Learning objectives

The objectives of this chapter involve understanding:

■ the difference between primary and secondary data and between
quantitative and qualitative data
■ the range of different research methods and sources of data
used by sociologists and an assessment of their strengths and
limitations
■ the stages of research design: deciding on research strategy;
formulating research problems and hypotheses; sampling and
pilot studies; conducting the research; interpreting the results and
reporting the findings.
 ,,.,?-~·.-.·

; Chapter 3:,. Methods o.f.re~earch·

Strength s
The distinctions between
The researcher is able to save time, money and effort by using
primary and secondary data existing data such as official government statistics about
and between quantitative and crime, marriage or divorce. There may also be situations,
where secondary data is the only available resource, such
qualitative data as when researching suicide. Secondary data is also useful
Primary data for historical and comparative purposes. Aries (1962), for
example, used historical paintings and documents to support
Primary data involves information collected personally
by a researcher. They may use a range of methods, such as his argument that childhood was a relatively recent invention.
questionnaires, interviews and observational studies.
Some forms of secondary data, such as official
statistics, may be highly reliable because the data is
collected consistently, in the same way from the same
sources. This type of data is also more likely to represent
Primary data: information collected personally by a what it claims to represent. Many countries, including
resea rcher. Britain, India and Mauritius, conduct a census of every

St rengths
household every 10 years, which collects representative
data that can be used as a reliable secondary source. I
The researcher has complete control over how data is
collected, by whom and for what purpose. In addition,
where a researcher designs and carries out their own Official statistics: government-generated secondary
research they have greater control over the reliability and source of data on areas such as crime, marriage and
employment.
validity of the data, as well as how representative it is.

Limitations
Limitations
Primary research can be time-consuming to design,
construct and carry out, especially if it involves personally Secondary data is not always produced with the needs
interviewing large numbers of people. Primary research of sociologists in mind. For example, official definitions
can also be expensive. In addition, the researcher may of poverty, class or ethnicity may be different from
have difficulty gaining access to the target group. Some
people may refuse to participate or, in the case of historical
research, potential respondents are no longer alive.

Secondary data
Secondary data is data that already exists in some form,
such as documents (government reports and statistics,
personal letters and diaries) or previous research
completed by other sociologists.

Secondary data: data that already exists; data not

personally generated by the researcher.
Why might personal websites involve unreliable secondary data?
 Cambridge International AS and A Level Sociology

sociological definitions. Sources, such as personal Matveev (2002) argues that quantitative research is more
documents, can be unreliable. Official crime statistics reliable because it is easier to repeat (replicate) the study.
may not measure all crimes but only those reported to Standardised questions that do not change, for example,
the authorities. Some forms of secondary data, such can be asked of different groups or the same group
as historical documents, may only reflect the views at different times. The results can then be quantified
of a single individual rather than representing wider and compared. If the answers are the same, or very
opinions. similar, then the research is more likely to be reliable.
Quantitative data also makes it easier for the researcher to
Quantitative data remain objective. They not need to have a close personal
Quantitative data expresses information numerically, in involvement with the subjects of the study, so their
one of three ways: personal biases are less likely to intrude into the data-
collection process.
■ a raw number, such as the total number of people who
live in a society
■ a percentage, or the number of people per 100, in a
population; for example, around 80% of Indians follow
the Hindu religion
a rate, or the number of people per 1,000 in a
population; a birth rate of 1, for example, means that
for every 1,000 people in a population, one baby is born
each year.

I Quantitative data: information expressed numerically

that captures the 'who, what, when and where' of
behaviour.

Strength s
Quantitative data, such as that collected from questionnaires,
The ability to express relationships statistically can be
is intended to limit subjective judgements by posing uniform
useful if the researcher does not need to explore the questions and even the choice of responses. Does quantitative
reasons for people's behaviour - if they simply need to data minimise subjective judgements?
compare the number of murders committed each year in
different societies. Kruger (2003) argues that quantitative
Li mit at ions
data 'allows us to summarize vast sources of information
Quantification is often achieved by placing the respondent
and make comparisons across categories and over time'.
in an 'artificial social setting' in order to control the
Statistical comparisons and correlations can test whether
responses and the data collected. People rarely, if ever,
a hypothesis is true or false. They can also track changes in
encounter situations where they are asked to respond to a
the behaviour of the same group over time (a longitudinal
list of questions from a stranger, or have their behaviour
study) .
observed in a laboratory. Some argue that it is impossible
to capture people's 'normal' behaviour or collect 'real'
responses when the subjects are placed in such an artificial
environment.
Correlation: a statistical relationship between two or
A further problem is that quantitative data only
more variables that expresses a level of probability. A high
(positive) correlation suggests the strong probabi lity of captures a relatively narrow range of information. Day
a relationship; a low (negative) correlation suggests the (1998) calls this the 'who, what, when and where' of
probability of little or no relationship. people's behaviour. Quantitative data does not usually
reveal the reasons for behaviour because it lacks
 Chapter 3: Methods of research

depth; the more detailed the behavioural data, the

more difficult it is to quantify. As a result, quantitative
Participant observation: research method that invo lves
data is often seen as superficial. As Kruger argues, it's
the resea rch er parti cipatin g, openly or secretly, in the
'difficult to get the real meaning of an issue by looking at
behaviour they are stud yin g.
numbers'.
McCullough (1988) suggests that a significant
limitation of quantitative data is that 'issues are only
measured if they are known prior to the beginning of establishing a strong personal relationship or rapport
the research'. To quantify behaviour, the researcher with respondents in order to experience their lives.
must decide in advance what is and is not significant By collecting qualitative data in this way, researchers
in terms of the behaviour being studied. There is no have greater freedom to study people in their 'normal'
opportunity to develop the research beyond its original settings. The results are more likely to show how people
boundaries. A final limitation is what Sorokin (1956) really behave and what they really believe. The ability to
calls 'quantophrenia' - quantification for its own sake, capture the richness of people's lives through qualitative
regardless of whether it tells us anything useful about the data is also an important strength. Matveev suggests that
behaviour being quantified. qualitative methods and data allow the researcher to gain a
'more realistic feel of the world that cannot be experienced
. ., . .
through numerical data and statistical analysis'.
·, TEST YOURSELF
• ~. C \ • • • •

Limitations
Briefly explain the idea that quantitative data captures
only the 'who, what, when and where' of behaviour. Qualitative research focuses on the intensive study of
relatively small groups, which may limit the opportunity
for applying the data more widely - such groups may not
be representative of anything but themselves. For similar
Qualitative data
Qualitative data aims to capture the quality of people's
behaviour by exploring the 'why' rather than the 'what,
reasons, it is difficult to compare qualitative data across
time and location because no two groups will ever be
I
qualitatively the same (the research does not compare
when and where'. It involves questions about how people
'like with like'). The depth and detail of the data also
feel about their experiences and can be used to understand
makes such research difficult to replicate, which means
the meanings applied to behaviour. For example, in the
its reliability is generally lower than that of quantitative
USA Venkatesh (2009) studied a juvenile gang from the
research.
viewpoint of its members, while Goffman ( 1961) examined
the experiences of patients in a mental institution. Both
were trying to capture the quality of people's behaviour: TEST
.
YOUR.SELF
. .
what the subjects understand, how they feel and, most
Why is qualitative data useful for capturing the meaning
importantly, why they behave in particular ways in different of people's behaviour?
situations.

ACTIVITY

Qualitative data: non-numeric data that expresses the Which of the following types of data do you think is
quality of a relationship. most suitable to collect in sociological research? Give
reasons for your answer.
l quantitative
Strengths
2 qualitative
The objective of qualitative behaviour is to understand
people's behaviour, so they must be allowed to talk 3 quantitative and qualitative.
and act freely. This allows the researcher to capture the Make a list of the strengths and limitations of each type
complex reasons for behaviour. Qualitative methods, of data.
such as participant observation involve the researcher
 Cambridge International AS and A Level Sociology

Quantitative and qualitative

methods and sources of data Identify two differences between open and closed
questions.
Primary quantitative methods
Questionnaires
Questionnaires consist of written questions that take one Strengths: Pre-coded questions make it easier to quantify
of two forms: data, because the options are already known, they are
limited in number and easy to count. Such questions are
Postal questionnaires are normally completed also quick and easy to code and interpret; in some cases
in private; respondents write their answers without this is just a simple count of the number of responses.
the presence of, or guidance from, the researcher. Pre-coded questions are useful when the researcher
Researcher-administered questionnaires are completed needs to contact large numbers of people quickly and
in the presence of the researcher, with respondents efficiently. The respondents do the time-consuming work
answering questions verbally. of completing the questionnaire.
Questionnaires can result in highly reliable data;
because everyone answers the same questions, it it easy
to replicate the research. The fact that respondents often
Questionnaire: research method consisting of a list of remain anonymous means that the validity of the research
written questions. Closed-ended questions provide possib le is improved, especially when it involves questions about
answers from whic h the respondent can choose, wh ile
potentially embarrassing or criminal behaviour. In
open-ended questions mean the respondent may answer in
addition, without face-to-face interaction, there is less
their own words.
risk that the respondent will give biased answers or try to
anticipate what the researcher wants to hear.

