview of the distribution of the sample data, identifying any
feature and characteristic of interest, which may be present. This usually requires the
data to be arranged or organized into some form of table, as for instance, putting the
age-range of the post retirees with active, gainful employment, how many of these are
still married, remarried, widowed, etc. The representation of the organized data
maybe done through graphs, charts, or bars. You have learned these in the earlier
modules. Numeric summary values may be determined as for example, the
proportion of those who have specific leisure hobbies who are still gainfully
employed.
 To determine a numeric summary measure of the average of the sample values – for
example the average age of the post-retirees who are still actively working. Very
often there will be more than one such suitable measure from which a choice may be
made. Such measures of average are known as measures of central tendency or
measures of location.
 To determine a numeric summary measure of the degree to which the sample values
are spread out. For example, in the sample about older people group if we are
studying those with active employment after retirement, is there a similar proportion
of those who are in the 70’s or 80’s who are still employed? Again there will usually
be more than one measure to choose from. We call such measures as measures of
spread or dispersion. In sum, descriptive statistics are methods used for organizing
sample data, calculating average values and calculating values of spread.

8.1.2 INFERENTIAL STATISTICS

We have discussed in Module I that it is not practical to collect data from a population
especially when that population is quite large. Fortunately, we can gather data from a sample
representing the population. Each value that describes the population is called a parameter and
each value coming from the sample is called parameter and each value coming from the sample
is called statistic. When we try to estimate population parameters through the analysis of sample
statistics, we have inferential statistics.
Inferential Statistics enables us to determine if the results derived from the
representative sample can be generalized to the population from which it was drawn. For our
sample to be representative we must use random or probably sampling techniques in selecting
the sample. These techniques include simple random sampling, systematic sampling, stratified
sampling, and cluster sampling, which we will discuss in detail in the next module.
Furthermore, for us to be able to generalize to the population, we must do hypothesis
testing. In hypothesis testing, the validity of a hypothesis about the population is tested against
sample outcomes.
Inferential Statistics uses values determined through descriptive statistics. Thus, these
two branches of statistics complement each other. Inferential statistics usually use values such as
the mean, the standard deviation, and/or frequency counts in analyzing sample data.

Let us have some activities and questions on what we have just done.

SAQ 8-1

What is the difference between descriptive and inferential statistics?

SAQ 8-2

Give examples of studies that are require descriptive statistics vis-à-vis studies that
require inferential statistics?

SUMMARY

This module has introduced you to inferential statistics. You have seen the difference
between descriptive statistics and inferential statistics. Along the application of inferential
statistics is the exigency of sampling. The sample must be real representatives of the population
in random draw. Go on and move to the next module, to learn the various random or probability
sampling techniques. However, it is your fidelity to read and do the exercises that will gain you
skills and competence in Statistics. So, good luck!

9
Random Sampling Designs

INTRODUCTION
Designs are like patterns of dresses, of houses, or any project to suit the purpose of the designer.
In health sciences, academe, and business, research is conducted by studying population through
the use of samples. It is not possible to study the entire population. When the population size is
very large such as, say the population becomes a practical method to study the characteristics of
the population.

Recall from Module 1, a sample is a collection of individuals, or items, or characteristics taken

from a population. A population, on the other hand, contains every person, items, or
characteristics in an entire group. in this context, a population might be: every person with a
history of coronary heart disease in Manila, every mother who breastfeeds within 20 to 35 years
old and who is a career person; every BMW car bought in 1999; or every nurse who has a
doctoral degree. We can go on infinitely giving examples, but what you have to remember is that
whatever the population, it is defined to include every possible eligible member.

In this module, you will learn the different random sampling designs and procedures. Sampling
distributions are a central concept within statistics and are fundamental to significance testing
which you will study in later modules. There are numerous sampling procedures, some of them
complex. We shall examine some methods that are widely used in different field of researches.
We shall also see the reasons why we do sampling designs in researches.

