Chapter 6 .Survey - Method 1
Chapter 6 .Survey - Method 1
Chhaapptteerr 66:: S
SU
UR
RV
VE
EY
YR
RE
ES
SE
EA
AR
RC
CH
HM
ME
ETTH
HO
OD
DO
OLLO
OG
GY
Y
Malaysian Teens Grade Dad a CA sample of 3212 young people from Hong Kong, Indonesia, Philippines,
Singapore, South Korea, Taiwan, Thailand and Malaysia aged between 14 and 18
took part in the survey. The survey required participants to grade their parents
from A to F for 35 statements given.
The survey was designed to allow children to reflect on and rank their
parents ability to show respect and love, have integrity and patience and be openminded, among a myriad of other skills and characteristics.
Although the Malaysian dad ranks lower than the Malaysian mum, a
general analysis of the survey results reveals fascinating details of Malaysian
fathers as seen through the eyes of young adults. For instance:
Dad got a B for telling jokes and C for being fashionable
Dad does not really allow them much independence (C-) but respects their
privacy (B) and trusts them (B).
Dad takes the trouble to teach them right from wrong (B+)
Dad accepts their own taste in music (B)
Dad talking about sex (D).
[Source: Sunday Star, 27 July, 2005, p.22]
LEARNING ACTIVITY
Read the newspaper report on Malaysian teens grade dad a C-
1. What are your views on the findings of the survey?
2. What other information would you need about the survey to accept the
findings?
INTRODUCTION
Do you want to know what people are thinking, feeling or doing? If you do, than
the survey would be the best research method. You could survey them orally through an
interview or you could give them a questionnaire. By asking questions, you are tapping
into peoples attitudes, beliefs, perceptions and behaviour. The survey has proven to be a
powerful tool in gathering information about the opinions, attitudes and intentions of
people concerning different social, cultural, economic and political issues. It is quite
common for newspapers, radio and television to report on the opinions and perceptions of
people obtained through surveys (at times using survey data to sensationalise issues).
Surveys are used by various government agencies, non-governmental agencies, business
and scientific organisations probing into peoples reactions to different issues and trends.
For example:
TV stations rely on surveys to profile the people who watch programmes over
their stations.
Automobile companies use survey to find out about customer satisfaction with
the cars they produce or sell.
Statistics Departments conduct surveys to measure the consumer price index with
the purpose of advising policy.
Engineering companies conduct surveys gathering information about the habits
of road users.
Magazine companies conduct surveys to find out about the reading habits and
interests of their subscribers.
Manufacturing companies conduct surveys to determine consumer acceptance of
their products.
WHAT IS A SURVEY?
What is a survey? According to Kerlinger (1973), survey research involves the
studying of large and small populations selecting and studying samples chosen from the
2
What they know? (Who was the first Prime Minister of Malaysia?)
What they believe? (Should students be given freedom to express themselves?)
What they expect? (Do you think you expect to be a famous person?)
What they feel? (Do you think your school principal is fair?)
What they have done? (How often do you use the computer in a week?)
What they plan? (Do you intend to continue studying or work?)
TYPES OF SURVEYS
Cross-Sectional
Survey
TYPES OF
SURVEY
Trend studies
Longitudinal
Survey
Cohort studies
Panel studies
the effects on family life of women working outside the home and so forth. To serve
these different needs, there are two main types of survey (see Figure 6.1). The types of
survey used will depend on the objectives of the study. If the study aims to get a snapshot
of opinions and practices than the cross-sectional survey would be most appropriate. If
the objective is to compare differences in opinion and practices over time than the
longitudinal survey would be the obvious choice.
a) Cross-Sectional Survey
Just like all surveys, the cross-sectional collects information from a sample drawn
from a population. It involves collecting data at one point of time. However, the time
period for collection of data can vary from 1 week to 6 months. If you are using a
questionnaire to collect data, you can ask respondents about the past, present or the
future. For example, you administered a questionnaire on habits and attitude towards
smoking to 500 students in secondary school aged between 14-16 years on 6th September,
2005. The students included male and females from different socioeconomic
backgrounds in the state of Selangor. The data you obtained is a cross-section of the
population at one point of time.
