Consensus Process for Audios
How to use the script for audio consensus
Consensus Process for Audios
Creating differences file
Confirm that both .cha files have been checked initially for mistakes (see: Checking CLAN Files)
Launch Audio_Consensor.jar (on /Desktop/GitHub/Audio_Consensor).
Attach Coder 1's file.
Attach Coder 2's file.
Click OK.
There should now be a _differences.cex (ex. 01_06_differences.cex) file created in the same directory as the two coder's files. If this file does not appear, there is a formatting error in someone's file!
Process for audio consensus
This process is for two coders meeting to do consensus on a file after creating a differences file (above).
Each difference is labeled as "Difference #" with Choice 1 as the difference in Coder 1's file and Choice 2 as the difference in Coder 2's file.
For each difference, make your choice in the empty brackets [] as 1 or [2].
Edit the closest choice and make your decision in the brackets if you need to.
Find the next occurrence of "Difference" using Ctrl+G to run through the differences quickly
After finishing with consensus:
Check for instances for "[]" for missed differences and make your decisions.
After all the decisions have been made, open Audio_Consensor.jar again, click "Completed Consensus" and attach the differences file (01_06_differences.cex) to create the final consensus file in the same directory (ex. 01_06_consensus.cex).
Go through the steps from the Checking CLAN Files section, but now with the consensus file.
Alternative Audio Consensus (old version)
Meet with your partner. Find a time that you and your partner are both available to compare your coding
Make a new File. Choose one of your files to serve as the 'consensus' version and make a duplicate copy of it. Change the file' name to say consensus instead of coder XX, e.g. 01_06_consensus.cex.
You'll see a list of differences between the files. Pink are differences, green are lines one file has that the other does not.
Here are the kinds of differences that are not a problem:
a. one person having comments that the other doesn't
b. being off by one line from each other about where the word occurred (it's okay if there's an extra comment line in between too)
c. there's a difference but one coder had a note (or clearly remembers) that they weren't sure or committed to their code/word
d. the difference is about object presence: since this is a best guess, it'll likely vary a good bit between coders. that's okay, just pick one.
differences. There may be differences in the files for several reasons: one coder may have missed something, or there may be a genuine disagreement about whether something is directed to the baby, or is a certain kind of utterance. For these kinds of differences, make a list of lines to check in the consensus file in an excel sheet called, e.g. 01_06_consensuscheckXX_YY.xlsx. XX and YY are the two coders, and XX is the person whose annotation got copied in step 2 above; it is that file's line #s you're righting down to check.
a. call the first column "line_number": put the line # of the discrepency
b. call the second column "difference": put the difference between the files here.
This should be what the other file has that the copied one does not, words or codes. e.g.:
if XX had 'dog' but YY didn't, put 'no word'
if XX has nothing but YY has hat &d|y|MOT , but ' hat &d|y|MOT ' (you can copy/paste from the matlab comparison screen)
if xx has 'n' for code 2 and yy has 'y' put 'y' in this column
Here is an example difference: the file on the left is missing 'dogs' but the file on the right put it after the bullet!
c. check differences once you've made this list, open up the consensus copy of the clan file, and load the audio file. nb: this step can be done by just one person, or together with your partner
decide for each difference, decide whether to change or keep the entry on the consensus copy. put a 'c' or 'k' in column 3 of the excel to indicate your choice.
add a notes column in column 4 if you'd like to make any notes
Last updated