Flanders - Differentiating among categories
Here are the Points of Contention between Xiao and I so far:
2 vs. 3: Simple Okay's, Umhm's, repeating of the answer don't count as 2's. There must be some indication of judgement by the teacher that the answer is a good one.
4 vs. 5: Is reading a question off a worksheet, a question? No. There is no verbal response by the student. I'm still not sure about this, but it feels like this is the right way to go.
8 vs. 9: Xiao and I disagreed over a response to an open-ended question was an 8 or a 9. I said 9 because the teacher is not looking for an answer they have in their own head, but is listening to the student's idea. Also this distinction will help us identify patterns of open vs. closed questions.
Coding across time intervals: I had advised Xiao to code the extra segments, whereas I myself was not doing so. If a code goes into an interval by only a syllable then I don't code it. Whereas if there is a change or sequence of codes, then I do code it. We also resolved to avoid multiple codes in single intervals as much as possible, breaking Amidon's Rule #4. This was in an effort to increase reliability, but might threaten validity of our results.
We also conflicted on the when the time interval started and the code represents the interval before or after the stated time. This was why we were off by 3 seconds overall. We resolved to have the code stand for the stated time until the next stated time. (eg 0:00:03, 5 means the interval 0:00:03 to 0:00:06 was coded 5)
Our overall reliability was 41% for BA_2004_03_29. This is unsatisfactory. Having had the conversation over the above issues, it is hoped that reliability will increase significantly.
We've found that in sections of high interaction, it had been difficult to get reliable results so far. The intervals might need to be reduced to 1 sec in such cases. Xiao and I achieved 58%.
2 vs. 3: Simple Okay's, Umhm's, repeating of the answer don't count as 2's. There must be some indication of judgement by the teacher that the answer is a good one.
4 vs. 5: Is reading a question off a worksheet, a question? No. There is no verbal response by the student. I'm still not sure about this, but it feels like this is the right way to go.
8 vs. 9: Xiao and I disagreed over a response to an open-ended question was an 8 or a 9. I said 9 because the teacher is not looking for an answer they have in their own head, but is listening to the student's idea. Also this distinction will help us identify patterns of open vs. closed questions.
Coding across time intervals: I had advised Xiao to code the extra segments, whereas I myself was not doing so. If a code goes into an interval by only a syllable then I don't code it. Whereas if there is a change or sequence of codes, then I do code it. We also resolved to avoid multiple codes in single intervals as much as possible, breaking Amidon's Rule #4. This was in an effort to increase reliability, but might threaten validity of our results.
We also conflicted on the when the time interval started and the code represents the interval before or after the stated time. This was why we were off by 3 seconds overall. We resolved to have the code stand for the stated time until the next stated time. (eg 0:00:03, 5 means the interval 0:00:03 to 0:00:06 was coded 5)
Our overall reliability was 41% for BA_2004_03_29. This is unsatisfactory. Having had the conversation over the above issues, it is hoped that reliability will increase significantly.
We've found that in sections of high interaction, it had been difficult to get reliable results so far. The intervals might need to be reduced to 1 sec in such cases. Xiao and I achieved 58%.
posted by John Baek @ 4:30 PM
0 Comments:
Post a Comment
<< Home