For many qualitative researchers, code and

retrieve is synonym to doing qualitative analysis.

Especially for people that use Content Analysis.

In Content Analysis, you code for themes in text.

Those text could be any kind of text.

Those text could be transcripts, interview transcripts.

But, also, documents you gathered during your participant observation.

Twitter messages, so Tweets, or any other social media text.

Or photos probably, maybe even films.

So it could be all kinds of texts.

And what you do when you do content analysis, you try to look for

things in this text, and you try to organize them.

And usually this organization is done by using codes.

Hsieh and Shannon make a distinction between three different forms

of content analysis.

The first form is Conventional Content Analysis.

And in conventional content analysis, you simply start out.

You start by reading texts and then coding texts.

You don't start with a theoretical framework or something like it.

No, you start to read and try to develop codes and themes from the text.

So bottom up, more inductively.

And they call it conventional.

A second form of content analysis they recognize is Directed Content Analysis.

In directed content analysis, you bring in a coding scheme or

a coding system, whatever you call it.

You bring it in from theory.

From other research, or other researchers.

Or probably your own conventional content analysis that you did beforehand.

And then, you apply it to a number of tests.

A third form they recognize is a more Summative Content Analysis, and

that form of content analysis is more quantitative.

So when you start Googling on content analysis,

you'll quickly run into any of these three forms of content analysis.

So be aware, when you look up for a more conventional content analysis,

you really pick literature on conventional content analysis,

because there has been a lot written about content analysis.

Probably more quantitative point of view.

When starting content analysis, you have to think about your text.

First, you have to think, what kind of themes do I want from this text?

Are these themes more manifest, or more latent?

So can I read them directly, or do I need to read between the lines?

Or do I need to combine it?

A little bit of both.

And if I look at manifest meanings and

manifest themes, do I need to code them exactly as they're worded?

So code in vivo.

Or can I use more abstract codes?

These are aspects you have to think about when doing any form of content analysis.

Usually, when you use content analysis,

in the end you do a more directed, more top-down form of content analysis.

And therefore, you need a coding frame, a coding scheme.

That's what you develop.

And that's the outcome of this conventional content analysis.

You need definitions of codes, because you have to take care that you code every text

in the same way, with the same definitions.

You need coding and interpretation rules.

So applications of those definitions.

And very often you work with multiple coders.

So you need great instructions for coders.

I specifically said great instruction, because it's very important in content

analysis that every coder does exactly the same.

And therefore, many people also use intercoder reliability measures,

such as Krippendorff's alpha.

Now, let e give you an example.

Let's say I did interviews about flower power, and

I spoke to hippies about what the meaning of flower power is to them.

So I did interviews and I transcribe these interviews.

And in the white part, I spoke about the meaning of flower power.

And this person talked about it quite extensively.

And this person talked about it a lot.

Flower power meant everything to her.

And then, this person didn't spoke about much.

And this one spoke about it.

And this one as well.

So I have all these interviews with fragments about flower power.

Now, I can count these fragments.

I can look at the content of these fragments and

maybe I can organize the themes in it.

Now, lets say this one was talking about love and peace and

this one was talking about marijuana, that's why I didn't say much.

This one was talking about love and peace.

And this one's about flowers, and birds, and rainbows, and so on, and so forth.

Now, how to organize it then?

I could look at those interviews and just compare those interviews.

But it works way more easy when you put it in a matrix.

Now, let's do that.

If you create a matrix with the themes on top, so

this theme, theme four, is flower power.

This theme is thoughts about gender.

And this is interview one, interview two, interview three, and so on.

Now, what you would do then is you cut and paste the fragments in this matrix.

And it works tremendously well,

because what you can do now is look at a single interview and

see what people say on theme one, on gender, on any other topic,

on theme four, the meaning of flower power to them.

And then, you can compare this interview with this interview.

And you can compare within a column.

And within a case and a column.

So you can combine cases and columns.

Now, that works great for fragments, but it's a lot, and it might be a bit messy.

And people tend to talk for hours and hours, especially about flower power.

So what you also can do is, not just use the fragment here, but use the codes here.

For instance, the first two interviews were about ecological issues,

flowers, and a dove, and a rainbow.

Whereas, the third interview was only about marijuana.

And the last two interviews were a bit more political,

because they were about love and peace.

So what does flower power mean to different persons?

Well, drugs, ecology, or more political, love, peace.

So this is a way of organizing your analysis.

Using a matrix it looks a bit like what people used to do in

more quantitative analysis, but

it definitely is a qualitative analysis form, quote and retrieve.

[BLANK AUDIO]

5.2 Code and Retrieve

定性研究方法

Meet the Instructors