I Questionnaires involve two basic types of question.

Closed-ended or pre-coded questions involve the
researcher providing a set of answers from which
the respondent can choose. The researcher limits
the responses that can be given, as in the following
example:

There are variations on this type of question, such as

those that measure respondent attitudes, but their
defining characteristic is that they do not allow the
respondent to develop an answer beyond the listed
categories.
In open-ended questions, the researcher does not
provide possible answers. Rather, the respondent answers
in their own words. For example:
'What do you like about studying sociology?'
This type of question finds out more about the
respondent's opinions and produces a limited form of
qualitative data - although the main objective of
a questionnaire is usually to quantify responses. Some
questionnaires contain a mix of open and closed
questions. Are questionnaires reliable, unreliable, don't know?
 Chapter 3: Methods of research

Limitations: One significant practical problem with ('If you were the Prime Minister how would you run the
questionnaires is a low response rate, where only a small country?') - and imaginary questions run the risk of
proportion of those receiving a questionnaire return it. producing imaginary answers.
This can result in a carefully designed sample becoming
unrepresentative, because it effectively selects itself. There
is also nothing the researcher can do if respondents ignore
questions or respond incorrectly, such as choosing two Suggest one reason why biased questions lower the
answers when only one was requested. validity of data collected using a questionnaire.
The questionnaire format makes it difficult to examine
complex issues and opinions. In addition, the lack of
detailed information means that potentially significant data
Structured inte rviews
is not collected. These factors can limit the validity of the A structured interview is where the researcher asks
research. Another weakness is the fact that the researcher questions to respondents in person. To achieve consistent
has to decide at the start of the study what is and is not and comparable results, the same questions are asked in
significant. There is no opportunity to amend this later on. the same order each time.
The researcher has no way of knowing whether a
respondent has understood a question properly. The
researcher also has to trust that the questions mean the
Structured interview: set of stand ard questions asked
same thing to all respondents. While anonymity may
by the researcher of the respondent. It is sim il ar to a
encourage honesty, if someone other than the intended
questionna ire, but is delivered by the researcher rather than
respondent completes the questionnaire, it will affect the comp leted by a respondent.
validity and representativeness of the research. Some of
these problems can be avoided by pilot studies (see below),
but they cannot be totally eliminated. Strengths: One strength of the structured interview
A further problem involves (unintentionally) biased format is that potential reliability problems, such
questions. These can take a number of forms: as respondents misunderstanding or not answering
questions, can be resolved by the researcher. In addition,
■ If a question has more than one meaning (ambiguity),
it avoids the problem of unrepresentative samples - the
people will be answering different questions. 'Do you
response rates will be 100%.
agree most people believe the Prime Minister is doing a
Limitations: Structured interviews involve
good job?', for example, is actually two questions; you
prejudgements about people's behaviour and, like
could agree or disagree that the Prime Minister is doing a
questionnaires, can also contain unintentionally biased
good job, but you could also agree or disagree with 'most
questions. The lack of anonymity in an interview also
people's belief'.
contributes to two related limitations:
■ Leading questions suggest a required answer; by
saying 'most people believe', for example, the question 1 The interview effect occurs when a respondent tries to
challenges the respondent to go against the majority. 'help' the researcher by providing answers designed
When giving respondents a range of answers, they must to please. This reduces validity because respondents
be weighted equally to avoid leading answers. The simply provide answers they think the researcher wants.
following possible answers to the question 'How do you This can be caused by a 'halo effect', a situation Draper
rate Sociology as a subject?', for example, are too heavily (2006) describes as occurring when the novelty of being
weighted in favour of a positive answer: interviewed and a desire to reward the interviewer
for giving the respondent the chance to experience
it, results in unintentionally dishonest answers.
Brilliant! Incredible! Fantastic! Marvellous! Not bad
Conversely, prestige bias occurs when a respondent
gives an answer designed to not make themselves
■ If an option is not precisely defined, it will mean different look bod. Opinion polls, for example, sometimes show
things to different people. For example, people may respondents saying they would willingly pay more taxes
define the word 'occasionally' in different ways. if it helped to improve hospitals or care of the elderly, but
Hypothetical questions ask respondents to imagine in reality they vote for political parties that promise to
themselves in a position they do not actually hold reduce taxes.
 Cambridge International AS and A Level Sociology

2 The researcher effect refers to how the relationship identified and analysed the stereotypical roles played by
between researcher and respondent may bias responses: female characters in soap operas. Harwood (1997) used
content analysis to demonstrate that television viewers
Aggressive interviewers, for example, may introduce
generally prefer to watch characters of their own age.
bias by intimidating a respondent into giving answers
Strengths: Content analysis can identify underlying
they do not really believe.
themes and patterns of behaviour that may not be
Status considerations, based on factors such as
immediately apparent. Recurrent themes, such as female
gender, age, class and ethnicity, may also bias the
associations with housework, in complex forms of social
data . Afemale respondent may feel embarrassed
interaction can also be identified. Hogenraad (2003) used
about answering questions about her sexuality posed
computer-based analysis to search historical accounts of
by a male researcher.
war in order to identify recurring themes and words in
the lead-up to conflicts. This suggests that quantitative
analysis can have predictive qualities. By identifying a
Researcher effect: also called the interviewer effect, this pattern of past behaviour that always leads to war, it would
refers to how the relationship between researcher and be possible to predict future conflict. Similarly, Kosinski
respondent may bias responses and lead to inva lid data . et al. (2013) used content analysis of Facebook to show how
a user's personal characteristics, such as their intelligence
quotient (IQ), sexuality and political views, could be
inferred from the things they 'liked'.

Suggest one similarity and one difference between a

structured interview and a questionnaire.

I Conte nt analysis
Content analysis has both quantitative and qualitative
forms. What both types have in common is the study of
tex ts (data sources such as television, written documents,
etc.). Quantitative analysis of media texts, for example,
uses statistical techniques to categorise and count the
frequency of people's behaviour using a content analysis
grid (Table 3.1).

Content analysis can be used to reveal hidden social processes,

Content analysis: research method used for the
such as how websites collect private information
systematic analysis of media texts and commun ications.
Content analysis can also be used for 'concept mapping'.
Page (2005) tracked how media professionals portrayed
Although the grid below is a simple example, content global warming in order to show how far global warming
analysis can be complex and wide-ranging. Meehan's was reported in terms of 'natural' or 'social' causes. The
(1983) study of US daytime television, for example, quantification of such behaviour allows researchers to
draw complex conclusions from quite simple
data-collection techniques. The use of a
Character Gender Age Place and purpose On screen standardised framework (the grid) also means
Azir Khan Male 25 Office (employee) 30 seconds that data can be checked and replicated.
Limitations: In some types of content
Safiq Dhonna Male 56 Office (customer) 43 seconds
analysis reliability may be limited because
Angelique Female 37 Shop (customer) 84 seconds researchers must make subjective judgements
Basson
about behaviour. Not only do they have to
Table 3.1 Simple content analysis grid to record the behaviour of characters in decide which categories will and will not
a television programme be used, they may also have to judge which
 Chapter 3: Methods of research

forms of behaviour fit which categories. This raises or coincidence. However, there are two ways to separate
questions about whether all observed behaviour can be correlation from causality:
neatly categorised, In this instance, data can be difficult to
1 Test and retest a relationship. The more times a test is
replicate because different researchers studying the same
replicated with the same result, the greater the chances
behaviour may not categorise it in the same way.
that the relationship is causal.
Content analysis does not tell us very much about how
or why audiences receive, understand, accept or ignore 2 Use different groups with exactly the same characteristics:
themes and patterns discovered by the research. This is an experimental group whose behaviour is
why content analysis is often used in combination with a manipulated
qualitative method such as semiology (see below). 11 a control group whose behaviour is not.

Experiments
Experiments involve testing the relationship between
different variables - things that can change under Experimental group: the subjects of an expe riment. The
controlled conditions. The researcher changes researcher changes different variables to test their effect on
(manipulates) independent variables to see if they behaviou r.
produce a change in dependent variables that are not Control group: in an experiment, the characteristics of
changed by the researcher; any changes must be caused the control group exactly match those of the experimenta l
by a change in the independent variable. group. While the be haviou r of the expe rimental group is
manipu lated in some way, no attempt is made to sim il arly
manipu late the control group. This allows compa risons to
be made between the contro l and experimental groups.
Variables: factors that can be changed (manipulated) by
the researcher to understand their effect on behaviour.
Laboratory experiments
Bandura et al.'s (1963) 'Bobo doll' experiment, designed
Experiments, therefore, are based on changing an to measure the relationship between media violence and
independent variable and measuring any subsequent violent behaviour in young people, used four groups:
change in a dependent variable. This relationship that can
■ Three experimental groups were shown a film depicting
be one of two types:
different types of violence
1 Correlations occur when two or more things happen at A control group was not shown violent behaviour.
roughly the same time. These only suggest a relationship,
however, because it is possible for them to occur by Each group was observed to see whether those shown
chance. For example, waking up in bed fully clothed violent behaviour then played violently with a specially
may correlate with feeling unwell - but that does not designed inflatable doll. The experimental groups
mean the former causes the latter. A third factor, such as demonstrated violent behaviour, but the control group
drinking a lot of alcohol the previous evening, might be a did not. This suggested at worst a correlation and, at best,
cause of both. a causal relationship between seeing violence and acting
2 Causation involves the idea that when one action violently. A control group can be used to check that changes
occurs, another always follows. Causal relationships are in the experimental groups' behaviour were not the result
powerful because they allow a researcher to predict the of chance. Without a control group, Bandura et al. could
future behaviour of something. not have been sure that the violent behaviour of the groups
shown violent films was not simply their normal behaviour.
This research is an example of a laboratory experiment,
one that takes place in a closed environment where conditions
Causation: the idea that when one action occurs, another
always follows because the latter is caused by the former.

