How to create a table which tells me how I coded each of my units of analysis (video data)

[hendrik]

Hi everyone,

I am coding video data. Each filmed observation gets 'cut' into units of information. Each unit of information gets some specific codes.

For example, one of the filmed observation lasts for approximately 32 minutes and consists of 353 units of information. Each of these units gets 5 codes that relate to my 5 specific coding categories.

I can run a matrix coding quiery which will give me a summary of of the codings for the whole filmed observation. But I also would like to create a table of which the rows are defined by the units of information and the columns by the specific codings. In other words, the table should contain 353 rows (these represent the 353 units of information) and 5 colums (which tell me how I coded the specific units of information):

 

                  Coding Cat.1     Coding Cat.2     Coding Cat.3     Coding Cat.4     Coding Cat.5

unit 1         Code A               Code K               Code R              Code P                 Code X

unit 2         Code B              Code K                Code T               Code P                 Code Y

unit 3         Code B              Code L                 Code T              Code Q                  Code X

…

unit 353    Code C              Code M               Code S              Code P                  Code Y

 

Afterwards, I would like to export the table to Excell, SPSS, …

Can someone help me?

 

Kind regards,

Hendrik

 

[Anonymous]

Hello Hendrik,

This problem is easily addressed. You need to create a customizable column in your source file. To do this go to file->info_project properties. In the dialogue box that opens, select the Audio/Video Tab. In the sub-tabs below select the video sub-tab. Click on new and call this field 'Unit of Information Number'. Select OK. When you open the video the next time you will now see your new column. Click on the 'Click to Edit' option at the top of the media transcript and give each unit of meaning a unique identifier. This could be a simple numbering system going from 1.1 to 1.353 (the 1 denoting the video number so the numbers are unique across all videos) as in the example used. 

Now, auto-code the media file to a new node. Right click on the media files and select 'Autocode' from the menu. You will now see your column called 'Unit of Information Number' and move this from the left pane to the right pane using the arrow between the panes. 

Select new node from the drop down menu and give it a name and a location. Select 'OK' and NVivo will now create a node with a child node for every unit of information correctly coded. 

Now place these nodes in the column of your matrix rather than than the media files and your five codes should make up the rows and you have the desired result. 

Hope this helps but do let us know if you need clarity on this.

Kind regards,

The QDATRAINING Support Team

[hendrik]

Hi,

Thanks for responding that soon! I have two more questions.

First, I can carry out the steps as written above (creation of a new column in the media transcript, numbering the newly created cells from 1-353, autocode these to a new node which results in a creation of a parent node 'Units' which in turn consists of all the autocoded units. However, when I run the matrix query with the newly created nodes in the colums, the content of all cells is zero. This should not be possible since all the units of analysis (autocoded into new nodes – these are my columns) have been coded. Does anyone know how I could circumvent this problem?

Second, and not as big a problem as the first one, is there a way to get in the colums for every maincategory (my parentnodes) how I coded this maincategory (the childnodes that belong to the specific parent node)

 

this is wat I get now (if my cells wouldn't be empty – see problem 1):

 

                           Code A.1                 Code A.2              Code A.3           Code B.1           Code B.2               ….              Code E.3

unit 1                      1                                                                                                                    1                                                  1

unit 2                                                        1                                                          1                                                                          1

…

unit 353                                                   1                                                                                     1                                                1

 

 

and this is how I preferably would like to see it:

 

 

                       Code A                 Code B              Code E         

unit 1                      1                            2                     1                                 

unit 3                      2                            1                     1                               

…

unit 353                 2                            2                      1                                                          

 

Many thanks in advance,

Hendrik

         

 

 

 

 

 

 

 

[hendrik]

Hi everyone,

 

This message is partly in line with message I posted a couple of days ago, but is more extended now (see the three problems).

 

I am coding in NVIVO9 and I have three problems which I list below. I will first give a short description of the project.