So, take a relaxed mood and go over the text slowly. Be sure to do the SAQs for your mastery in
skills. Good luck and read the modules regularly.

OBJECTIVES

At the end of this module study, you will be able to:

1. Explain the reasons why we study samples instead of populations

2. Discuss the different types of sampling method such as:
2.1 simple random sampling
2.1.1 with replacement
2.1.2 without replacement
2.2 systematic random sampling
2.3 cluster sampling
2.4 stratified random sampling
2.5 multistage random sampling
2.6 discuss sampling distribution
3. Discuss sampling distributions
9.1 WHY DO WE STUDY SAMPLES AND NOT POPULATION?

I will cite here six reasons why samples are studied instead of studying the whole population:

1. Populations are usually too big for us that we do not have either the time or the
resources to locate and study every single member of them. Some populations are
infinite in size. Just imagine the difficulty and expenses you will encounter if you
interview everybody in the Philippines who has a daughter or son taking up nursing.
2. There is a need to restrict the size of the sample for ethical and safety reason. Trials of
any intervention require few volunteers who give their full consent in order to limit
any untoward or unethical consequences.
3. In experimental research designs, it may be hard to find very many suitably matched
individuals as the subjects especially in a case control design. Thus, a good sample
representative of the population might be sufficient to answer the demands of a
scientifically done study.
4. It might be very difficult to identify individuals with unusual characteristics like for
example, those carrying a virus which produces few prodromal signs and symptoms
of the disease. A sample of them is manageable.
5. Some population by nature may be too transient and too volatile to measure. An
example of this is identifying all teenagers who take “ecstasies” or “shabus”. This
type of population constantly moves residences to escape the police force.
6. The sampling process itself may be destructive. For example, you cannot take too
many biopsy or blood extractions from a patient, or test new respirators for its
efficiency and its usability limits.

What I to be borne in mind is that the sample is representative of that population. This means all
the features of the population are accurately reflected in the sample. Remember that the more
closely the sample resembles the population, the more reliable will be the conclusions about the
population based on the sample.

9.2 RANDOM SAMPLING TECHNIQUES

As mentioned In Module 8, we can make inferences about a population from a drawn sample
that is a representative of a population. To make the sample representative, we must use
probability or random sampling techniques.

Probability or random sampling is the only type of sampling that fully supports the use of
inferential statistics. In probability sampling techniques, the main goal is to select cases that
are representative of the population from which the cases were drawn. A sample is
representative if it produces the important characteristics of the population. Representativeness is
crucial for inferential statistics. If the sample is not representative, generalizing to the population
will be erroneous.
The fundamental principle of probability sampling is that a sample is very likely to be
representative if it is selected by a principle called EPSEM. This term stands for “Equal
Probability of S Election Method”. The EPSEM principle states that every element or case in
the population must have an equal probability of being selected for the sample. The goal then
is to select a representative sample and the method to be used should be the equal probability of
selection method.

9.2.1 Simple Random Sampling

Simple random sampling technique makes use of a list of all elements or cases in the
population and employs a system of selecting cases from the list that will guarantee that every
case has an equal chance of being selected. The selection process could be based on two major
kinds of operations, such as:
(a)drawing numbers or cases from a “hat” and
(b) using tables of random numbers.

This is how to use the table of random numbers

1. Assign each case on the population list a unique identification number.

2. Select cases for the sample, making sure that the identification number corresponds to
the number chosen from the table of random numbers.
3. The numbers in the table are in random order and any number is just as likely as any
other number, then the EPSEM criteria are met.
4. Stop selecting cases when you have reached your desired sample size.

To define such a sample, there is a need for you to distinguish between sampling with
replacement and sampling without replacement.