b) Longitudinal Survey
In longitudinal surveys, data are collected at different points in time in order to
study changes. There are 2 common types of longitudinal surveys:
Cohort studies: You identify a specific population (e.g. teachers in Perak who
have a masters degree) and list the names of all members of this population. At
each data collection point, you select a sample of respondents from the population
of Perak teachers with a masters degree and administer a questionnaire (e.g. about
their aspirations). At another point you might select another sample from the same
population of teachers and administer the same questionnaire. Thus, although the
population remains the same, different individuals are sampled each time. Your
aim is to see if there are changes in perceptions or trends that are present.
Panel studies: You identify a sample from the beginning and follow the
individuals over a period of time with the aim of noting changes in specific
respondents and explore reasons why these individuals have changed. For
example, you want to find out about changes in racial attitudes among a group of
primary school children. You administer an attitude scale at year 5 and than
administer the same scale when they in year 5 and so on. You analyse the data to
see if there are changes in racial attitudes as children grow older. The only
problem is the loss of subjects which you cannot replace.
LEARNING ACTIVITY
1. Define survey in your own words.
2. What is the main difference between cross-sectional surveys
and longitudinal surveys?
SAMPLING
Unlike a census, where all members of the population are studies, survey gather
information from only a portion of the population of interest the size of the sample will
depend on the purpose of the study. [We will discuss this issue later]. In a good survey
the sample is not selected haphazardly or only from persons who volunteer to participate.
It is scientifically chosen so that each person in the population will have a measurable
chance of being selected. This procedure is called sampling. This way the results can be
reliably projected from the sample to the larger population. Two key words involved in
sampling are population and sample. The word population has is defined as all people,
objects or events found in a particular group the researcher is planning to generalise to
(Borg and Borg, 1983).
POPULATION:
(Total = 10,000 persons)
Characteristics:
- Gender: 5100 males & 4900 females
- Socioeconomic status: (2500 high, 4000
middle, 3500 low)
- Ethnicity: 5000 Malays, 3000 Chinese,
1000 Indians, 1000 Ibans.
GENERALISATION
SAMPLE (n=500)
Characteristics:
- Gender: 255 males & 245 females
- Socioeconomic status: (125 high, 200
middle & 175 low)
- Ethnicity: 250 Malays, 150 Chinese,
50 Indians & 50 Ibans.
SAMPLING
Sample
Selected
from
population
(5%)
LEARNING ACTIVITY
Refer to Figure 6.2 and explain how the sample selected is representative
of the population.
SAMPLING TECHNIQUES
As mentioned earlier, surveys rely on samples to make projections about the
population. How does one select a sample? The sampling techniques used will depend on
the objectives of the study and the resources available. Generally, there are two types of
sampling techniques:
Probability sampling: Probability sampling includes techniques that select
samples based on the concept of random selection. Among the techniques that are
based on the concept of random sampling are; random sampling, systematic
sampling, stratified sampling, and cluster sampling.
Non-probability sampling: Non-probability sampling techniques are not based
on random selection. Among the common techniques are; quota sampling,
purposive sampling and convenience sampling.
ACCESSIBLE
POPULATION
SAMPLE
and independent chance of being selected as a member of the sample. Independent means
the selection of an individual is not affected by the selection of another individual. In
other words, each individual, event or object has the same probability of being selected.
For example, the number of graduate teachers in Perak is 10,000 and you intend to draw a
sample of 100 teachers (see Figure 6.3). When you select the first teacher, he or she has a
1:10,000 chance of being selected. Once this teacher is selected, however there are only
9999 cases remaining so that each teacher has 1:9999 of being selected as the second
case. Thus, as each case is selected, the probability of being selected next changes
slightly because the population from which you are selecting has become one case
smaller.