I am coding 32 videotaped fragments. Each fragment gets ‘cut’ into meaningful units of information. As such, most videotaped fragment get cut in about 500 meaningful units. I am coding it by means of play mode ‘transcribe’. This means that each time I play & stop the video fragment, another line in the transcription part is added (containing the timespan for that specific line). So far so good.

My aim is to see for each meaningful unit (these are my lines in the transcription part) how I coded these. A  matrix query can solve this, after I created a new column in the media transcript, numbering the newly created cells from 1-500, autocode these to a new node which results in a creation of a parent node 'Units' which in turn consists of all the autocoded units. That enables me to see the information in the following way:

 

                               CodeA.1              CodeA.2              CodeA.3              Code B.1             Code B.2

Unit 1                   1                                                                                            1

Unit 2                                                   1                                                                                            1

Unit 3                   1                                                                                            1

Unit 4                                                                                  1                             1

…

Unit 500

 

Problem 1

Numbering each newly created cell (with the aim of autocoding – see above) for 32 videotaped lessons takes tons of time (32 videotaped fragments, +/-500 units of meaning for each videotaped fragment = 16000 cells to number manually). Is there a way to number these cells automatically so that I can use these cells with the aim of autocoding?

Problem 2

I would like to see the information in this way (and not in the way as depicted above):

 

                               CodeA                  CodeB                 

Unit 1                   1                             1                                                           

Unit 2                   2                             2                                            

Unit 3                   1                             1                                                           

Unit 4                   3                             1                            

…

Unit 500

 

Is this possible?

 

Problem 3

I have problems with video fragments that I coded by means of drag & drop from the timeline. Although coding stripes show me that I coded these fragments (both coding stripes below the time line and the ones on the right of the transcript field), when I run a matrix query (rows = units of meaning; columns = my nodes), I get the message that all cells of the matrix are empty. Two more rarities along the same line:

–          When I coded three subsequent meaningful units with the same code, and did this for one of these three erroneous, and I uncode this by clicking on the specific coding stripe (and then select uncode), both the other two coding stripes disappear too.

–          When I try to uncode by means of selecting the transcript line, then right mouse click and select ‘uncode’ in the dropdown menu, all my nodes are greyed out (meaning that I did not use any of the nodes to code the fragment?? – this is in contrast with what the coding stripes tell me).

–          When I move the cursor on the horizontal ‘coding density bar’ (the one below the time line) I can see how I coded the specific fragment (a list of nodes that I applied on the fragment shows up). BUT When I move the cursor on the vertical ‘coding density bar’ (the one on the right hand side of the transcript field) I cannot see how I coded the specific fragment (no list of nodes that I applied on the fragment shows up).

All these problems as described under ‘problem 3’ disappear for the video fragments that I coded by means of selecting the transcript line, then right mouse click and select ‘code’ in the dropdown menu, an then thick all the nodes that I want the fragment to be coded.

Any help would be greatly appreciated, since I am running toward an upcoming deadline!

Best, hendrik

[QDATRAINING Admin]

Hi Hendrik,

Let me deal with your second question first because it is the easiest. All you need do here is aggregate the parents (right click on the parent node and choose 'aggregate from child node' in the menu) and the coding from the child nodes will be subsumed into the parents. Then, remove the child nodes form the matrix and replace them with just the parents and you will have achieved the desired results. 

With regard to the first question; it sounds like you did it correctly but I would need to see the data file to investigate what the problem might be. I'm going to e-mail you instructions and passwords for uploading your NVivo project file to our secure server. Once I see what you have done, I should be able to offer a solution and report back to the forum. 

Kind regards,

Ben Meehan

The QDATRAINING Team

[hendrik]

Hi Ben,

Thank you for your post.

It still doesn't give me what I was hoping for. After aggrgating the parent nodes and running an adjusted matrix query, this is what I get:

(the second column is the parent node which is aggregated, now; the next three columns represent the child nodes). It seems that the column representing the parent node, just gives information on wheter any of its' childnodes were coded (1) or not (0). I get the same result if I run the quiry only with the parent node and without the childnodes (then I get only the second column).

