Inter-variation and Intra-variation

[kellogg]

발표자	Independent Variables (Input, treatment, manipulation variable)
정윤희	Group discussion about English literature for children Good! Now, one of the peculiarities of ACTION research is that the variable doesn't really vary--we just give the input, we do the treatment, we perform the manipulation, but we do NOT give any variation, because we don't have any control group. Actually, I think this is a GOOD move. We're teachers first and researchers only second. When you're a teacher, you teach the best lesson you can, whether the kids are in an experimental group or a control group. When a teacher finds something that works, the teacher shares it with everybody. But of course there ARE lots of variables WITHIN the treatment, which we can call "intra-variable variations". For example, there are different books. That's a variation. And there are different ROLES for the children. That's another variation. The discussion can take place in Korean, or it can take place in English. That's a variation too. With a good transcript, we can compare the SAME child with DIFFERENT books. We can compare the SAME roles with DIFFERENT children and the SAME children in DIFFERENT roles. We can look at (e.g.) the emotional "color" of a discussion in English and compare it with the emotional 'color" of one in Korean. I think that if our transcripts are good enough, and long enough, and well interpreted, we can probably make generalizations that hold true not just for one classroom, but for many.
한지희	Digital text book in Blended-learning One of the criticisms that has been made of "Blended Learning" is that there is now no such thing as non-blended learning: ALL learning involves a "blend" of modes of delivery. So "blended learning" is not a variable, because it cannot be varied. There are a lot of things I don't like about the phrase "blended learning" (I don't like "immersion" or "absorption" or any metaphor for learning that suggests that language is a liquid or that children are basically aquatic animals, and of course "blending" suggets the mixing of liquids.) But you can see that one ADVANTAGE of "blended learning" is that it DOES recognize the possiblity of intra-variable variations: we can easily compare a "blended learning" programme with MORE digital delivery and one with LESS (and in fact we can do it with the same class, if we have a good transcript).
박성경	Activities focused on Collocation This was the only thesis project that used a control group, if I remember correctly.
김상민	Peer-assessment in an assessment process of English oral proficiency??? (I don't think that peer-assesment is the right independent variable for this thesis, but I can't find any other variable...) Ms. Choe is right.  It's true that there are TWO variables under study. But they aren't really independent, because the researcher cannot manipulate them (they are manipulated by the subjects), and because there are no variables which depend on them them ("reliability" and "validity" are not variables; they are theoretical constructs).  When a researcher performs a CORRELATION, there has to be more than one variable. A correlation is a look at the extent to which two variables CO-VARY, that is, the variation in one variable can be mathematically predicted from variation in another. If they correlate as 1.0, that means that the value of one variable can be completely predicted from the value of another. If they correlate at 0.1, it means that only about one percent (0.1 SQUARED) can be predicted. And of course if they do not correlate, that means that there is no relationship at all. Now, the key problem for Mr. Kim was to determine the CORRELATION between peer assessment and TEACHER assessment. Right?  The "outcome" Mr. Kim expects is some measure of the "reliability" or the "validity" of peer assessment. Of course, it's not really a dependent variable, because it doesn't "depend" on the variables of peer assessment or teacher assessment.
이세운	predicting activities for reading stories Good. The problem here is the dependent variable, of course. What "depends" upon the predictive activities is comprehension. We know that comprehension is VERY hard to measure. Tests for elementary school kids are problematic: because they are often more difficult than the actual text, they often test ONLY the child's understanding of the test question. One possible way around this is to use a FUNCTIONAL definition of understanding; the children understand if and only if they can make an appropriate response or continue the story in an appropriate way. There problem is that, as we know from Chapter  Five, the FUNCTIONAL equivalent of a concept can often be STRUCTURALLY quite different. Children can re-enact a story and even improvise upon it, without completely understanding the WHY behind the story.