The table of random numbers is often used in the selection of a random sample
(see Table 6.1). You need to obtain a list of the 10,000 graduate teachers in Perak and
assign number 1 to 10,000. Using the table of random numbers, randomly select a row or
column as the starting point. Say, that you select Column 5. Select all the numbers that
follow in that column. So, you will select teacher assigned number 7332, followed by
teacher assigned number 6516, then teacher assigned number 4553 and so forth. If you
need more numbers, you can proceed to the next column until you have enough numbers
to make up the desired sample (i.e. n=100).
1
2
3
4
5
6
7
8
9
10
1
2345
7612
6823
4598
8956
9059
2312
8787
9063
1298
2
3445
7989
5732
2564
357
2691
6120
8236
7190
3820
3
2678
5456
3702
9860
3934
8395
6425
5153
2590
6737
4
9156
4523
3808
6360
3495
2634
6556
0202
8696
0932
5
7332
6516
4553
5245
9112
2189
3720
9530
6267
6719
6
9332
1345
4589
2347
9472
3465
0639
6490
3923
3154
7
8345
1123
0467
0391
1254
8223
5490
6220
5360
4532
8
5950
5636
8506
3623
3998
0745
2614
3523
4937
4046
9
5023
9189
6612
0620
1390
3487
2950
9691
4854
2860
10
5189
27452
12136
0850
3194
6709
7556
4917
5223
3191
You want to select a sample of 100 students from a population of 1000 students.
You divide the population by the sample needed (1000 / 10) = 10
You select at random a number smaller than 10
8
You start with that number (e.g. 7) and select every seventh name from the list of
the population.
After the required sample size has been calculated, every Nth record is selected from a list
of population members. As long as the list does not contain any hidden order, this
sampling method is as good as the random sampling method. Its only advantage over the
random sampling technique is simplicity. Systematic sampling is frequently used to select
a specified number of records from a computer file.
iii) Stratified sampling
In some surveys, you want to ensure that individuals with certain characteristics
are included in the sample to be studied. For this purpose, the stratified sampling
technique is used. For example, if you are interested in studying inductive reasoning
among 12 year olds according to ability and gender in Petaling District. You want to
ensure that you have a proportionate number of high and low academic achievers as well
as a proportionate number of males and female. In order to avoid a sample that does not
include a sufficient number of students of each sex at each ability level, a stratified
sample may be selected. All 12 year old students in the district are divided into one of the
following four groups: male high achievers, female high achievers, male low achievers
and female low achievers. Subsamples are then selected at random from the population to
fill each of the four groups (see Table 6.2).
High achievers
Low achievers
Males
Females
n=?
n=?
n=?
n=?
ACTIVITY 5.3
You have been appointed to lead a research team assigned with the task
of finding the reasons teenagers smoke. The team has decided to conduct
a nation-wide survey involving students between 14-16 years of age in
secondary schools.
1. Suggest THREE research questions that will guide data collection.
2. Explain how you plan to draw the sample of students using stratified
sampling. What subgroups would you include?
3. What further information do you need to draw a representative
sample?
LEARNING ACTIVITY
1. What is meant by random sampling?
2. What is the difference between stratified sampling and
systematic sampling?
3. How is cluster sampling different from all the other types of
sampling techniques discussed?
LEARNING ACTIVITY
N=118
N=134
N=105
N=113
N=154
N=111
N=129
N=98
N=109
The above is a diagram showing the location of 9 housing areas dispersed all over a city
with the number of senior citizens indicated in each area.
1. Explain how you would select a sample of about 200 senior citizens
you intend to interview regarding how they spend their time using
cluster sampling.
2. What factors should you consider when using cluster sampling?
10
B) Non-Probability Sampling
In many situations it is not possible to ensure that the sample will be selected based on
random selection. So the sample has to chosen by some other way. Non-probability
sampling is based on the researchers judgement and hence biasness will enter in sample
selection and distort the findings of the study. Nonetheless, non-probability sampling
techniques are used because of practical reasons. For example, non-probability
techniques are used to save costs and time; when its use can satisfactorily meet the
objectives of the study and it may be the only feasible method given the spread and
features of the population. Among the more common non-probability techniques are:
quota sampling, purposive sampling and convenience sampling.
a) Quota sampling
In situations where you are unable to ensure that subjects are randomly assigned to the
subgroups you can use quota sampling. Like stratified sampling, you first identify the
subgroups and their proportions as they are represented in the population. Following
which you select subjects based on convenience and your judgment to fill each subgroup.