 

	A : 0. V, T, N	V	T	N
1 : 212.1	1	1	0	0
2 : 212.10	1	1	0	0
3 : 212.100	1	0	1	0
4 : 212.101	1	0	1	0
5 : 212.102	1	0	1	0

 

What I would like to see as an output is the following:

	A : 0. V, T, N
1 : 212.1	1
2 : 212.10	1
3 : 212.100	2
4 : 212.101	2
5 : 212.102	2

 

where a 1 indicates that I coded the fragment childnode 'V'; and a 2 indicates that I coded the fragment the childnode 'T' (and a 3 would indicate the childnode 'N').

 

Changing the cell contents (right click on a cell still doesn't change the thing). I am going crazy on all these 'practical' issues 🙁

 

Another thing I just remarked. I did these codings now by means of selecting the transcription in the transcription field, right click, select 'code' from the dropdown menu, assign the codes.

when I now run a matrix query, it cannot change the cell content into 'duration coded'. What is this all about?

 

Kind regards,

Hendrik

 

 

[QDATRAINING Admin]

 

Hi Hendrik,

I am sending you a link (privately) to download your project file with the matrix configured to give the results as set out above. You can see from the sample table below that the matrix is now working:

Units of Meaning	0. V, T, N	0.1. Voor	0.2. Tijdens	0.3. Na	0.9999. NVT	888. Onverstaanbaar
1	0	0	0	0	0	0
10	0	0	0	0	0	0
100	0	0	0	0	0	0
101	0	0	0	0	0	0
102	0	0	0	0	0	0
103	0	0	0	0	0	0
104	0	0	0	0	0	0
105	0	0	0	0	0	0
106	0	0	0	0	0	0
107	0	0	0	0	0	0
108	6	0	4	1	0	1
109	1	0	1	0	0	0
11	0	0	0	0	0	0
110	0	0	0	0	0	0

You only missed one step which was you were intersecting the sources with the codes rather than the ‘Units of Meaning’ nodes. If you look at ‘Unit of Meaning’ 108, you will see that the parent node (O.V.T.N.) is aggregated with each sub-code reporting separately. This matrix was created using the units of meaning identified in the source video ‘TB2deel1-Valérievideo 2011_10_21_14_39_37’(auto-coded), and intersecting them with the one of the major five codes, ‘O.V.T.N’ in the ‘TB Test’ folder and its child nodes. The matrix is in the Queries folder and is a copy of your matrix so you can compare the differences between yours and mine. Your matrix has been renamed to ‘valerie_alles – ORIGINAL’ and mine is called ‘valerie_alles – REVISED’ The result is in the results folder and is called ‘OVTN x Units 1 Source’.

I’m afraid I gave you an incorrect instruction in an earlier post. I told you to use a unique identifier for each ‘Unit of Meaning’ which is incorrect in this context. You should use the same identifier as you are comparing multiple source videos. Once you do this you will be able to update the ‘Units of Meaning’ nodes in this matrix with the remaining video sources and when you re-run the query, the matrix will update the tables/results automatically. I manually coded a few O.V.T.N child nodes to ‘Unit of Meaning’ 108 to demonstrate that the matrix was working properly.

One caveat, you will need to download the trail version of NVivo 10 to view this file. This is because I am travelling at the moment and don’t have NVivo 9 on the laptop I’m using on this tour. Like most software, NVivo 10 is not backward compatible. Once you see what I did, you will be able to re-create this matrix in your licenced version.

I’m sorry about the incorrect instruction earlier as it’s much easier to read the situation when I see the project data file.

I hope this helps!

Kind regards,

Ben Meehan

The QDATRAINING Team

 

[hendrik]

Hi Ben,

Thank you a lot for all your effort. I managed to open the file in NVIVO 10, but still have some problems:

For unit 108: I do not need the parent node to give me the sum of all its child nodes (as in unit of meaning 108 seems to be the case). What I want to see in the parent node, is how  I coded its subnodes. For example,

• if I coded subnode A to be present, then I want the parent node to display '1'
• if I coded subnode B to be present, then I want the parent node to display '2'
• if I coded subnode C to be present, then I want the parent node to display '3'

Each unit of meaning gets one code: or the subnode A, or the subnode B, or the subnode C.