1. Sampling with replacement means that, as members of the population are selected from
the sample, they are “replaced” into the population and can be selected again. In sampling
with replacement, a simple random sample is one in which all members of the population
have the same probability to be selected. It is therefore possible that the case may be
drawn again for inclusion in the sample.
2. In sampling without replacement, once a population member is selected for the sample,
that member is not replaced into the populations and cannot be selected a second time. In
this case, a simple random sample is one in which all possible samples of a given size
have the same probability of selection. Once a case has been selected for inclusion in a
simple, it makes little sense to select that case again for the sample.

In health science research, however, the more common practice is to sample without
replacement. In both cases, though, all members of the sample are selected independently of one
another.

Conceptually, simple random sampling involves recording all names (on some identification
code) on slips of paper and placing them in a container. Suppose you want a random sample of
50 members. The 50 slips of paper are drawn without bias from the container, either with or
without replacement, the names drawn constitute the simple random sample.

Obviously, this procedure is cumbersome, and there are more practical methods for selecting a
simple random sample (with or without replacement) from a finite population. Suppose in a
study, we want a sample of 20 children from a population of 500. First, we assign each child an
identification number from 1 to 500 (that is, N is the population). Then we obtain 20 numbers by
using, either a table of random numbers, a computer, or a programmable hand calculator that will
generate a list of random numbers. Members are included in the sample if their identification
numbers are included on this list of random numbers.

A simple random sample is one, which al population members have the same probability of
being selected and the selection of each member is independent of the selection of all other
members.

Systematic Random Sampling

When a list of the members of a population is readily available, it is practical to use the
systematic random sampling techniques in selecting your sample. In systematic random sampling
only the first case is randomly selected. The succeeding sample consists if every kth case on the
list.
The first step in systematic sampling is to determine the sample size and the sampling fraction.
The sampling fraction is the ratio of the size of the sample (n) to the size of the population (N).
For example, suppose you have a population of 500 members and want a select of 5 percent
sample, that is, 5 percent or 25 of its members. The sampling fraction is
n 25 1
= =
N 500 20

Assuming that a list containing the names of the 500 population members is available and that
the list in random order, systematic sampling involves the following steps:

1. Randomly select a number between 1 and the denominator of the sampling fraction,
inclusive, using the methods described above. This is the first sample member
selected. For our example, this means randomly selecting a number between 1 and
20. If we select 16, then the first sample member between is the 16 th member on the
list.
2. To identify subsequent members, add multiples of the denominators of the sampling
fraction to the first number selected. In our example, then, the second member is (1 x
20) + 16 = 36: the third member is 56; the fourth member is 76, and so on.
3. Continue this procedure until the list is exhausted, at which point the sample members
will have been selected.

In short, the steps for this procedure can be simplified as:

 1. Divide the population size by the sample size and round the result down to the
nearest whole number, m
2. Use a random table (or similar device) to obtain a number k, between 1 and m
3. Select for the sample those members of the population numbered k, k + m, k + 2m…

Systematic random sampling is more convenient than simple random sampling when a list of the
population members is available, but it may not be as easy as this example suggests. Notice that
our example is simple because the denominator of the sampling procedure becomes slightly
more complex when the denominator is a decimal.

Moreover, systematic random sampling is not easier to execute than simple random sampling but
it also provides results comparable to simple random sampling unless there is some kind of
cyclical pattern in the listing of the members of the population (e.g. male, female, male,
female….)

A systematic random sample is a probability sample in which every kth member of the
population is selected and in which 1/k is the sampling fraction. The first member of the sample
is determined by randomly selecting an integer between 1 and k.

Cluster Sampling

Sometimes, members of a population occur in naturally formed groups called clusters. For
example, group of staff nurses in a certain hospital or even total hospitals, constitute clusters.
When clusters like these exist and it is impractical to select individual members for the sample,
you can use the procedure called cluster sampling.

In cluster sampling, clusters – not individual members – are randomly selected from the
population of clusters. Each members of the population must belong to only one cluster, but the
clusters may have different numbers of members. Once a cluster is selected, all members of that
cluster are included in the sample.