Laboratory experiment: experiment that takes place in

It is not always easy to distinguish between correlation a closed environment where co nditio ns can be precise ly
and causation in the real world of sociological research, monitored and controlled.
because things often happen at the same time by chance
 Cambridge International AS and A Level Sociology

can be precisely monitored and controlled. This ensures that

no 'outside' or uncontrolled variables affect the relationship
between the dependent and independent variables.

Field/natural experiments
Laboratory experiments are rare in sociology, because they
raise ethical issues and questions about validity. Therefore,
a more common type is the field experiment, which is
conducted outside of a closed, controlled environment.

Field experiments: experiments that take place in the 'real

world', beyond the closed, controlled environment of the
laboratory.

Briefly explain the difference between a laboratory

and a field experiment. Suggest one strength and one
limitation of each type of experiment.

I It is very difficult to control all possible independent

variables in a natural setting, which means that natural
experiments tend to establish correlations rather
A specially designed inflatable doll can be used in laboratory
experiments to test triggers for violent behaviour in young people

than causation. However, the basic principles of the instructions to behave as if their parents (the dependent
experiments are the same. Researchers use dependent and variable) were strangers and to observe and record how
independent variables to test a hypothesis or answer a the parents' behaviour changed towards their 'oddly
research question: behaving' offspring.

■ To test the hypothesis that teachers' expectations Strengths: Laboratory experiments are easier to replicate
influence how well their pupils do in school, Rosenthal than field experiments because the researcher has
and Jacobson (1968) conducted a study of low more control over both the research conditions and the
educational achievement in Mexican children. The variables being tested. Standardised research conditions
dependent variable was their level of achievement give experiments a high level of reliability. Experiments
and the independent variable was the expectations can also create powerful, highly valid statements about
teachers had about the ability of their pupils. Rosenthal behaviour based on cause-and-effect relationships that can
and Jacobson manipulated the independent variable be extended from the lab to understand people's behaviour
by pretending to be psychologists who could, on the in the real world. Similarly, field experiments can be used
basis of a sophisticated IQ test, identify children who to manipulate situations in the real world to understand the
would display 'dramatic intellectual growth'. In fact, underlying reasons for everyday behaviour.
they tested the pupils and then randomly classed some Limitations: It can be difficult to control all possible
students as 'later developers'. The researchers informed influences on behaviour, even in a laboratory setting.
the teachers of their 'findings'. They retested the pupils A simple awareness of being studied, for example, may
at a later date and discovered that the IQ scores of those introduce an uncontrolled independent variable into an
pupils whose teachers believed were 'late-developing experiment. The Hawthorne (or observer) effect, named
high flyers' had significantly improved. after a study by Mayo (1933) at the Hawthorne factory in
■ Garfinkel's (1967) breaching experiments showed how Chicago, refers to changes in people's behaviour directly
people 'construct reality' through everyday routines and resulting from their knowledge of being studied. The
assumptions. In one experiment, student researchers working conditions at the factory were manipulated in
(the independent variable) were sent home with different ways, such as changing the brightness of the
 Chapter 3: Methods of research

lighting and the temperature in the factory. However, contact with the study group throughout the period of
the results were always the same: the productivity of the the research.
workers increased. As Draper argues, 'the important effect
here was the feeling of being studied'.

Longitudinal survey: a form of comparative analysis that

invo lves tracking changes among a representative samp le
over time.
Hawthorne (or observer) effect: changes in people's
behaviour that result from their knowledge of being
observed. Strengths: Kruger argues that one strength of
longitudinal surveys is that they can be used 'to
summarize vast sources of information and facilitate
The ecological validity of experiments can also be comparisons across categories and over time'. This is
questioned because they take place in an artificial because they exploit the ability of quantitative methods
environment. A controlled experiment is an unusual to identify and track personal and social changes. Hills
situation and respondents may behave differently in a et al. (2010), for example, used data from the English
laboratory or field setting. One solution to this problem Longitudinal Survey of Ageing to analyse the relationship
is to conduct the experiment secretly, as in the case of between mortality rates and levels of wealth. The study
Milgram (1974) or Rosenhan (1973), but this raises ethical found a strong correlation between low wealth and
questions about the right to experiment on people who premature death. This demonstrates a significant strength
may be unwitting (and unwilling) participants. oflongitudinal studies: the ability to reveal trends that
would otherwise remain hidden.
A further advantage of longitudinal surveys is that they
can generate reliable representative samples to suggest
causal relationships. Power et al.'s (20ll) 10-year study
of 200 families raising children in highly disadvantaged
neighbourhoods, for example, found a 'clear cause and
effect between physical or environmental improvements to
an area and the well-being of its families'.
Limitations: Sample attrition, or the number of people
who withdraw from the original sample over time, is a
major limitation of these surveys. High levels of attrition
can reduce the representativeness of the sample over
time - a problem that grows the longer the study lasts.
While longitudinal studies can identify trends or allow
How might the knowledge of being watched change people's researchers to make correlations and causal connections
behaviour? between phenomena, such as income and life expectancy, they
are only ever a glimpse of behaviour at any given moment.
longitudinal surveys They can, therefore, be criticised for lacking depth and validity.
These are a form of comparative analysis that involves
tracking changes among a representative sample over
time, from a few months to many years. The same group
is analysed at different stages in their lives, using methods Suggest one reason why a researcher might want to
ranging from questionnaires to non-participant observation. study the same group at different times.
Like participant observation studies (see below),
longitudinal surveys are carried out over a significant Cross-sectional surveys
period of time; they can last many years. However, in
This type of survey is explicitly designed to produce a
longitudinal surveys the researcher remains detached
'snapshot' of behaviour at any given time:
from the study group, having contact with the research
subjects only on a limited basis at set intervals. By 11 Qualitative forms are generally descriptive, with the
contrast, with participant observation studies the objective being to illustrate a particular type of behaviour.
researcher usually maintains more or less complete It may involve, for example, looking at a certain
 Cambridge International AS and A Level Sociology

population characteristic, such as suicide, income or changed. For example, education statistics can track
poverty, applied to a single country, a large area within a changes in levels of achievement. Statistics can also be
country or a specific feature of different countries. used to track changes in behaviour, such as before and
a Quantitative forms, the most common type, are analytic: after the introduction of a new law.
the objective is to analyse both correlations and
causations between different phenomena. Durkheim's
study of suicide, for example, used cross-sectional
surveys taken from different societies to build up a Comparative analysis: a co mparison of different cu ltures,
cases and situatio ns to understa nd their sim ilarities and
comparative analysis of variable suicide rates. He
differe nces.
used these as the basis for a theoretical explanation of
different types of suicide.

Statistics can be used for comparisons within groups,

such as differences in middle- and working-class family
Cross-sectional survey: research method focused on size, and between societies. Bakewell (1999) suggests
identifying groups that share broad similarities, such as that official refugee statistics are useful because they
level of education, and measuring differences in a single quickly and easily demonstrate the size and scale of an
variable; whether, for example, people with a high level of international social problem.
education have higher rates of suicide than those with a Strengths: In practical terms, official statistics may
lower level of education.
be the only available source covering a particular area
of study, such as suicide. In addition, data that would
Both types of cross -sectional survey normally require be costly, time-consuming and difficult to collect, such
representative samples because one of the main objectives as statistics on marriage, divorce or crime, is readily
is to make generalisations about behaviour. In this respect, available - especially since the development of the

I cross-sectional surveys tend to focus on identifying groups

that share broad similarities, such as income, education
internet. Another strength of official statistics is their
representativeness. As Marshall (1998) notes, statistical
and gender. They measure differences using a single 'data are almost invariably nationally representative,
variable, such as death or suicide rates. By comparing because they are obtained from complete censuses or very
standardised groups, it is possible to explain differences large-scale national sample surveys'.
in death or suicide rates using variations in standardised More theoretically, many official statistical sources,
variables - whether, for example, people with a high level in areas including crime, unemployment, marriages,
of education have higher rates of suicide than those with a births and divorces, are recorded by law. Data is usually
lower level of education. collected in the same way from the same sources
(iteration). This adds to its reliability because research
Secondary quantitative methods can be replicated and compared. Although definitions,
Official statistics of areas such as 'unemployment' may change over time,
Official statistics created and published by governments most - such as 'birth' or 'murder' - remain the same.
are a major source of secondary quantitative data used by Some statistical data has low validity, but this is not true
sociologists to examine trends and patterns within and of all official statistics. For example, data about marriage,
between societies: divorce, birth and death can record these events with a
high degree of accuracy.
Patterns of behaviour may be picked up by statistical Sociologists use the term 'hard statistics' to refer
analysis because they provide a broad overview of to quantitative data that demonstrates such accuracy.
behaviour across potentially wide areas: local, national For example, statistics about the number of divorces
and international. Durkheim (1897), for example, in a society can be viewed as 'hard' evidence. This is
identified distinct patterns to suicidal behaviour based because a divorce has to be legally registered and so
on a comparative analysis of official suicide statistics clear and accurate records are available. Statistics that
across a range of different societies. are considered to be less accurate are referred to as 'soft
■ In terms of trends, statistical data drawn from different statistics'. Official statistics about the unemployment
years can be used to understand how something has rate may be 'soft' in, because there are different ways of
 Chapter 3: Methods of research

defining 'unemployment'; depending on which definition

is used, the figures may vary greatly.
Limitations: Apart from not providing any great depth
or detail, official statistics involve of problems of validity
due to what governments include in or exclude from
published data. Such data may only give a partial picture of
reality for two reasons:

1 While official crime statistics provide valuable data

about crimes reported to the police, they tell us little
or nothing about the 'dark figure of crime' - crimes that
are not reported or recorded. Research has shown that
in many societies as much as 75% of all crime 'is in the
dark figure'. Does having to legally record statistical information make it more
2 Statistical data does not reveal much about the reliable?
reasons for people's behaviour. For example,
although we have a reasonably precise figure for the
Primary qualitative methods
number of year-on-year murders in countries around
the world, this data tells us little about why people kill Se mi-structu red (' fo cused ') inte rviews
each other. Nichols (1991) defines this method as 'an informal
interview, not structured by a standard list of questions.
Although quantitative data is normally considered more Researchers are free to deal with the topics of interest in
objective than qualitative data, its significance must always any order and to phrase their questions as they think best.'
be interpreted by researchers; they have to decide what the Semi-structured interviews, therefore, allow a respondent
data means. A statistical rise in crime, for example, may be to talk at length and in depth about a particular subject. The
the result of: focus or topic of the interview is decided by the researcher.
The interview has a structure or 'interview schedule' - the
a real rise
areas the interviewer wants to focus on - but there is no
■ a different way of defining and counting crime
list of specific questions. Different respondents may be
11 police targeting certain types of crime and arresting asked different questions on the same topic, depending on
more people. how the interview develops. The objective is to understand
Governments occasionally change the definition of key things from the respondent's viewpoint, rather than make
concepts. Different governments may also define a concept generalisations about behaviour.
differently, as Bakewell discovered in relation to official
definitions of'a refugee' around the world. Such changes
and differences bring into question the reliability of the
Semi-structured interview: research method in whic h
data, because when making statistical comparisons the
a respondent is encouraged to talk at length about a
researcher must ensure they are comparing like with like. particular subject. Also cal led focused interviews because
Two further problems arise over how behaviour is officially the topic is decided by the researcher and is the focus of
categorised: their questions.

To estimate the extent of 'knife crime', data is increasingly

taken from hospital records because victims do not Open-ended questions are frequently used in semi-
always report the incidents. While this official data gives structured interviews. Some of these are created before
us an idea of the general extent of knife wounding, it the interview, while others arise naturally from whatever
does not distinguish between deliberate and accidental the respondent wants to talk about. For example, if the
wounding, for example. interview focused on understanding family life, the
11 The categories used by governments to define things interviewer might begin with a question like: 'Tell me
such as social class or poverty are not necessarily the about your family.' If the respondent then mentions their
same as those used by sociologists. children the interviewer might decide to continue with
 Cambridge International AS and A Level Sociology

a question such as: 'Tell me about your relationship with knowing if someone is telling the truth, a further problem
your children.' is imperfect recall; it can not only be difficult to remember
Strengths: As there are no specific questions prepared, things that may have happened months or years ago, but
there is less risk of the researcher predetermining what memories can also be selective - respondents only recall
will be discussed. Where the respondent can talk about those things that seem important to them.
things that interest them it is possible to pick up ideas Finally, semi-structured interviews lack standardisation;
and information that may not have occurred to the the same questions are not necessarily put to all respondents
interviewer or of which they had no prior knowledge. and similar questions may be phrased differently. Can
This new knowledge can be used to inform subsequent reduce the reliability of the data and make it difficult to
interviews with different respondents and to suggest generalise the research.
further questions.
Unstructured interviews
By allowing respondents to develop their ideas, the
researcher tries to discover what someone really means, Unstructured interviews are built on a general idea
thinks or believes. The focus on issues that the respondent or topic that the researcher wants to understand.
considers important results in a much greater depth of Respondents are encouraged to talk freely about the
information. This may increase the validity of the data as things they feel are important. Kvale (1996) states that
it is more likely that the research will achieve its real aims. 'behaviour is understood from the perspective of those
Oatey (1999) suggests that 'freedom for the respondent being studied; their perceptions, attitudes and experiences
to answer how they wish is important in giving them are the focus'.
a feeling of control in the interview situation'. Within
limits, face-to-face interaction allows the researcher to
help and guide respondents. To explain, rephrase or Unstructured interviews: free-form interview method
clarify a question or answer, for example, may improve where the objective is to get the respondent to talk, without

I overall validity. prompting or interrupti on, about whatever they fee l is

important about a topic.

Strengths: The researcher's limited input means that

Suggest two differences between structured and semi-
structured interviews. data reflects the interests of the respondent. It is therefore
more likely to be an accurate and detailed expression of
their beliefs. Hamid et al. (2010) used this method in their
Limitations: This method demands certain skills study of young Pakistani females because 'unstructured
in the researcher, such as asking the right questions, interviews helped elaborate on the topics of participants'
establishing a good rapport and thinking quickly about choice [marriage and sexuality] and probed further
relevant question opportunities. It also requires skill their concerns'. This technique avoids the problem of
from the respondent; an inarticulate respondent will the researcher prejudging what constitutes important or
probably be unable to talk openly and in detail about irrelevant data.
the research topic. Oatey also argues that open-ended The researcher must establish a strong rapport with
questions 'can cause confusion either because of the respondents. If this rapport is achieved, people who may
lack of understanding of the question or by the lack of be naturally wary of being studied can open up to the
understanding of the respondent's answer'. researcher, allowing sensitive issues to be explored in
Semi-structured interviews are not only more time- depth. Hamid et al. established a relationship with their
consuming than questionnaires but the large amounts respondents by meeting them a number of times before
of information they produce must also be analysed their research started. This 'helped the participants to
and interpreted. This data is rarely tightly focused on a open up ... and discuss sensitive issues regarding sexuality
particular topic, so a researcher may spend a lot of time and growing up with reference to their marriage and other
analysing data that has little or no use to the study. related topics of their choice'. If the research is relatively
A theoretical problem is the idea that all interviews are informal it can take place somewhere the respondent
reconstructions. Respondents must remember and recount will feel at ease, such as in their own home. For Hamid
past events, and this creates problems for both researcher et al. this helped 'overcome the barrier of talking about
and respondent. While a researcher has no way of sensitive issues'.
 Chapter 3: Methods of research

- :J for example - or they may simply represent a group the

i researcher wants to explore in detail. Nichols suggests
that focus groups may also be same-sex and from similar
backgrounds to prevent gender and class variables affecting
the reliability and validity of the data. For Morgan (1997),
the effectiveness of group interviewing is based on:

■ an interview structure with clear guidelines for the

participants, to avoid arguments within the group
■ predetermined questions through which the experiences
of participants can be explored
111 interaction within the group, which Gibbs (1997)
argues gives 'unique insights into people's shared
How are unstructured interviews like a conversation? understandings of everyday life'.

Strengths: In group interviews, the researcher can help

Limitations: Unstructured interviews demand skills of both
the discussion. They can:
the researcher and the respondent. The researcher must
resist the temptation to influence, encourage or interrupt. El control the pace and scope of the discussion
Respondents must be able to express themselves clearly and ■ plan a schedule that allows them to focus and refocus the
understandably. The researcher, by design, has little control discussion
over the direction of the interview and the conversation may
■ intervene to ask questions, stop or redirect aimless
stray into areas that later prove irrelevant to the research.
discussions
The interviews are time-consuming in themselves, but so
■ create a situation that reflects how people naturally share
is analysing and interpreting all the data they generate.
Reliability also tends to be low because the non-standardised
format makes the interview impossible to replicate.
and discuss ideas.

Gibbs argues that one strength of this method is the

I
A more severe limitation is the argument that all forms ability to 'draw on respondents' attitudes, feelings, beliefs,
of interviewing are inherently biased by interview effects. experiences and reactions'. In a group, respondents are
Cohen and Taylor (1977) argue that the most significant of encouraged to elaborate and reflect on their beliefs. Pain
these effects is that through the act of questioning people, a et al:s (2000) study of the fear of crime used this format to:
series of status manipulations come into play. The outcome
of this is that respondents try to please the researcher by ■ generate large amounts of detailed information quickly
telling them what they believe they want to hear. and efficiently
uncover attitudes, beliefs and ideas that would not have
t :-resfvou R·sEi.F•"
:c:· , •,~ - - - . ' •" ', - -,
been revealed by less flexible methods.