When using this method, you must be confident that the chosen sample is truly
representative of the population. Obviously, you should state clearly the criteria for your
selection of the sample; especially when you make generalisations of the results to the
population.
b) Purposive sampling
The sample is selected on the basis that members conform to certain criteria. The
researcher uses his or her judgement to select cases to answer certain research questions.
The form of sampling is usually used when the population is small (such as in case study
research) and when the main purpose is to select cases that are particularly informative. It
is very useful in the early stages of an exploratory study. The main weakness of this
technique is that the sample may have characteristics which differ from population
characteristics.
c) Convenience sampling
In this sampling technique, researchers have the freedom to choose whomever they find.
You simply sample people who are easy to survey. The sample is chosen rather
haphazardly until the required sample size is met. It is less expensive and easy to
conduct; and is considered the most useful procedure to test ideas in exploratory research.
This sampling technique is considered the least reliable because of its high incidence of
biasness. Newspapers ask people to mail in their responses to a survey question; radio
stations ask people to call in their opinions to a question. As you can see, the convenience
sample can provide you with a lot of data but you do not know whether the sample
represents your population.
LEARNING ACTIVITY
1. What is the major difference between probability and nonprobability sampling techniques?
2. Why are non-probability sampling techniques used despite
their many weaknesses?
11
When the sample selected has to be broken down into smaller groups involving
comparisons of groups, a larger sample is advisable. For example, you want to
compare the self-esteem of males from low socioeconomic backgrounds from
single-parent families against males from high socioeconomic backgrounds from
single-parent families.
When attrition is expected, especially in longitudinal studies. The longer the
duration of a study, the higher will be the number of subjects who drop-out. To
reduce attrition you should keep demands on subjects to the minimum, fully
inform subjects about the study, obtain a strong commitment from subjects before
the study and make frequent contacts with subjects to maintain interest and
rapport.
When the population is highly heterogeneous on the variables being studied, you
need to have a larger sample so that the different characteristics of individuals are
satisfactorily represented. If members in the population are more or less similar,
then you will need a small sample as most of the characteristics can be captured.
LEARNING ACTIVITY
What are the factors you will consider when making decisions
about sample size in a survey?
12
LEARNING ACTIVITY
Sample Size Calculator (www.surveysystem.com/sscalc.htm)
Click the above website to access a Sample Size Calculator.
1. What is meant by confidence level and confidence interval?
2. Calculate the sample size required from a population of 9000
students with a confidence level of 95% and confidence interval (or
margin of error) of 5.
1) Defining the
objectives of
the survey
13
Order of Questions:
14
o Take into account the sensitivities of the person to whom they are addressed it is better to start with "What is your date of birth?" rather than jump straight
into "Has there been a death in the family?"
o Begin with interesting items. If the first items are boring, there is little chance
that the person will complete the questionnaire.
Step 3: Method of Conducting the Survey
You have to decide the procedures for conducting the survey; how many people
you will survey (the sample size and how they will be obtained)
Decide how you will survey your subjects (by phone, in class, mail format,
interview).
Decide how you intend to distribute and collect the questionnaires; make followup contacts.
Decide on the level of response that is acceptable (refer to your research questions
and the population the question is being asked). What percentage of
questionnaires returned or answered will you accept? What will you do if you do
not have enough respondents? Or a certain sector of respondents is underrepresented? e.g. insufficient number of females.
Step 4: Pilot-Testing of the Questionnaire
Pilot-testing a questionnaire is necessary to avoid problems that may arise when
the questionnaire is administered to the whole sample. Unfortunately, some
people consider pilot-testing nothing more than a ritual while others do not do it.
Before pilot-testing, get some of your colleagues to review the questionnaire,
particularly those who are interested or familiar with what you are doing.