Further, I do not understand: "You should use the same identifier as you are comparing multiple source videos. Once you do this you will be able to update the ‘Units of Meaning’ nodes in this matrix with the remaining video sources and when you re-run the query, the matrix will update the tables/results automatically. I manually coded a few O.V.T.N child nodes to ‘Unit of Meaning’ 108 to demonstrate that the matrix was working properly."

and "ou were intersecting the sources with the codes rather than the ‘Units of Meaning’ nodes."

What I did: I autocoded each unit of meaning to be represented in the nodes folder, and I crossed these autocodes with the codes.

Finally, the file I have downloaded, is a read-only file, and I am not allowed to open your revised query.

 

Once again,

Thanks a lot for all your efforts so far!!

Kind regards,

Hendrik

 

 

 

[QDATRAINING Admin]

Hi Hendrik,

Let me deal with your points as you raise them because I think this is a communication issue more than an NVivo problem.

Question 1:

"For unit 108: I do not need the parent node to give me the sum of all its child nodes (as in unit of meaning 108 seems to be the case). What I want to see in the parent node, is how  I coded its subnodes. For example,

• if I coded subnode A to be present, then I want the parent node to display '1'
• if I coded subnode B to be present, then I want the parent node to display '2'
• if I coded subnode C to be present, then I want the parent node to display '3'

Each unit of meaning gets one code: or the subnode A, or the subnode B, or the subnode C."

I think I must be missing something here. The sub-node is displayed in this example so you can se exactly which sub-nodes have been coded or not to  the 'Unit of Meaning', 108. Plus, you can see a total in the aggregated parent or not if you don't want to by simply removing the parent from the matrix and just showing the child nodes. You can see which codes were coded to which behaviors by reading down the the column and which codes were coded to 108 by reading across the matrix. I don't understand the problem? 

Question 2:

"Further, I do not understand: "You should use the same identifier as you are comparing multiple source videos. Once you do this you will be able to update the ‘Units of Meaning’ nodes in this matrix with the remaining video sources and when you re-run the query, the matrix will update the tables/results automatically. I manually coded a few O.V.T.N child nodes to ‘Unit of Meaning’ 108 to demonstrate that the matrix was working properly."

This means that 'Unit of Meaning' 108 is named as '108' in all videos and not given any other name such as '212.108' as was the case with video "TB2deel2-Valérievideo 2011_10_21_18_37_292" This was because of an earlier incorrect instruction from me. Each video is then auto-coded to the units of meaning nodes which form the first column of the matrix. The example used contained just one source. because I could not use the video "TB2deel2-Valérievideo 2011_10_21_18_37_29' because of this naming inconsistency. 

Question 3:

What I did: I autocoded each unit of meaning to be represented in the nodes folder, and I crossed these auto-codes with the codes.

You did not!

This is why I left your original query so you could compare yours and mine. Your query placed the sources in the column of the matrix. This is incorrect. Your query is still there called 'valerie_alles – ORIGINAL'. Mine is called 'valerie_alles – REVISED'. Please compare these two matrices. They are not the same. You placed the sources in the column – see screenshot:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

However, I placed the auto-coded nodes 'Units of Meaning' in the matrix row. You will find these nodes in Nodes/Ben/Units of Meaning

Question 4:

Finally, the file I have downloaded, is a read-only file, and I am not allowed to open your revised query.

 This is because you misread the instructions. 

"You will now see your file as a link at the bottom of the dialogue box. Click on this link to download and save your file. Chose the SAVE option rather than RUN. "

You are getting a read only file because you ran it on the server. If you SAVE the file, you will be able to write to it. 

Kind Regards,

Ben Meehan,

The QDATRAINING Team

[Author]

Posts