Suggest two advantages to understanding behaviour Kitzinger (1995) suggests that group interaction 'enables
from the respondent's point of view. participants to ask questions of each other, as well as to
re-evaluate and reconsider their own understanding of
their specific experiences'. Respondents can also 'explore
Group interviews solutions to a particular problem as a group rather than as
Group interviews (also called focus groups) involve individuals'. Morley (1980) felt that individual interviews
respondents gathering to discuss a topic decided in were 'abnormal situations' for most people, but that group
advance by the researcher. These groups may be selected interviews created an environment that encouraged people
as representative samples - a cross-section of society, to talk at length.
Limitations: The researcher must control the behaviour
of the group to allow people to speak freely and openly
Group interviews: also ca lled focus groups, th ese involve
about an issue while maintaining the focus of the research.
respon dents discussing a top ic as a group rather tha n This, Gibbs notes, means 'good levels of group leadership
individually. and interpersonal skill are required'. The more people there
are to co-ordinate, the more likely it is that there will be
 Cambridge International AS and A Level Sociology

problems with resources, such as time, money and effort. literally - when the research subject does not know they
There may also be problems with representativeness: if in a are being observed. However, non-participant observation
carefully selected group of ten, one person does not show usually means that the researcher does not become
up, the sample becomes unrepresentative. personally involved in the behaviour they are studying.
Morgan argues that any failure to stop individuals and In this way, the researcher's presence cannot influence the
alliances taking control and 'setting their own agenda for behaviour of those being watched. As Parke and Griffiths
discussion' raises questions about reliability. Researchers (2002) noted in their study of gambling: 'Non-participant
have less control over the data because they have less observation usually relies on the researcher being unknown
control over the group and the direction of the discussion. to the group under study ... [they] can study a situation in
This may also affect validity because the focus of the its natural setting without altering that setting.'
interview may end up moving away from the researcher's Strengths: Access is one practical advantage of non-
intended focus. Group interviews are also at risk from participant observation. It allows research on people who
another type of'interview effect' - one that Janis (1982) may not want to be studied because their behaviour is
calls 'Groupthink'. This refers to the pressure people feel illegal, secret or personally embarrassing, for example.
to arrive at 'desired outcomes', such as saying what they People may also be suspicious of a researcher or, as in the
believe the researcher wants to hear. Group interviews also case of the gamblers Parke and Griffiths studied, 'dishonest
run the risk of simply reflecting a 'group consensus' rather about the extent of their gambling activities'. Since the
than revealing what individuals really believe. researchers could not simply ask gamblers about their
behaviour, one solution was to observe them 'at work'.
u When the researcher does not participate in the

I
behaviour being observe, respondents can be objectively
studied in a natural setting. The researcher gets to see
'everyday behaviour' just as it would normally occur - a

I technique Yule (1986) used to discover how mothers really

treated their children in public places.

I,
What practical advantages do group interviews have over
individual interviews?
How does this picture illustrate
Observation the idea that we can't always
trust the evidence of our
Observational methods are based on the idea that data is
own eyes?
more valid if it is gathered by seeing how people behave,
rather than taking on trust that people do what they say
Limitations: Observational studies cannot be easily
they do. There are two main observational techniques:
or exactly replicated because the characteristics and
non-participant and participant.
composition of a group may change over time. Observing
Non -participant obse rvation people 'from a distance' may also produce data that fails
Non-participant observation involves observing to capture the depth, richness and intimate details of their
behaviour from a distance. This sometimes takes place behaviour. This type of study also raises ethical questions,
because people are being observed without their permission.
Parke and Griffiths argue that this method requires certain
practical skills, such as 'the art of being inconspicuous'.
Non-participant observation: when the researcher They found that 'if the researcher fails to blend in, then slot
observes behaviour without participating in that behaviour. machine gamblers soon realise they are being watched and
are highly likely to change their behaviour'.
 Chapter 3: Methods of research

the gang and eventually gained access to some of its

more powerful members. This situation echoed Whyte's
What advantages are there to watching rather than (1943) entrance into an Italian street gang through the
questioning people? sponsorship of its leader, 'Doc'.
Sponsorship makes it easier to separate the roles of
participant and observer. It reduces the chance of the
Participant observat ion
researcher becoming so involved in a group that they
Participant observation is when the researcher takes part
stop observing and simply become a participant (going
in the behaviour being studied. It is based, in part, on
native). Even so, Venkatesh found there were times when
what Weber (1922) termed verstehen - 'to understand
his involvement was so complete that he acted 'like one
by experiencing' or, as Mead (1934) described it, the
of them' and effectively ceased to be an objective and
researcher's ability to take the part of the other and see things
impartial observer.
from their viewpoint (empathy) . Participant observation
The ability to ask questions, observe individual
reflects what Downes and Rock (2003) call 'the claim social
behaviours and experience the day-to-day life of respondents
behaviour cannot be understood unless it is personally
helps the researcher to build up a highly detailed picture of
experienced' or, as Parker (1974) puts it, 'by cornering people
the lives they are describing. This '360-degree' view means
in classrooms to answer questionnaires the sociologist misses
the researcher not only gets to understand what people 'say
meeting them as people in their normal society'. Participant
they do', but also witnesses and experiences what people
observation can take two forms: overt and covert.
actually do. This can increase the validity of the data.
Where the observer's role is clearly defined there is also
less risk of involvement in unethical, criminal, dangerous
Verstehen: research strategy central to participant or destructive behaviours. The researcher can, for example,
observation, which tak~dvantage of the resea.Lcher~s- __ withdraw from risky situations without necessarily losing
abil ity to see things from the subject's viewpo int (empathy). the trust or arousing the suspicions of those being studied.
I
Overt observation involves participating in the behaviour
of people who know they are being studied. The researcher
joins the group openly, and usually conducts the research
with the permission and co-operation of the group (or
significant members of the group).

Overt observation: participant observation in wh ich those

being studi ed are aware they are being researched .

Strengths: On a practical level, recording data is relatively

easy because the group knows and understands the role What advantages are there to experiencing behaviour rather than
of the researcher. The researcher can ask questions, take simply observing it?
notes and observe behaviour openly. With groups that
have hierarchical structures, such as large businesses, the Limitations: If a group refuses the researcher permission
researcher can gain access to all levels - the boardroom as to observe it, then the research cannot be carried out. In
well as the shop floor. addition, overt observation requires substantial amounts
It can be difficult to gain access to some groups, of time, effort and money. Venkatesh, for example, spent
so researchers may use sponsorship to find a way in. around eight years on his study of a single gang in a small
This involves gaining the trust and co-operation of an area of one US city.
important group member. Venkatesh's (2009) study of Theoretical criticisms initially focus on the observer/
a black American gang, for example, was only possible Hawthorne effect and the extent to which knowledge
because a middle-ranking gang member called 'JT' of being watched changes how people behave. While
'sponsored' and protected Venkatesh while he observed people may be studied in their natural environment, an
 Cambridge International AS and A Level Sociology

awareness of the presence of the researcher may make Defensive groups: Ray (1987) covertly studied Australian
them behave unnaturally. Venkatesh witnessed a staged environmental groups who would have been suspicious
punishment beating designed to demonstrate the limits of his motives if he had tried to study them openly.
of his observational role; higher-level gang members were
'putting on a show' for his benefit.
A further limitation is the researcher's level of
involvement: Covert observation: th ose being studied are unawa re th ey
are pa rt of a research project.
11 Without full participation, researcher involvement may be
too superficial to allow a true understanding of behaviour.
Ethical concerns, such as not participating in illegal Full participation in a group, where the researcher
behaviour, may affect the extent to which the researcher lives, works and socialises with the people they are
is truly experiencing how people normally behave. studying, means this method produces highly detailed,
■ There is a risk that the researcher will become too insightful, personally observed and experienced, data.
involved and effectively 'become the story' they are Covert participation also avoids the observer effect - the
reporting. Their presence becomes the focal point around subjects' behaviour is largely unaffected by the researcher's
which people orientate their behaviour. Venkatesh was presence.
given 'special treatment'; he was invited to meetings Through personal experience, the researcher gains
and was introduced to people he would not have met if valuable insights into the meanings, motivations and
he had not been known as a researcher sponsored by a relationships within a group. These can explain why people
powerful gang member. behave in certain ways. The ability to experience things
from the point of view of those involved, coupled with
Overt participant observation is impossible to replicate; the sociological insights a researcher brings to the role of
others must trust that the researcher saw and experienced