이수진	TEE(I can't find any independent variable here) Remember that this is a study of RELIABILITY. That means that the variables are going to be different ratings by different raters. You then perform a CORRELATION (see the study by Mr. Kim Sangmin on peer assessment) to decide if the ratings are reliable. There was also a questionnaire given to the subjects, of course. They were asked if they thought the rating system was fair. Now, in theory, questionnaires like this are dependent variables. Each item "depends" on a given treatment (the experience of being tested). Usually when you do a questionnaire like this you actually have to "correlate" related items with each other to see if the questionnaire is valid (e.g. "I rather like TEE" has to correlate negatively with "I really hate teaching English through English"). I don't remember if Ms. Yi Sujin did this or not.  But I think that Ms. Choe has the gist of the research correctly when she says there is no independent variable, because, just like Mr. Kim's research, the research depends on covariance, on relating TWO variables that BOTH vary, rather than seeing how one variable brings about changes in another. I think that Ms. Yi's questionnaire research was just kind of an afterthought (although it was the most interesting part of the research for me!)
이지현	a content text in a process drama Good! Actually, that's EXACTLY how I understood Jihyeon's research. It's easy to see ONLY another piece of action research, in which the treatment is simply "content text + process drama" and the dependent variable is something like "proficiency" or "real communication" or "authenticity" or something like that. But what is really interesting about this research is not the extent to which it produces something that is inherently unmeasurable (e.g. "authenticity") but the extent to which it produces something that is really VERY measurable: the novel use of a content text. After the presentations, Koala and I had a nice long chat about, among other things, songs and chants, and rote repetition, and Ms. Kim Jihyeon's remarkable mastery of pronunciation. Koala argued that ROTE practice, not conscious control, was key; she took a strongly ASSOCIATIVE and ANTI-INTELLECTUAL stance, the sort of thing we find in Thorndike and the behaviorists (and also in input-based theories of acquisition which do not recognize the importance of noticing or "focus on form). I complained that it is really NOT possible to teach a language without any creativity; at some point we HAVE to ask children to produce utterances that they have never heard before. I think this is particularly true of children: they remember literally, but that means they do not remember very much. So when they try to remember a whole story, they will ALWAYS end up recreating rather than recalling it. That difference--between recreating the source text and literally recalling it--seems to me to be the most interesting thing about this research. For Koala, recall is more interesting than recreation (because of course it DOES sound better!). But for me recreation is far more interesting than recall (because only being able to say things that you have never heard before makes you an active user of the language).
조주은	strategies of using a student CD-ROM Good. Actually, the strategy use is a good example of intra-variable variation: the study is not simply kids who use strategies vs. kids who do NOT use strategies, but rather different KINDS of strategies. The problem is that it's hard to see how we can KNOW what strategies kids are using without asking them. And when we ask them, we are often asking for a kind of SELF consciousness, a kind of metacognitive awareness, that children may not actually have.  The result, of course, is a pseudoconcept: a questionnaire response that LOOKS like a concept but is really just a functional response to the question.

I'm still learning about variables. I think that independent variables are clearer to understand than the dependant variables.

Dependent variables have to be MEASURED. That can be very difficult (and in the case of something like "understanding" and "authenticity", impossible).

Handyman is supposed to tell us about the dependent variables! 

As far as I know, Independent variables are what I want to know and they seem to be very clear to catch when we have an experiment like a horse race.

Oh, I think it's probably truer to say that the DEPENDENT variable is what you want to know. The independent variable is the thing you know already, because you control it. But you are right, some of the things here (e.g. "blended learning" or "collocation") are not really very well known, because they are poorly defined. That is a THEORETICAL problem, not a problem of research design.

 I think, however, there are some kind of thesis like Mr. 김상민's or Ms. 이수진's and including mine that do not have independent variables. Am I correct?  

I think they have variables. And one way to understand them is that each rater is a DEPENDENT variable with respect to the other raters; they are being compared to see if they are mathematically dependent on each other, and therefore dependent on some other construct (the child's proficiency in speaking, or the teacher's ability to teach English through English).

Nice work, Ms. Choe! It's VERY good to hear from you at along last. (Now...let's see what Handyman can do for us!) 

K-Dragon!