As another alternative to simple random sampling, this method is particularly useful when the
members of the population are widely scattered geographically.

In the simplest case, cluster sampling is implemented using three steps: (1) divide the population
into groups (clusters); (2) obtain a simple random sample of the clusters; (3) use all of the
members of the clusters obtained in step 2 as the sample.

Although cluster sampling can save time and money, it does have disadvantages. Ideally, each
cluster should mirror the entire population. However that is often not the case – members of a
cluster are frequently more homogenous than the members of the population as a whole. This
situation can cause problems.
Cluster sampling involves the random selection of clusters (group of population members)
rather than individual population members. When a cluster is selected for the sample, all
members of that cluster are included in the sample.

Stratified Random Sampling

In simple random sampling, systematic sampling, and cluster sampling, the population is
assumed to be generally homogenous. However, a population may instead be heterogeneous and
consist of several subpopulations, which are called strata.

Another sampling method, known as stratified random sampling, is often more reliable than
cluster sampling. In here, the population is first divided into subpopulations or strata, and then
sampling is done for each stratum. Ideally, the member of each stratum should be homogenous
relative to the characteristics under consideration.

Suppose the people in a certain town are classified according to income-size. Using the stratified
sampling method, the population can be divided into three strata: upper income, middle income,
and low income. A simple random sample can then be taken from each of the three strata.

This stratified sampling procedure will ensure that no income group is missed. It will also
improve the precision of the statistical estimates (since the individuals within each income group
tend to be somewhat homogenous) and would make it possible to estimate the separate opinions
of each of three strata.

In stratified sampling, the strata are often sampled in proportion to their size: this is called
proportional allocation. For instance, suppose the strata consisting of three income groups
(upper, middle, low) comprise respectively, 10%, 60%, and 30% of the town. Then for a sample
size of 100, say the number of upper income, middle income, and low income individuals
sampled will be, respectively 10 (10% of 100), 70 (70% of 100), and 20 (20% of 100).

The simple type of stratified random sampling, called stratified random sampling with
proportional allocation, is implemented using three steps: (1) divide the population into
subpopulations (strata); (2) from each stratum, that is, the sample size equals the total sample
size divided by the population size; (3) use all of the members obtained in step 2 as the sample.

Stratified random sampling yields a representative sample. By providing for a sampling

throughout the entire population, it ensures that members from each stratum are included in the
sample.

In stratified random sampling the population, into subpopulations called strata. All strata are
represented in the sample, often through proportional allocation according to size.
Multistage Sampling

Most large-scale surveys combine random sampling, systematic random sampling, cluster
sampling, and stratified random sampling in ways that can be quite complex. Such multistage
sampling is used frequently by pollsters and government agencies.

for instance, the National Census and Statistics Office conducts surveys on all residents of the
Philippines to obtain information on sociodemographic characteristics (age, sex, income,
occupation, educational attainment, etc.) of each of those included in the population. Data
collection is by a multistage probability sample of approximately 65 million Filipinos.
Information obtained from the surveys will then be of help to the Philippine government in terms
of planning and implementing various socio-economic-political programs.

In multistage sampling, the combination of simple random sampling, systematic random

sampling, cluster sampling, and stratified random sampling will be utilized.

9.3 SAMPLING DISTRIBUTIONS

The frequency distributions, which we dealt in Module 2, represented single samples. These
distributions showed how many times various scores were obtained within a single sample. In
contrast, sampling distributions may be defined as theoretical frequency distributions that
depict the frequency of occurrence of values of some statistic computed for all possible samples
of size N (or pairs, triplets or other multi-sample sets) drawn from some population. Let us look
more closely at this definition of sampling distribution.

First, sampling distributions are theoretical or imaginary entities. Although it is theoretically

possible to draw all possible samples (or sets of two or more sample) from a population, to do so
would be a practical impossibility, at least if the population is of any substantial size.