I the behaviour they document. In addition, it can be

difficult for a researcher to accurately record behaviour
while they are in the middle of it. No researcher can record
observer, means they can make sense of behaviour even in
situations where group members may not fully understand
the reasons for that behaviour. As Parke and Griffiths note,
it is possible to overestimate 'the subjects' knowledge and
and document everything that happens, which means this
understanding of their own behaviour' and their ability
method will always involve the selection, interpretation to explain why they do something in an interview or
and reconstruction of ideas and events.
questionnaire.
In covert observation, the researcher secretly (covertly) Limitations: Goffman's study of a US mental
joins the group, so the subjects are unaware that they
institution identified three major problems for the covert
are being studied. The main objective is to experience participant observer:
behaviour in its 'natural setting'; to watch people behave
as they normally behave. Unlike overt participation, 1 Getting in: while gaining covert entry to any group can
the researcher must balance the roles of researcher and be a problem, some groups are more difficult to enter
participant without revealing their true role to other than others:
group members.
Entry to some groups is by invitation only. Unless the
Strengths: Covert observation may be the only way to
researcher is invited, they cannot join.
study people who would not normally allow themselves to
111 Some groups have entry requirements. To covertly
be researched. Such people may include:
study accountants or doctors, the researcher
Criminal or deviant groups: Ward (2008) 'was a member would need to hold the qualifications these
of the rave dance drugs culture' when she began her professions require.
five-year study 'in London nightclubs, dance parties, bars, ■ The characteristics of the observer must match
pubs and people's houses'. Her knowledge of the 'dance those of the observed. A man, for example, could
scene', coupled with her friendship with those involved not covertly participate in a group of nuns. (There
meant she was able to gain easy access to this world. are, however, ways around this problem . Goffman,
■ Closed groups: Lofland and Stark (1965) secretly studied while neither a doctor nor mentally ill, was able to
the behaviour of a religious sect because this was the covertly observe by taking a manual job within the
only way to gain access. institution.)
 Chapter 3: Methods of research

..~--~
~ ~, _.! ~ . . ... ~ ' ~· ,. ·. 11 At the other, the researcher may become so much a

. ""
' -y: . ··. ·~·:,

\
•
. ~. • • .

•... :.,. 6 . . -
.
~
.
••

~
b
~ .
,.7,
~ : .

.... ~
"
• • ~
-

- ,
. •
~- -, . , ;,,, ~ -
.

.....,~'
.
I/ ~
:,;
,

_._
' t
'
,
~ • •.
~
. .,,,_•·· .
~&

-
..
,;, • I
part of the group they go native and stop being an
observer, which can raise doubts about the validity of
{
' -sJ

•
,·. ; #.,,,, '.:;a,
. • '"'.:aj b
,f ' -~

O
•

;; I ''
• '
. ;~ .
•
.

,;;1.
, _;
;)f
\
the research. Whyte, for example, became so involved
with the lives of gang members he came to see himself

.i9., , ~ . .
. -~ -.,, C) ' ;,
as 'one of the gang' and not as a researcher.
~
~ ~

f
, { \ I!') ' @
3 Getting out: it can be difficult to stop participating.
, · t4 ' -~ , ' f ..
A member of a criminal gang, for example, cannot
simply leave. In other groups leaving may raise ethical
·~- questions, such as the effect of deserting people who
\w ~ -
'.':'>
have grown to trust and depend on the researcher.
,\. This type of research raises further ethical questions,
A police officer in
uniform is intended to stand out from the such as whether a researcher has the right to pretend
crowd. Why are the characteristics of the researcher and the to be one of the group or use its members for their
group significant in covert participant observation? own purposes.
In addition to problems of entrance, acceptance and
departure, further limitations include:
2 Staying in: once inside, the researcher may not have
access to all areas. For example, an observer posing 11 research cannot be replicated
as a school student could not freely enter places, ■ we have to trust that the researcher saw what they
such as staff rooms, reserved for teachers. Someone claim to have seen
being where they are not supposed to be would raise 111 recording data is frequently difficult; the researcher
suspicions, and the researcher runs the risk of being cannot take notes, ask too many questions or openly
discovered by 'gatekeepers' - those whose job it is to
restrict access.
record conversations
the impossibility, as Parke and Griffiths note, of
I
The researcher has to quickly learn the culture and studying 'everyone at all times and locations', where it
dynamics of a group if they are to participate fully. becomes 'a matter of personal choice as to what data
This can require a range of skills, including the ability are recorded, collected and observed', all things that
to mix easily with strangers, create and maintain affect reliability and validity.
a plausible and convincing 'back story' (past) and
to think quickly on their feet when questioned or '..' .. -
· TEST YOURSELF
challenged. For example, Parker had to make instant
decisions about whether or not to participate in Identify two differences between overt and covert
the criminal activities of the gang of youths he was participant observation.
secretly studying.
If a researcher lacks the 'insider knowledge' they
should have, they risk exposure. Parke and Griffiths, Case studies
for example, noted how 'street knowledge about slot This type of research studies the characteristics of a
machine gamblers and their environments', such as the particular group or 'case', such as Westwood's (1984)
terminology players use, machine features and gambling 12-month participant observation study of female workers
etiquette, is an essential requirement for covert in a 'Stitchco' factory. A case study is not really a research
participation in the gambling world. method, but rather a research technique, in which different
The ability to successfully blend into a group carries methods can be used to generate data. Such studies are
its own problems. It can be difficult to separate the roles usually based on qualitative methods.
of participant and observer, especially if the researcher is
well integrated into a group:
At one extreme, the researcher may have to choose Case study: in-depth, qua litative, study of a particular
between participation and observation, such as if group or 'case'.
a group participates in criminal activities.
 Cambridge International AS and A Level Sociology

~•~-i:• .. z-: "i! •, -,-;'!"' ~)~•,-~f.;_1"~~•,:T,j~'lt'••,.;~:;,~;:-~;~;..,i;CK",f•'•:•.-:~::-!;r ~ ~•:;~.::~•~•'-_,j-1;>i!, - J . < . • ,," _.- •

particular group, there is no need
~:rype - :* ~~'·:.··-
:,-i..,_~..Ji•1"-l .. ..._'J;;',.~,\
,:.,:·_;/'Overt or.covert ~r{it-.~~•.,1-'t'1t.i",'-.:c"':!tl,,.ll"JJ~~,f•2"1~~!!h.._.,...'l';"'l:,_~
P.art1c1pat1on •-;1 ·characterist1cs i.~,-•~
,-·.,
,'<''f~':,.\ll•~•~.._,-. ,i",1,,:f,«._:{_i;;..J,l,'Ci;l,-,:vUli:,iir,._..t,ff'_'; ':.,,,,._,,
for the research to be generalised.
Non-participant Overt/covert None/minimal Complete observer
Covert participant Covert Full Complete participant Semiology
Semiology is the study of the
Overt participant Overt Full Participant-as-observer
cultural meanings embedded
Table 3.2 Types of observation in media forms. Stokes (2003)
Source: Brewer (2000) suggests that semiology is frequently combined with
content analysis to produce a more rounded picture of
Strengths: The focus on a single group studied over time behaviour through its ability to explore and interpret the
provides great depth and detail of information that has 'hidden messages' within texts. The theoretical justification
greater validity than simple quantitative studies. In cases for this is the idea that texts always involve two levels of
where the sample is relatively small and self-contained, understanding:
such as a factory (Westwood), a school (Lacey, 1970) or
■ The denotative refers to what something is; a literal
even a restaurant (Marshall, 1986), large amounts of data
representation, such as picture of a bunch of roses.
can be collected in a relatively cost-effective way.
The connotative refers to what something means and this
Such studies help to uncover the meanings people
can vary between individuals, groups and cultures. In
give to everyday behaviour. They often make use of
some cultures, for example, a man giving a woman 12 red
participant observation, but other methods are also
roses might symbolise his love. Doing the same in Russia
used in case studies. The research objective is normally
symbolises death - even numbers of flowers are only
to develop narrative data that 'tells a story' about the
given to people at funerals.
people being studied, so qualitative methods are usually
most useful.

I A significant strength of case studies is that they

allow an in-depth understanding of how people see and
understand their particular world, using their own eyes,
Semiology: the study of cultural meanings embedded in
media forms, often used to explore and interpret 'hidden
words and experiences. People can 'speak for themselves' meanings' embedded within texts.
through the channel of sociological research. Small-scale
case studies can also be used as pilot studies to allow a
researcher to develop hypotheses, test data-collection These features of language and culture are significant
methods and identify potential problems in preparation for because it is possible to examine both behaviour and texts,
a larger study. such as television programmes and websites, on two levels:
Limitations: Case studies have a range of practical
what it claims to be (denotative level)
limitations that depend to some extent on their size
and scope. Large-scale, in-depth studies can take a lot 11 what it means to others (connotative level).
of time, effort and money. Regardless of their size, the For example, the Glasgow Media Group (1976) showed
intensive and detailed nature of case studies means that how semiological analysis can identify the assumptions
they make higher demands on the skills of researchers, that lie behind the presentation of television news. Its
who may spend months or years living and working with analysis of how industrial disputes were portrayed
their subjects, and respondents, who may be subject illustrated subtle (and not so subtle) forms of bias:
to extensive and detailed questioning and observation
throughout the study. ■ Employers were generally filmed in a relatively calm
It is difficult to generalise from case studies because environment, such as in an office, behind a desk, and
they tend to focus on small groups that may only be the reporter would ask respectful questions that the
representative of themselves. It might, however, be possible employer was allowed to answer without interruption.
to draw comparisons between similar groups. The validity ■ Employees, often simply identified as 'striking workers',
of such criticism depends on the researcher's aims. If the were often pictured outside and the questioning was
sample (or case) is the target population, and the aim is to more aggressive, with an emphasis on the employee
simply provide a detailed insight into the behaviour of a having to justify their actions.
 Chapter 3: Methods of research

