CHAPTER 6

POPULATION AND SAMPLE

6.1 POPULATION

According to Husaini Usman, the population is all the values of the results
of calculations and measurements, quantitative and qualitative, of certain
characteristics regarding a group of objects that are complete and clear. Example of
the population is: All accounts registered in the BRI branch in Renon

6.2 SAMPLE

According to Riduwan (2007: 56) the sample is part of the population that
has certain characteristics or circumstances researched. Example of the sample: BRI
branch in Renon is being audited for the level of error in recording its account.
Instead of observing all the accounts registered that have total of 5,500 accounts,
an auditor can only choose and observe sample as many as 100 accounts.

6.3 RESEARCH USING POPULATION AND SAMPLE

6.3.1 Population
1. Basic Concepts. Basically, population or universe research is a
psychological object or item that is limited by certain criteria. Therefore in the
research it needs a clear population size. Population size refers to the number of
psychological objects in the population. Usually, population size in research is
represented by the letter N (capital).
2. Form of Research Population Size. The population quantity there are
countable and some uncountable. Therefore, population size in business research
can be divided into 2 forms, namely:

a. Finite Population. Finite population is a research population whose numbers can

still be determined or can be calculated. For example: The number of individual
taxpayers registered in the Tax Office Pratama North Badung on 2016 was 60,528
people

b. Infinite Population. Infinite Population is a population size that has been so

large that the number cannot be calculated again. Sometimes the number of
individuals in the population is not fixed or the number is infinite. For example:
The number of people visiting the Tax Office Pratama North Badung.

3. Method of Collect Information on Population in Research. Data

obtained in a research is objective, valid and reliable data. In addition,
representative data can be obtained from the population researched. To achieve
these objectives, a method of collecting population information is needed, namely:

a. Complete Enumeration. Collected by counting each unit or unit in the

population. In social research including business, researchers often face an infinite
population. Therefore researchers must first determine the target population. This
population will later become the scope of the conclusions in the study.

b. Sample Enumeration. Collected by only counting a portion of the population

according to the characteristics desired by the population. So the description of the
population is taken from the representative of the population or from the sample
(part of the population that researched).

6.3.2 Sample
1) Basic Concepts. Basically the sample is some psychological object or
member of the population taken according to certain procedures. Sample as part of
a population that has certain characteristics or circumstances to be researched.
Whereas sampling is the process of choosing part of a psychological object from a
population.In the process of selecting a sample, a sampling unit is obtained from
the sampling frame of the object that are researched.
2) Sampling Unit. Sampling unit is everything that the researcher makes as
a unit that will later be used as the object of research. The form of a sampling group
can be an individual and can also be a collection of individuals or groups. Example:
If in the Tax Office Pratama North Badung on 2016 there are 60,528 people
registered as individual taxpayers and in the research, individual taxpayers selected
as research data, so that is mean the individual taxpayers is the sampling units.
3) Sampling Frame. Sampling frame is a list of sampling units in a
population. For example: If in Tax Office Pratama North Badung on the 2016 there
are 60,528 people registered as individual taxpayers, and in the research, individual
taxpayers selected as research data, so the list of individual taxpayers is the
sampling frame

6.4 GOOD SAMPLE CRITERIA

6.4.1 A good sample allows the researcher to make decisions related to the size
of the sample to obtain the desired answer
6.4.2 A good sample identifies the probability of each unit of analysis to be a
sample.
6.4.3 A good sample allows the researcher to calculate accuracy and influence
(for example errors) in sample selection rather than having to do a census.
6.4.4 A good sample allows the researcher to calculate the degree of confidence
specified in the population estimate compiled from a statistical sample.
6.5 SAMPLE SIZE
Roscoe (1975) quoted by Uma Sekaran (2006) provides a general reference
for determining sample size:
6.5.1 Sample sizes of more than 30 and less than 500 are appropriate for most
research
6.5.2 If the sample is broken into subsamples (male / female, junior / senior, etc.),
the minimum sample size of 30 for each category is appropriate
6.5.3 In mutivariate research (including multiple regression analysis), the sample
size should be 10 times greater than the number of variables in the research
6.5.4 For simple experimental research with strict experimental controls,
successful research is possible with a small sample size of 10 to 20
The sample size is very depend on the amount of accuracy or error that the
researcher wants. However, in terms of error rates, the maximum social error rate
is 5% (0.05). The greater the error rate, the smaller the number of samples. But what
needs to be considered is that the greater the number of samples (the closer to the
population) it’s mean the smaller opportunity of generalization errors, the smaller
the number of samples (away from the population) the greater opportunity of
generalization errors. Some formulas for determining the number of samples
include:
 Slovin formula (in Riduwan, 2005: 65)
 n = N / N (d) 2 + 1

n = sample; N = population; d = 95% precision value or sig. = 0.05.