Second, sampling distributions do not display the frequency of occurrence of scores on some
variable, but depict instead how often various values of some statistic are obtained as the statistic
is computed for samples drawn from a population. For example, the sampling distribution of the
mean, which we will consider in greater detail in a moment, would be formed by drawing all
possible samples of size N from a population, computing the mean for each sample, and then
plotting the frequency of occurrence of each of the various values of the men obtained from one
sample to the next. Other sampling distributions might involve computing a statistic other than
the mean. We can imagine, for example, drawing all possible samples of size N from a
population, computing the variance of each sample, and plotting the number of times different
variance values occurred across the many samples. This would be the sampling distribution of
the variance.
Third, sampling distributions are not always based on drawing all possible samples from a
population. Some sampling distributions are created by drawing all possible pairs of samples,
triplets of samples, quadruplets of samples, or other sets of several samples. Then, a statistic is
computed for each pair, triplet, quadruplet, or other set of samples, and the frequency of
occurrence of the various values of this statistic is plotted, forming the sampling distribution.

Bear in mind that useful sampling distribution exist only in the imagination. No one would
actually draw all possible samples from a population, compute a statistic on each, and plot
frequency of occurrence of the values obtained. It is not necessary to do this because we know
what the sampling distributions look like even without actually forming them.

9.4 THE SAMPLING DISTRIBUTION OF THE MEAN

If you were to draw all possible samples of some size (N) from a population and compute the
mean for each sample, you would find that the values of the means would vary somewhat from
sample to sample. This variability should seem intuitively reasonable. Even though the samples
were all drawn from the same population, we would not expect their means to be perfectly
identical to each other. This variability in sample means seen from one sample to the next is due
to sampling error. Because no sample is likely to match the population from which it was drawn
in every way, the samples can be said to be “in error,” thus the term sampling error. Because the
means computed from one sample to the next vary, it is possible to count and plot the number of
times each value of the mean occurs, forming the sample distribution of the mean.

As stated earlier, the sampling distribution of the mean is usually a theoretical entity, existing
only in the imagination. Nonetheless, if the population from which samples are being drawn is
very small, it is possible to actually draw all possible samples of size N, compute the mean for
each sample, and plot the distribution of the mean. For the sake of making the concept of the
sampling distribution of the mean more concrete, we will construct one here.

Table 7.1 Scores for the Entire Population of Sto. Niño Town in Cebu

Inhabitant IQ
A 80
B 90
C 100
D 110
E 120
F 130
Shown in Table 7.1 are IQ scores for the entire population of the Sto. Niño town in Cebu. In
order to construct the sampling distribution of the mean for sampling of size N=2, we would
draw all possible samples of size 2 from this population, compute the mean for each, and plot the
frequency of occurrence of each of the mean values. All possible samples of size N=2 and their
sample means are listed in Table 7.2

Sampling distributions are formed using sampling with replacement. Once a case has been
selected for inclusion in sample, it is put back into population, that is, “replaced”. It is therefore
possible that the same case may be drawn again for inclusion in the same sample. That
statisticians have developed sampling distributions based upon the procedure of sampling with
replacement is evidence of the discrepancy between pure and applied statistics. In research
practice, sampling without replacement is the established procedure. Once a case has been
selected for inclusion in the sample, it makes little sense to select that the case again for the same
sample! Using sampling with replacement, we could conceivably end up studying a sample size
N=100 consisting of the same case drawn 100 times! Although this is mathematically acceptable,
it is hardly acceptable from a practical standpoint. The fact that applied statistical procedures rely
upon sampling without replacement, while statistical theory assumes sampling with replacement,
is one of those things in statistics that that we try not to worry too much about. In fact, there is
little or no cause for concern, since the practice of sampling without replacement has no
noticeable effect on the accuracy of the conclusions that are reached, provide we confine our
attention to relatively large samples drawn from very large populations.