Strengths: Rose (2007) argues that semiology

has a number of practical advantages: it
requires very few resources, is relatively cheap
to carry out and can be applied quickly and
efficiently to very large samples.
Semiology also provides useful tools
for analysing the meaning of media texts,
to demonstrate how the media constructs
realities and identities, from selling products
to selling ideas. Marxism, in particular, has
used semiology to reveal how the media in
capitalist societies has presented various
social conventions, such as social and
economic inequality, as 'natural', inevitable
and unchangeable. Various studies have used
semiology to show how 'hidden meanings' are
embedded in texts:
Cumberbatch (1990) found that TV adverts
used male and female identities in different
ways. Older men and younger women
What is the connotation (meaning) of these pictures to you? Think about the
were more likely to be used than other age
cultural meanings embedded in the images
groups. Older men featured heavily when
an advertiser wanted to convey authority,
especially when an advert featured technical expertise,
while young women were used to convey sexiness.
to identify and understand the underlying meaning
of texts or behaviours. A semiological analysis of
multiplayer video games, for example, would be
I
■ Best (1992) demonstrated how pre-school texts designed
difficult for a researcher with little or no experience of
to develop reading skills are populated by sexist
online gaming.
assumptions and stereotypes about males and females.
More fundamental limitations relate to the idea that we
Although it is an interpretive method, semiology can cannot simply assume that because a researcher uncovers
be grounded in empirical research and can be easily 'hidden meanings' in a text, a casual audience would
combined with different types of content analysis: do the same. Similarly, semiology is based on the belief
■ Conceptual analysis focuses on the concepts or themes
that media messages are layered in terms of denotations
embedded in texts - an extension of quantitative
and connotations - 'core ideas' are embedded in layers
content analysis. Philo and Berry (2004), for example,
of seemingly inconsequential ideas. The 'real message'
identified recurring themes in news reports of the Israeli-
of advertising, for example, is not about choosing one
Palestinian conflict, such as language differences when
product over another but about creating a consumption
referring to similar forms of behaviour: Palestinians were
culture in which people define themselves in terms of what
often described as 'terrorists', while Israeli settlers were
they buy.
called 'extremists' or 'vigilantes'.
However, this argument assumes that 'real' messages
can be found by peeling back the layers and that the
• Relational analysis examines how texts encourage the
researcher has a privileged position in this process
reader to see something in a particular way by relating
as someone who 'knows the real meaning' of media
one idea to something different. Hall (1980) calls this a
messages. Hebdidge, for example, states that some
'preferred reading' the way text is constructed through
youth sub-cultures wore Nazi swastikas in an 'ironic
language, pictures and illustrations 'tells' the audience
way', while Young argues that this is unsupported by
how to interpret the information presented.
evidence and that semiological analysis runs the risk of
Limitations: Rose suggests that a researcher needs a the researcher projecting their own interpretations and
thorough grounding in their subject matter if they are prejudices on the study.
 Cambridge International AS and A Level Sociology

Official Organisational •
;. Individual , ·
, • • - 1
·: • ~ •'
_: .. ~: ._~_:_..
- ' r .,. -;,l;:.L

Sources Governments Private companies and Personal documents

organisations
Historical and Official reports; court Newspapers (local/national); Letters; diaries (paper and video); oral
current types reports; academic film; magazines; books; church histories; personal websites; biographies;
studies; websites records; academic studies autobiographies; social networking sites;
photographs

Table 3.3 Types and sources of documentary evidence

Secondary qualitative methods

Documentary sources Personal documents: secondary source of data covering
There are many documentary sources available to areas such as persona l letters, diaries, oral (verbal) histories,
sociologists and it is difficult to classify them in websites, socia l networking sites and photographs.
any meaningful way. We can, however, think about
documentary evidence (Table 3.3) in terms of different
sources and types. Newspaper articles, for example, may tell us more about
Strengths: In terms of practical strengths, their writers and how they see social problems than they
documentary sources give the researcher access to data do about the topic of the article.
that would cost a lot of money, time and effort to collect
personally. Such sources can provide secondary data
in situations where it is not possible to collect primary
data (about things that happened in the distant past, Suggest one reason why official government websites

I for example). Historical documents can also be used

for comparative purposes; contrasting how people once
lived with how we live now is useful for tracking and
may not be a valid source of data.

Limitations: Practical limitations tend to focus on

understanding social change. Historical analysis also
the availability of documentary sources - they are not
reveals the diversity of people's behaviour - things we now
always easy to find - and where they come from. Paper
take for granted may have been seen differently in the past,
documents can be forged and a researcher needs to know
and vice versa.
whether they are originals or copies that may have been
More theoretically, documents can provide qualitative
changed by other authors. Similar considerations apply to
data of great depth and detail. For example, diaries such
digital text, photographic and video sources. We do not
as those of Samuel Pepys, who recorded life in England
always know why or by whom a document was created,
during the 1660s, or Anne Frank, who recorded her
which means we cannot always be sure if it is a credible
life in hiding from the Nazis in Amsterdam during
source. Did the author have first-hand experience of
the Second World War, provide extensive and valuable
the things they describe, or are they simply repeating
details about people and their daily lives. In addition, it
something they heard?
is sometimes possible to compare accounts across time to
More theoretically, documents pose reliability
test the validity of current accounts of social behaviour.
problems in that they may be:
Comparisons of past and present accounts of family and
working lives can help us understand the continuities ■ incomplete
and changes in individual and institutional behaviour. ■ inaccurate
Pearson (1983), for example, used media accounts going
unrepresentative - diaries, for example, may simply be
back over 100 years to demonstrate that 'hooligan' or one individual's view.
'yobbish' behaviour is not a recent phenomenon in
the UK. These problems make documents difficult to generalise.
Documents can also be used for semiological analysis This may not be an issue if there are a large and diverse
that compares their literal (what they actually say) and number of sources from which to reconstruct social
metaphorical meanings - what they tell us about the behaviour, but it is a problem if the researcher is dealing
hopes, fears and beliefs of whoever produced them. with a single documentary source.
 Chapter 3: Methods of research

under consideration. A review like this may generate ideas

ACTIVITY
about what to study, whether to replicate previous research
A simple way to do some sociological research is to and how to avoid errors made in previous research.
take a short walk around the area where you live
Research hypothesis or question
or go to school. As you walk, make a note of the
This is the research focus that sets the basic theme for
people and things you see. What sort of sociological
a study:
picture of your area have you discovered? Order your
observations into meaningful categories, such as: 11 If a hypothesis is used - Ginn and Arber's (2002) analysis
■ housing of how motherhood affects the lives of graduate women
■ car ownership was based on the hypothesis 'The effect of motherhood
on full-time employment is minimal for graduate
■ evidence of crime
women' - it must be tested and this means using research
■ population density
methods suitable for this purpose.
■ cultural life (theatre, museums, type of shops etc.).
If a research question is used - Conway's (1997)
Once you have collected your data, write up a short examination of parental choice in secondary education
report outlining the sociological profile of your was based on the question 'Does parental choice help
area, explaining what common patterns of social to strengthen the advantage of the middle classes over
organisation you have seen. the working class?'- the research method used must be
capable of generating high levels of descriptive data.

Research design
This section looks more generally at the design of Suggest two reasons why a researcher might want to
sociological research. Oberg (1999) suggests there are four review previous research in the area they plan to study.
interconnected stages of research design:

1 Planning is where the researcher decides on the Collecting data

strategy- such as what to research and how to research Before data can be collected, the researcher needs to
it- and formulates research hypotheses or questions. identity the people - or respondents - who will be the
2 Information gathering involves identifying a sample to subject of the research. Although it would be ideal to
study, conducting an initial pilot study and applying select and study everyone in a particular group (the target
research methods to collect data. population), this is not always possible. For example, if
3 Information processing relates to the idea that once data the target population was 'doctors in India', the size and
has been gathered, its meaning must be analysed and geographic distribution of such a population would make
interpreted. it impossible to observe or question everyone personally.
4 Evaluation involves both an internal analysis that asks This is where sampling enters the research process.
questions about how the research was conducted (whether A sample is a relatively small proportion of people who
the research method was appropriate, for example) and an belong to the target population. In the example above, the
external analysis, whereby conclusions are reported to a researcher might choose 1,000 doctors and, by studying
wider public audience for their analysis and criticism. their behaviour draw certain conclusions about all doctors.
However, this only works if the sample is representative
This outline can be developed to show a more detailed of the target population. Representativeness may be more
representation of research design across a range of significant than sample size because it relates to whether
categories. the characteristics of the sample accurately reflect those of
the target population. If:
The research problem 60% of the target population are male
This is the initial stage, when the sociologist decides things
60% of the sample should be male.
such as the general topic to study and then develops more
specific ideas about what aspect to study. This may be If the sample is representative anything discovered can
accompanied by a review of previous research in the area be generalised to the target population. A researcher can
 !
I

Cambridge International AS and A Level Sociology i

,,
I

make statements about a group they have not studied (the

target population) based on the behaviour of a group they

''•'
have studied (the sample).