For example, the total population is 125, and the desired error rate is 5%,
then the number of samples used are: N = 125/125 (0.05) 2 + 1 = 95.23,
rounded 95

 Formula Jacob Cohen (in Suharsimi Arikunto, 2010: 179)

N=L/F^2+u+1

Information :

N = sample size

F ^ 2 = Effect Size

u = Number of changes related to research

L = Power function of u, obtained from the table

Power (p) = 0.95 and effect size (f ^ 2) = 0.1

Price of L table with t.s 1% power 0.95 and u = 5 is 19.76

then with the formula the sample size is obtained

N = 19.76 / 0.1 + 5 + 1 = 203.6, rounded 203

 Formulas based on the proportion or table of Isaac and

Michael
The table for determining the number of samples from Isaac and
Michael provides the ease of determining the number of samples based on
1%, 5% and 10% error rates. With this table, researchers can directly
determine the sample size based on the number of population and the desired
error rate.

6.5 CONSIDERATION OF SAMPLE SIZE DETERMINATION

The number of samples suitable for a study is influenced by several factors, namely
as follows (Davis & Cosenza, 1993: 222-2231.

6.5.1 Homogeneity. The more homogeneous a sample selection unit is, the
smaller the number of samples needed.
6.5.2 Degree of Trust. The degree of trust measures how far the researcher
believes in estimating the population parameters correctly. The degree of trust is
usually expressed in probability, for example 95%.
6.5.3 Precision. Precision measures the standard errors of the estimates made. In
other words, the expectation of deviation from the population is calculated by the
standard deviation.
6.5.4 Analysis Procedure. Some specific analysis models require a certain number
of samples. Researchers need to consider the number of samples needed
according to the analysis model that will be used.

6.6 SIZE OF SAMPLE

The size or size of the sample is very dependent on the level of accuracy or error
that the researcher wants. However, in terms of error rates, the general reference to
social research has a maximum error rate of 5% (0.05).

Here are some conditions that need attention;

6.6.1 If the research being carried out is descriptive research, the sample size is at
least 10% of the total elements of the population.
6.6.2 If the research done is a correlation or related research, then the sample size
is at least 30 subjects (sample units).
6.6.3 If the research done is a comparative study, the recommended sample size
of the research is 30 subjects.
6.6.4 If the research done is a group experiment, the recommended sample size is
15 sample groups.

6.7 SOURCE OF SAMPLING ERROR

6.7.1 Random Variation. Random variations are the most common source of
sampling errors. The presence of a mistaken guess is rather easily detected if the
information obtained is clearly doubtful, but if the estimation error is not so large,
of course the error that appears to be difficult to detect so that ultimately the
information obtained will lead to the wrong conclusion
6.7.2 Specification error. Errors caused by misrepresentation of specifications
are very common in opinion making for elections. Specific errors can also arise due
to incorrect lists of population elements (population frames), incorrect information
in the inventory logbook, incorrect selection of sample members (such as replacing
the intended respondent with a neighbor if the respondent who was supposed to be
found was not in place) ), question sensitivity, errors in gathering information about
samples caused by intentional or unintentional interview bias, or errors in
processing sample information.
6.7.3 Error Determining Respondents (mis-specification of sample subjects).
The source of additional errors in the sample survey was caused by errors in the
determination of respondents from several sample members. In general, researchers
assume that respondents and non-responders represent similar layers of the
population when in fact this is a rare case.
6.7.4 Coverage errors. One of the keys to success from a good sample selection
is the availability of a list of relevant complete population elements. Errors due to
incomplete coverage of the population list (coverage error) arise because of the
unavailability of a list of certain groups in the list of population elements. These
conditions make the individual members of the group not likely to be selected as a
sample and result in a bias in the election.
6.7.5 Nonresponse error. Not every respondent is willing to respond to a survey.
Experience shows that individuals in the upper and lower economic classes tend to
be less responsive to surveys than those in the middle class. Errors due to
incomplete response (nonresponse errors) arise from failure to collect data from all
individuals in the sample
6.7.6 Sample Withdrawal Error (Sampling error). It is believed that a good sample
is a miniature of the population. Even so, repeated sampling usually results in a
different amount of population characteristics between one sample to another. In
this case the sampling error reflects the heterogeneity of the probability of the
emergence of differences from one sample to another because of the individual
differences chosen from the various samples.
6.7.7 Measurement error. Measurement error refers to inaccuracy in recording
the response given by the respondent because of the weakness of the instrument in
the main question, the inability of the question or because the statement made tends
to direct the respondent's answer.

