An introduction to educational research methods. Введение в образовательные исследовательские методы Білім беру-зерттеу әдістеріне кіріспе

жүктеу/скачать 32,4 Mb.

Pdf көрінісі

бет	39/85
Дата	06.03.2017
өлшемі	32,4 Mb.
	#8078

1 ... 35 36 37 38 39 40 41 42 ... 85

TRIANGULATION

This is a widely endorsed strategy for strengthening the internal validity of qualitative studies

in social science. It is based on the principle of confirming findings through the use of multiple

perspectives. The key aspect of the strategy is threats to the validity of the conclusions,

caused by the particular biases of any one source, method or agent of research, and which

will be lessened by employing a variety of type.

Why is it necessary?

Writers have argued that there is a need for several types of triangulation in relation to

qualitative research, but I shall focus on three modes which can have a bearing on the type

of research design discussed in this chapter: triangulation of sources, of methods and of

investigators. Triangulation of different sources can mean simply the inclusion of multiple

informants (i.e. interviewees or subjects under observation). However, by including different

categories of informants (e.g. teachers and pupils, secondary- and primary-based informants,

pupils of different gender or ability or from different classrooms), the converging perspectives

will arguably make the findings more powerful. Triangulation of methods involves the

use of different data-collection instruments with the same subjects. This has the merit of

counterbalancing the threats inherent to any one method. For instance, the likelihood of

the ‘Hawthorne effect’ (changes in participants’ behaviour during the course of a study

may be related only to the special social situation and social treatment received) occurring

is arguably stronger in the case of individual interviews than in the case of questionnaires,

particularly if the interviewer is known to the interviewee. Questionnaires, however, have

their own and different weaknesses, such as less freedom for the interviewee to develop

their own thoughts and a reliance on the written form of expression. Finally, triangulation of

investigators refers to the use of more than one person in collecting or analysing the data.

One can argue that triangulation as a research device is particularly valuable in the case of

introspective research, though as we shall see in our examples, it is also a useful strategy in

strengthening the credibility of observational studies.

Some commentators (e.g. Silverman, 2010: 371) argue that the use of triangulation is

inappropriate in qualitative research as it contravenes the ethnographic perspective in which

the context of each source of evidence, of each method used and of each investigator’s

approach is of intrinsic value. Comparing results from different angles of enquiry, therefore,

cannot be used to strengthen a common finding. However, a less extreme view would be

that it is possible to maintain the primacy of the researcher’s main source and methods of

research as well as of his or her own interpretation of the results, while also drawing on

multiple sources of evidence and analysis in support of the validity of that main focus.

How can the threat be reduced in both studies?

As an example of introspective research, John’s study would benefit from all three kinds

of triangulation. In particular, the type of study envisaged here calls out for the need for

triangulation of sources. By interviewing teachers and pupils, including, if possible, teachers

from the feeder primary school, John would be able to gather together the views of a range

of people on the issue of the effects of a KS2/3 transfer on the learning of geography based

Reliability And Validity In Qualitative Research By Teacher Researchers

338

on triangulation of sources of information. If the conclusions relied exclusively on evidence

from the pupils, we would feel less confident about them, since, taken alone, the pupil

interview data may be skewed by the ‘Hawthorne effect’ or by a partial recall of geography

learning in the primary school.

Despite the observational focus of Jane’s study, the internal validity of the findings would be

strengthened in particular by the use of triangulation of methods. In order to make general

claims about how her pupils collaborate in small-group interaction and what forms of target

language use this generates, she needs to have some evidence that the pupils’ performance

is not entirely dependent on the nature of the group task or on the dynamics of particular

peer interrelations. In order to do this, Jane can usefully think about different methods of

eliciting group target language use by asking them to engage in different types of language

tasks over her data-collection period. These could, for instance, include role plays, debates,

problem-solving tasks or description tasks. It would be important to include a variety of

task types since hidden pupil preferences may be determined by the type of language

needed in the different tasks (for instance, knowledge of particular vocabulary topics), or

by the possibility of different kinds of interaction (e.g. argument, collaboration) appealing to

different types of personality.

Jane’s interest in probing the reasons why her pupils perform linguistically in particular ways

could lead to conclusions whose validity may be under threat unless she took suitable

measures to strengthen her case. In order to offset exclusive reliance on her own explanation

of the data, Jane could carry out individual pupil interviews as another form of methods

triangulation in her study. This would be especially useful in investigating the reasons why

individuals code switch at particular points in the group dialogue. Though Jane may be able

to interpret some possible reasons by looking for patterns of switching in the transcripts,

her analysis would benefit from asking some of the pupils to come up with their own

explanations. She might also be able to use the interviews as an opportunity to test out

some of her own interpretations of the code-switching patterns on the pupils themselves

to see whether they confirmed her own views. Of course, even if the pupils rejected some

of her conclusions, this would not necessarily invalidate them, since the impulse might be

operating at a subconscious level. However, having this discussion with the pupils in itself

would strengthen her study by providing a thicker layering of interpreted data, and by

demonstrating more of an awareness of the complexities of the research process.

As far as triangulation of investigators is concerned, both John and Jane would gain some

benefit from asking a colleague to examine a sample of their data relating to a particular

aspect of their research, in order to see if the kind of interpretations they themselves are

making of the data are reiterated in their colleagues’ responses.

How can the threat be reduced in both studies?

As an example of introspective research, John’s study would benefit from all three kinds

of triangulation. In particular, the type of study envisaged here calls out for the need for

triangulation of sources. By interviewing teachers and pupils, including, if possible,

teachers from the feeder primary school, John would be able to gather together the

views of a range of people on the issue of the effects of KS2/3 transfer on the learning

of geography based on triangulation of sources of information. If the conclusions relied

exclusively on evidence from the pupils, we would feel less confident about them, since,

Reliability And Validity In Qualitative Research By Teacher Researchers

339

taken alone, the pupil interview data may be skewed by the ‘Hawthorne effect’ or by a

partial recall of geography learning in the primary school.

Despite the observational focus of Jane’s study, the internal validity of the findings

would be strengthened in particular by the use of triangulation of methods. In order to

make general claims about how her pupils collaborate in small-group interaction and

what forms of target language use this generates, she needs to have some evidence that

the pupils’ performance is not entirely dependent on the nature of the group task or on

the dynamics of particular peer interrelations. In order to do this, Jane can usefully think about

different methods of eliciting group target language use by asking them to engage in different

types of language tasks over her data-collection period. These could for instance include role

plays, debates, problem-solving tasks or description tasks. It would be important to include

a variety of task types since hidden pupil preferences may be determined by the type of

language needed in the different tasks (for instance, knowledge of particular vocabulary

topics), or by the possibility of different kinds of interaction (e.g. argument, collaboration)

appealing to different types of personality.