Вы находитесь на странице: 1из 6

Chapter 4

Observe observing the results of the plan

Preparing for data analysis

Analyzing AR data is a continuing process of reducing information to find


explanations and patterns. There are no quick-fix rules for this type of analysis. Try to
extract the key meanings and messages and centre our attention on what the data are
telling us. In this way we will refine our understanding about what is going on. we can
draw out new concepts, develop personal theories and find fresh possibilities for
classroom practice that we can test out in further cycles. Some people refer to this
analysis process as squashing or, as Jerry also called it, crunching the data. Calhoun
(1994, cited in Burnaford, 2001, p. 68) refers to it as squeezing the data to see what
meanings we can find. Although its a really good idea to begin scrutinizing and
reflecting on our data as soon as we can, we will inevitably get to a stopping point where
we are ready to look back over the whole data set. There are then some useful steps we
can follow to get an overall framework for the analysis.
1.
2.
3.
4.
5.

Assembling your data


Coding the data
Comparing the data
Building meanings and interpretations
Reporting the out comes

Analyzing and synthesizing qualitative data

Qualitative data are those that are analyzed without using numbers. we are likely
to get quite a lot of qualitative information from data such as journal/diary entries,
interviews, classroom recordings of interactions among ourselves and/or our students,
and observation notes. The main tools for analyzing qualitative data that were going to
examine in this section are: i) categorizing; and ii) analyzing talk.
Categorizing
Sorting objects and information into logical groupings is something that human
beings do constantly in daily life. Essentially, this is what categorizing data is all about.
We generally treat qualitative data in AR though a process of what is known as inductive
coding. In other words, we scan the data carefully, usually several times over, to see
what categories suggest themselves, or emerge, from the data. The research term for
this is an emic approach, which is sometimes also called an insider approach. It means
that we look at data from the perspectives of people closely involved in the research
context and analyze their opinions and views exactly as we find them. In other words,
the data give us the categories. This is a different approach from deductive coding where
the researcher develops categories based on the literature or on theoretical knowledge
and then looks for instances in the data to match those pre-arranged categories.
Coding the data in this way basically means that we are looking for the main
themes or concepts that will help to throw light on your research questions, puzzles or
dilemmas. Its a bit like listening for the most prominent refrains or beats that keep
repeating themselves in a melody. Here are some possible steps for working out
categories for coding our data.

1. Gather together all the data we want to look at for example, our/you students
journal entries or the comments made in an interview.
2. Read the data over several times to get a good feel for what is there. As we read,
use a highlighter or underline anything that seems to make an impression on us
3.

(interesting, surprising, unexpected, and so on).


Now take a section of the data and give a label to the main idea or theme that is
being expressed (e.g. the role of the teacher, learning strategy, feelings and
emotions, and so on) this will provide we with an initial category. Look for
other things in later sections of the data that could also go under that label (for

4.

example, are the same words mentioned over and over again?).
Keep doing this until we have developed a number of different labels for the
ideas in the data that group together. We may find that some statements dont fit
neatly into any of the categories. We can put these into a miscellaneous
category in the meantime. Go back over them later to see if they form
categorizing you are likely to have a better idea of where some of the

5.

miscellaneous ones might fit.


Make a list of our different initial categories on a sheet of paper or on a

computer.
6. Now look back over our broad categories and see whether we can group them
more into sub-categories that all relate to the main category label. Keep doing
this until we are satisfied that we have identified all the ideas that cluster
together.
7. Make a note on our category list (see point 5) of where the sub-categories fit in.
8. Show our categories to a colleague and see whether they agree with the way we
have grouped them. Discuss the ones where we have a difference of opinion and
decide how best to categorize them. Asking someone else to look over our

analysis can give we more confidence that we have come up with some reliable
categories.
To provide a very concrete example of how coding can be done, we might begin
by getting a large sheet of paper and giving it a provisional heading Views of good
teachers. A provisional heading is a temporary label that weve identified by reading
over our data set several times. We then place all the pieces of data weve cut up that
relate to views of good teachers on that sheet. The next step is to sift again through all
the pieces of data weve placed under your first heading. Andy found that his data broke
down into sub-categories Positive views and Negative views. Now you can refine the
first heading into two (or more) sub-headings.
Once the sub-heading step is completed you can go further and see whether there
is a third level of coding. For example, it might be possible to categorize the positive and
negative views of teachers in relation to: the role of the teacher, the behaviour of the
teacher, or the skills of the teacher. We can split the data up now across as many sheets
of paper as we need for the different sub-categories.
One of the challenges of doing coding in this way is that we may find that some
pieces of data fit into more than one category. One way to handle this is to copy them
and for the moment place them in both categories. Alternatively, we can put them onto a
sheet of paper with the title miscellaneous. Later, when we are further along in our
analysis and things are becoming clearer, review them again to see whether it is now
more obvious where they fit in. It might also be that the miscellaneous data start to form
new categories of their own.

Analyzing talk
If youve decided to focus our research on classroom talk our data will probably consist
of recordings and transcripts weve made from all, or parts of, these recordings. The
analysis of spoken interaction in the field of applied linguistics has become highly
developed both theoretically and practically in recent years. If you are keen
to learn more about the main approaches it would be worth consulting specialist books.
Richards (2003, ch. 4) is a good introduction to the main approaches of (critical)
discourse analysis, conversation analysis, interactional sociolinguistics and ethnography
of communication. When analyzing classroom talk it is worth referring back regularly
both to the recordings and to the transcripts to deepen our analysis. Remember that the
transcript can only ever be our written interpretation of the original interaction, so
checking out what was actually said from time to time is very worthwhile. The basic, but
most important, aspect of analyzing classroom talk is that we are listening deeply to
what we are hearing. Richards (2003, p. 185) suggests four steps in a basic analysis, and
in the discussion below will follow his model:
1. Providing a general characterization
This step involves getting a general sense of the kind of interaction you are
listening to and giving it a description.
2.

Identifying grossly apparent features


Richards suggests that it is best to hold back on interpretations at this point if
you can (2003, p. 187) and just focus on the obvious. Looking more closely at

the structural elements and what they tell us is part of the next step.
3. Focusing in on structural elements

This step is where we begin to examine the finer details of how the interactions
are structured and describe the most noticeable features.
4. Developing a description.
Richards notes that once steps 13 are completed we are well on the way to a
description (2003, p. 187) which involves giving an objective picture of the
interaction related to what we have seen in the data.

Вам также может понравиться