Table 7.2 Sampling Distributions of the Mean Formed Using Sampling with Replacement

Samples Scores X
A,A 80,80 80
A,B 80,90 85
A,C 80,100 90
A,D 80,110 95
A,E 80,120 100
A,F 80,130 105
B,B 90,90 90
B,C 90,100 95
B,D 90,110 100
B,E 90,120 105
B,F 90,130 110
C,C 100,100 100
C,D 100,110 105
C,E 100,120 110
C,F 100,130 115
D,D 110,110 110
D,E 110,120 115
D,F 120,130 120
E,E 120,120 120
E,F 120,130 125
F,F 130,130 130

All possible samples (drawn with replacement) of size N=2 from the population of six
inhabitants of Sto. Niño in Cebu City. Listed with each sample are IQ scores and the sample
mean (X)
Characteristics of the Sampling Distribution of the Mean

Although we were to construct an actual sampling distribution of the mean just above, it was
only because the population in the example was unrealistically small and our samples were much
tinier that would ever be the case in actual practice. Normally, populations are very large and the
samples we study are of a fair size as well. Under these conditions, constructing the sampling
distribution of the mean would be most tedious, to say the least. Fortunately, we need not
actually construct the sampling distribution of the mean in order to know its characteristics. If we
are drawing fairly large samples (> 50) from a very large population, the central limit theorem
tells us that the sampling distribution of the mean will have the following features:

1. The sampling distribution of the mean will be approximately normal in shape, and will
become increasingly normal as the size of the samples increases. This is true regardless
of the shape of the population’s distribution, although if the population is normally
distributed, the sampling distribution of the mean will approximate the normal
distribution at smaller sample sizes. That the sampling distribution of the mean is
normally distributed is important because this tells us about the relative likelihood of
finding the sample means that deviate by specified amounts from the mean of sampling
distribution of the mean.
2. The mean of this distribution, abbreviated with the symbol μχ will equal the population
mean μ.
3. The third thing we know about the sampling distribution of the mean from the central
limit theorem is the variability of its distribution. The standards deviation of the sampling
distribution of the mean called the standard error of the mean abbreviated with the
symbol σχ is computed in the following equation:

σ s
σχ = ∨
N N

Where: σ = population standard deviation

S = the corrected sample standard deviation
N = the size of the samples drawn from the population

From the sampling distribution of the mean depicted in Table 7.2, the equation given us:

σ s
σχ = ∨
N N

17.08
=
2

= 12.08
We are already familiar with the standard deviation as a measure of variability as discussed in
Module 3 and 4. As the standard deviation of the sampling distribution of the mean, the standard
error of the mean provides an approximate measure of the average absolute deviation of sample
means around μχ. When we say then that σχ = 12.08, we are saying that the average sample mean
deviates from μχ by about 12 points.

The equation reveals that the standard error of the mean will always be less than the standard
deviation of the population distribution. This is because sample means are already the averages
of all their constituent raw scores and can therefore never be as extreme as any single score.
Further, the larger the samples (N) drawn from the population, the lower σχ will be, since only
rarely will the mean of a large sample deviate substantially from the mean of the population.

That was a long text, is it not? It is sometimes unavoidable to break text discussions in statistics
because a certain line of continuity logic must be sustained. Also, texts like this need to be re-
read until the logic becomes familiar and clear. So, now before we go further, let us do some
SAQs.

SAQ 9-1

(a) Is it practical for your master’s thesis to study all the household heads in Palawan given a
study period of one semester?
(b) Why?
(c) What would you do to make the study possible within time and within your resources as a
master’s student?
(d) How would you go about the study?

SAQ 9-2

(a) What do you mean by multistage sampling?

(b) Can you give an example of a study that utilizes multistage sampling?

SAQ 9-3

(a) What is the basis for all types of sampling distribution?

SAQ 9-4

(a) What is the standard error of the mean?

 (b) What is its equation?