Sa mpling fra me
Constructing a representative sample often requires
a sampling frame . This is a list of everyone in a target
population, such as an electoral or school register, and it
is used for two main reasons:
,_•
1 Unless everyone in the target population can be
identified, the sample drawn may not accurately reflect
the characteristics of the population.
Random samples are based on chance distributions
2 For a researcher to contact people in their sample,
to interview them for example, they must know who
Systematic sampling
they are.
This is a variation on simple random sampling that is often
However, simply because a sampling frame exists does not used when the target population is very large. It involves
mean a researcher will automatically have access to it. This taking a sample directly from a sampling frame. For a 25%
may be denied for reasons of: sample of a target population containing 100 names, every
fourth name would be chosen. This technique is not truly
■ legality: names cannot be revealed by law
random - for example, the fifth name on the list could never
■ confidentiality: a business may deny access to its payroll be included in the sample so not everyone has an equal chance
records, for example of inclusion. However, it is random enough for most samples.

I privacy: some groups do not want to be studied.

There are a number of representative forms of sampling

Stratified random sampling
Although simple and stratified random samples can be
technique. used in many research situations, problems can occur
when a target population is made up of small groups,
such as a population with numerous age groups. A biased
sample can easily occur by chance, with some groups over-
Sampling and sampling techniques: a samp le invo lves a represented and others under-represented.
small number of subjects drawn from a much larger (target) Stratified random sampling avoids these problems by
population. Sociologists use a variety of random and non- dividing (stratifying) the target population into groups
random sampl ing techniq ues.
whose characteristics are known to the researcher, such
as different age groups. Each group is then treated as a
separate random sample in its own right. We can illustrate
Simple random sampling this by considering:
This is based on the probability that the random
■ a target population of 100 people (80 females and
selection of names from a sampling frame will produce
20 males)
a representative sample. For the sample to be truly
■ a 10% sample.
random, everyone in the target population must have an
equal chance of being chosen. A simple random sample, To exactly represent the gender balance of the target
therefore, is similar to a lottery: population, the researcher needs a sample of eight females
and two males. To ensure its randomness this involves:
■ Everyone in the target population is identified on a
sampling frame. ■ splitting the target population into two groups: 80
■ The sample is selected by randomly choosing people females and 20 males
from the frame until the sample is complete. selecting 10% of each:

A 30% sample of a target population of 100 people, for o eight females from the 'female only' group
example, would involve the random selection of 30 people. o two males from the 'male only' group.
 Chapter 3: Methods of research

When combined, the sample is representative of the target Opportunity sampling

population. In some circumstances it may not be possible to create a
representative sample. Here, the researcher may be forced
to settle for opportunity sampling, a general type of
sampling - with two main sub-divisions:
Target/general
population
= 100 people ~ 1 Best opportunity sampling involves deliberately choosing
[) sample \ a sample that gives the best possible opportunity to test
80% female
+ [10 people]
a hypothesis. If the hypothesis is false for this group, it will
20% male ?
( males probably be false for other similar groups. Goldthorpe
et al. (1968), for example, wanted to test the claim that the
working class in the UK was becoming indistinguishable
A simple worked example of stratified random sampling
from the middle class. Their best opportunity sample
consisted of highly paid car-assembly workers in Luton.
Stratified quota sampling This group was chosen because if any working-class
This variation uses the same basic technique, with two group was likely to show lifestyles similar to their middle-
main differences: class peers, it would be these 'affluent workers'.
2 Snowball samples work on the principle of 'rolling up'
1 Although a sampling frame is always useful, it is more and more people to include in the sample over
not strictly necessary. It is enough just to know the time. The researcher would identify someone in the
characteristics of the respondents in order to construct target population who was willing to participate in their
a sample. Using the previous example, the selection of research. This person then suggests more people who
females and males from their respective groups is done are also willing to participate. These then suggest further
on an opportunity basis. The researcher goes through the possible participants, until the researcher has a usable
group of20 males, asking each in turn to be part of the sample. Although this technique is unrepresentative, it
sample. Once two males have agreed, the quota for the may be the only option in certain situations:
males-only sample is complete and no further males can
Wallis (1977) used snowball sampling to contact
be selected.
(ex-)members of the Church of Scientology when his
2 This technique is not truly random because not
request to interview current members was rejected.
everyone in the target population has an equal chance
In Sappleton et al.'s {2006) research into gender
of being selected. If the first two males agree to be part
segregation, 'respondents were enlisted through
of the sample, the remaining 18 have no chance of being
personal referrals, prior contacts and cold calls'.
selected.

Non-representative sampling
Opportunity sampling can be a useful technique when no
sampling frame is available and the researcher knows little or
Researchers generally find representative samples useful,
nothing about the characteristics of their target population.
but there are times when a non-representative sample
serves the purpose. For some types of research the Pilot study
sociologist might not want to make generalisations about Before embarking on a full-scale study, many researchers
a very large group based only on a small sample. They choose to run a pilot study to test the various elements
might simply be interested in the behaviour of the group of their research design. Pilot studies are a research tool
itself, rather than what it represents. Case studies, for normally used for one of two reasons:
example, involve studying the behaviour of a particular
group (or case) in great detail and can be used to illustrate 1 As a 'mini version' of a full-scale study designed to test
how a non-representative sample works: the 'sample' is the the feasibility of carrying out such a study. In other
target population. In Ward's research on rave participants words, before embarking on a study that may take up
and Venkatesh's gang study, the fact that each group
was only representative of itself was unimportant, since
neither researcher wanted to generalise their findings; Pilot study: 'mini version' of a full-scale study designed to
they simply wanted to understand a particular group in test its feasib ility.
depth.
 Cambridge International AS and A Level Sociology

large amounts of time, money and effort, a researcher construction. Not everything the researcher saw, heard
may conduct a smaller study to identify any problems, or recorded is presented for public consumption. This
such as access to respondents, that may occur in a is partly because it would result in lengthy reports and
larger study. A pilot study is also helpful in determining partly because some data may be considered irrelevant
the resources, such as staffing and finance, needed to the research objectives.
for a study. The results of a pilot study can be used to
demonstrate to funding bodies that a full study would be
Presenting completed research
feasible and worthwhile.
Glaser and Strauss (1967) suggest that the final stage of the
2 To pre-test a research method, such as a questionnaire.
design process involves four related elements:
This might involve testing different types of question,
examining and analysing the data it produces to ensure 1 analysing related research to discover common themes
the questions will elicit the data required, and identifying and trends in the data
and eliminating possible sources of bias or unreliability, 2 reflecting on the research itself; does it, for example,
such as leading or ambiguous questions. support or disprove the hypothesis?
3 is it possible to discover patterns in the data?
4 does the research suggest ways the data can be linked to
create an overall theory?
Identify and explain one difference between a stratified
random and a stratified quota sample. Once the data has been analysed and interpreted it can be
presented in terms of its:

■ findings
Data analysis ■ conclusions about, for example, the hypothesis (has it
Foucault (1970) argues that data 'can never speak for itself';

I
been disproven?)
it has to be analysed, by bringing together and categorising ■ limitations, which might include discussion of
related ideas, and interpreted: what, in short, does the data various research problems that may have affected
mean? In this respect, analysis and interpretation takes the study
place on three levels: ■ suggestions for further research
1 Private/internal analysis involves using concepts like ■ improvements to the research design.
such as reliability and validity to ensure data is logical
and consistent.
ACTIVITY
2 Practical analysis relates to the purpose of data
collection - the idea of doing something with the Design a poster that identifies and explains the different
data. Wilkinson and Pickett (2009), for example, used stages of research design. Once you have done so, choose
comparative data drawn from secondary sources to make a sociological topic that interests you, and think about
extensive correlations between social inequality and how you would approach the research of this topic,
crime. These suggested a possible causation: the more based on your poster. Make a clear connection between
unequal a society, the higher its relative level of crime. the outcomes you want from the research and the
3 Public/external analysis relates to the idea that all methods employed to get to those outcomes.
research represents the outcome of a process of social
 Chapter 3: Methods of research

Exam-style questions

• There are two different types

of data:
■

■
non-participant observation
overt participant observation
a What is meant by the term primary
data? [2]
■ primary ■ covert participant b Describe any two differences between
secondary. observation
■ quantitative and qualitative data. [4]

• ■ cross-sectional surveys
Each can take one of two
■ semiology. C Explain how primary quantitative
different forms:

■
qualitative
quantitative.
• Secondary qualitative methods
include documents:
methods can be used to generate
sociological data. [8]
personal

•
■ d 'The strengths of qualitative methods
Primary quantitative official
■ lie in their ability to generate highly valid
methods include:
■ historical.
data.' Assess this claim. [11]
■ questionnaires (postal)
The four stages of research
■ structured interviews
design are:
■ content analysis Total available marks 25
■ Planning:
■ experiments
- the research problem
■ longitudinal surveys.
- hypothesis or question

• Secondary quantitative
methods include official
statistics.
- pilot study
-sampling:
• simple random

• Primary qualitative
methods include:
• systematic
• stratified
• non-representative.
■ semi-structured
('focused') interviews ■ Information gathering.
■ unstructured interviews ■ Information processing and
■ group (focused) analysis.
interviews ■ Evaluation.