The questionnaire should be pilot-tested with a small group of individuals who are
in many respects similar to the sample in the actual study. When they have
finished, ask for their opinions and suggestions (e.g. Was it too long? Which
items were difficult?).
The pilot-test helps you find out if the content and form of the questions are
satisfactory. You can also get information about:
o Length of time to complete the questionnaire
o The appropriate order of the questions
o Sufficient space is provided for responses
o The instructions or directions are understood
o Extent to which the questionnaire is reliable (if you give the questionnaire
again four weeks later would you get the same responses)
o Extent to which the questionnaire is valid (does it measure what it is
supposed to measure) [We will discuss further these issues in Module 7]
15
o results (data presented in the form of tables and graphs; statistical analysis,
description of the findings)
o conclusion (explanation for the findings, relate to previous literature and
theory, making generalisations and recommendation for future research)
LEARNING ACTIVITY
1. List some of the qualities of a good questionnaire.
2. Why is pilot-testing necessary?
confined to those who have internet access which may not be representative of the
population.
c) Telephone Interviews
Telephone interview are an efficient method of collecting some types of data and are
being increasingly used. They lend themselves particularly well to situations where
timeliness is a factor and the length of the survey is limited. The telephone interview
gives respondents the feeling of anonymity since the interviewer cannot see them.
d) Face-to-face Interview
Face-to-face or in-person interviews in a respondent's home or office are much more
expensive than mail or telephone surveys. They may be necessary, however, especially
when complex information is to be collected.
Self-Administered
Questionnaire
Survey
Mail Surveys
Telephone Interview
Group Administered
Questionnaire
e) Group-Administered Questionnaire
A sample of respondents are brought together and invited to respond to a
structured sequence of questions. This is convenient method because you are able to
capture a relatively large sample of respondents in one sitting (e.g. classroom). Also, the
response rate is relatively high. If the respondents are unclear about the meaning of
questions they could ask for clarification. However, the presence of the researcher may
make respondents feel that their answers are less anonymous and as such they may be
less candid.
ETHICS IN SURVEYS
18
As far as possible use only number codes to link the respondent to a questionnaire
and store the name-to-code linkage information separately from the
questionnaires. For example 001 for respondent Azlina Darus, 002 for respondent
Ong Mei Ling and so forth.
The names and addresses of survey respondents should not be made available to
anyone outside those involved in the survey after the responses have been entered
into the computer (Individuals and organisations have been known to sell such
databases to companies for marketing purposes without the consent of individuals
involved!).
Omitting the names and addresses of survey respondents from computer files used
for analysis
Presenting statistical tabulations using broad enough categories so that individual
respondents cannot be singled out.
Respondents should be informed about the purpose of the survey and have the
option not to participate or not to divulge information that he or she feels not comfortable
with. For example, respondents may be reluctant to disclose income. To overcome this,
you may want to use categories (e.g. RM1500 RM2000 per month) which may be less
intrusive. You should determine in the pilot-test which items respondents are
uncomfortable with, so that you do not have too many unanswered questions to the point
that some research questions cannot be answered. The questions asked should not in any
way attempt to deceive respondents. The integrity of a survey is enhanced if respondents
are clear about the purpose of the study.
19
DISCUSSION QUESTIONS:
1. You wish to study moral reasoning among fifteen year olds in a large
district. A total of 4100 fifteen year olds are enrolled in 105 classrooms in
the district. You plan to obtain a total of 250 students using cluster
sampling technique. Describe the steps you would take in selecting the
sample?
2. Discuss some of the problems with telephone interviews that may affect
the results of a survey.
3. One of the drawbacks of mail surveys is the low response rate. Suggest
how you would increase response rates.
4. When would you use a longitudinal survey rather than a cross-sectional
survey?
5. What are some problems with surveys conducted over the internet? How
can you overcome these problems?
Books
a) Mitchell, M. and Jolley, J. Research Design Explained (1988).
New York: Holt, Rinehart and Winston. Chapter 12: An introduction to survey
research. 284-311.
b) W. Borg & M. Borg, (1988), Educational Research: An Introduction, Chapter 6:
Population and sample.
20