SAQ 9-5

How do you go about selecting 300 people from a population of 10,000? Please show your
process in steps using systematic random sampling.

Please do these exercises for your own skills in doing sampling designs and distribution.

Now let us look at some other sample study designs.

9.5 OTHER SAMPLE STUDY DESIGNS

Ways to carry out investigations in the field of health care can be done through various ways
such as:
(a) Observational versus experimental studies
(b) Prospective versus retrospective studies
(c) Longitudinal versus cross-sectional studies

The observational versus experimental studies. What differs essentially between

observational and experimental studies is that in the observational designs, we simply observe
what is going on in different groups of individuals without having any control over who or what
is in the group. for example, we observe toddlers and adolescents at play. We simply observe and
record what is observable. In the experimental designs, we introduce an intervention like for
example, we give health teachings on diet preparation to one group and another intervention to
another group and get another group without any intervention to serve as the control group.
Observational studies include the following: (a) cohort studies, (b) case-control studies, and (c)
cross-sectional studies. Experimental studies include (a) interventions, (b) clinical trials, (c)
pre-post test, and (d) time-series repeated measures. In observational studies, the investigator
mainly observes and documents all the observable variables; in the experimental studies, the
investigator has control over the study group versus the control group.

Cohort Studies

Cohort studies may be longitudinal or prospective. A group of individuals with asthma or with
nephritis maybe studied over a period of time until some endpoint as for example until death or
until the development of the ailment is reached, Cohort studies are time-based such as those
diagnosed in 1999 only are the participants. Studies may also take into consideration groups of
people taking care of TB patients and over a period of time. Take note the probability of those
being infected with the disease for those exposed to TB. In this sense, we take the word risk, as
the probability of getting the disease when exposed to an individual with the same disease.
Case-control Studies

Case-control studies maybe retrospective because studies go backward to trace what had been
the cause or the consequence. To put this point clearer, let us take as an example a sample of
babies who had died from sudden infant death syndrome (SIDS) as the cases and a sample of
healthy babies of the same age as the controls. Then we try to find out what factors were
different from the two groups based on the hypothesis made. In this manner we shall see
variance between the case group and the control group. Usually this study is done on a particular
type of ailment.

Cross-sectional studies

Cross-sectional studies look at a particular population at some moment in time over a short
period of time rather than looking back or forward over a more extended period. The sample
studied may be from a definite population or from a group of patients with a particular disease.
Because of the nature of this study, cross-sectional studies are descriptive in nature. This study
makes use of surveys and data are gathered through questionnaire, interviews or focus group
discussions as well as key informant discussion.

Experimental Studies

Experimental studies make use of a treatment trial on one group and a placebo or conventional
treatment to one group. Many researches in nursing are done using this type of study, applying
one intervention on one sample and having a similar or matched sample as a control group. For
experimental studies, randomization of subjects is a must; samples should be more than 15 in
each the treatment and in the control group so that a statistical can be applied to its analysis.

One way of protecting biasness of sampling in experimental designs is the use of single blind and
double blind designs. To observe the effect of the treatment, it is advisable to subject the
observed scores of the research observers to a coefficient of reliability to avoid bias of
contamination of responses.

SAQ 9-6

Write a brief summary about study designs.

SAQ 9-7

Apply an appropriate sampling design to your projected research. How did you fare in your
responses when compared with the ASAQs? Statistics needs practice and going over the text
several times. Do not be discouraged if you did not get the answers right. Remember, practice
and exercises make your skills perfect.
SUMMARY

You have just gone through this module on sampling designs and sampling distribution. You
have seen the reasons why sampling is necessary and what to guard to help your representative
of your population.

With the knowledge on the different sampling methods, the best that you have learned in this
module is your having conceptualized a research problem and apply the appropriate sampling
design as well as sampling distribution.

So again, congratulation for having done this module and I encourage you to move ahead to the
next module. Do not give up, keep on trying until you finish the course. It will pay you off later.