6.8 SAMPLE SELECTION PHASE

6.8.1 Population Determination. The first process for selecting samples is

determining the population. The unit of analysis is as an individual head of
household, student, trader), organization (for example a distributor, manufacturing
company), or it can be a company (for example: car, toothpaste).
6.8.2 Determination of the Sample Selection Unit. The sample selection unit is
a group of elements. From the research population the elements that will be grouped
into one or several groups depend on the sample design used by the researcher.
6.8.3 Determination of Sample Selection Framework. The sample selection
framework is a list of elements from each sample selection unit. Research on first-
year students, for example, can use the list of names of first-year students that can
be obtained in the administration section.
6.8.4 Sample Design Determination. Sample design is a method for selecting
samples from existing populations. There are several types of sample designs that
can be used by researchers.
6.8.5 Sample Number Determination. As is known, the data to be analyzed is
obtained from the research sample. Thus the greater the number of samples, with
the correct sample design, of course the data obtained will increasingly represent
the population under study.
6.8.6 Sample Selection. The final step in the sample selection process is to
choose the sample needed. In this step the researcher determines the elements that
will be the sample of the research conducted.

6.9 SAMPLING MAKING METHOD OR SAMPLING TECHNIQUES

 In general, the sample design consists of two types, namely probability
design and non-probability design. There are 5 types of probability sample designs,
namely: random samples, systematic samples, stratification samples, cluster
samples, and multistage samples.

Sample Type Description Advantage Disadvantage

Simple Random Each element of Only requires a Requires more of
Sampling the population has little knowledge. more population
the same elements.
opportunity to be
selected as a
sample.
Systematic Select samples Cheaper than Periodic
Sampling from the simple random population allows
population from sampling data and results to
the start and follow be distributed
the sample abnormally.
selection based on
the order of
elements.
Stratification Researchers The results are If the subsample
Sampling divided the more selected on a
population into representative of different basis
groups and the overall will increase the
randomly selected population so that error.
subsamples from it increases
each group efficiency
statistically.
Cluster Sampling Groups that have More efficient Researchers must
heterogeneous because have the ability to
characteristics are economically divide into truly
identified first and specific clusters.
 then randomly than simple
selected random samples.
Multistage The researcher Provide more Researchers may
Sampling chooses a small accurate be reluctant to do
area for each stage information. so because they
and combines the have to be
four sample repetitive.
techniques above
There are 4 types of probability sample designs, namely: convenience
sampling, judgment sampling, quota sampling, snowball sampling.

Sample Type Description Advantage Disadvantage

Convenience The researcher Does not require a Biased variability
Sampling uses the simplest long population and estimation
or most list. cannot be
economical measured or
sample. controlled.
Judgment Experienced or The sample The results are
Sampling experienced ensures that the biased because
researchers choose objectives to be the sample is not
samples to fulfill achieved must be representative.
their objectives, achieved.
such as ensuring
that all populations
have certain
characteristics.
Quota Sampling The researcher Do not need a Deviations from
classifies the population list the population
population anymore cannot be
according to certain estimated due to
criteria (pertinent the use of non-
properties), random selection.
 determines the
desired sample
proportion for each
class, sets the quota
for each
interviewer
Snowball Initial respondents Useful in The result is
Sampling are selected with allocating biased because
probability members from a the number of
samples while the small population. samples is not
next respondents independent.
are obtained from
previous
respondents'
suggestions / input
Bibliography:

Cooper, Donald R. 2006. Metode Riset Bisnis Volume 1. Jakarta: PT Media Global
Edukasi

Sekaran, Uma. 2006. Research Methods for Business Buku 2. Jakarta: Salemba
Empat

Sekaran, Uma. 2007. Research Methods for Business Buku 1. Jakarta: Salemba
Empat

Sugiarto, dkk. 2001. Teknik Sampling. Jakarta: PT Gramedia Pustaka Utama

Supriyanto. 2009. Metodelogi Riset Bisnis. Malta Pritindo: Bandar Lampung