Jump to content

Kate Edwards-Davis

QSR Staff
  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Kate Edwards-Davis

  1. Hi Scott, Firstly, here's how to visually compare the coding from different users in your project. Once you have run a Coding Comparison Query, you can right-click on any single row in the query results, and select 'Open Source'. NVivo will open the image and show coding stripes for the node represented in that row, including coding stripes for each user in your comparison query. I've attached an image that shows an example of these stripes. In NVivo, if you click on any one of the stripes, the associated region of the image will be highlighted. Secondly, from your reported query results your Kappa result of 0.6942 looks correct, although our approach to calculating it would have been different. Following the example in the spreadsheet located at http://redirect.qsrinternational.com/examples-coding-comparison-nv10-en.htm, I would start with the average of each Agreement and Disagreement detail columns, rather than the totals. Deriving the averages from your totals I get: Agreement Values Total Average % Present-Present (A+B Agreement) 49 3.780864 Present-Absent (A but Not B ) 31 2.391975 Absent-Present (B but Not A) 9 0.694444 Absent-Absent (Neither A nor B ) 1207 93.13272 However, these two approaches are equivalent in their outcome. You can calculate the Kappa from either set of these values and achieve the same score. I verified this by: Putting my calculated averages into the Kappa calculator you specified at https://www.niwa.co.nz/node/104318/kappa, instead of the totals. I confirm that I get the same Kappa score. I also calculated Kappa independently, step by step, using Cohen's Kappa formula as detailed at https://en.wikipedia.org/wiki/Cohen . I did this twice - with your totals and my averages - and on both occassions I got the same Kappa score. This score of 0.6942 represents the Kappa coefficient averaged across all nodes and sources in your original query, where each source is considered to have the same weight. Regards, Kate
  2. Hi TanjaC, You can still use the "Classify Cases from Dataset" function to update the existing nodes in each of your subfolders. What you can try is: On Step 2 of the Wizard, set the location for the nodes to "Existing Folder" and choose one of the subfolders you are using Underneath the 'Name' field on this same page, make sure the option 'Create cases if they do not exist in this location' is NOT checked Continue through the wizard as usual This will classify the nodes in the subfolder you selected in step 1, without creating any new nodes for participants not filed under that subfolder. You can then repeat the process for each subfolder you are using. I hope that helps! regards, Kate
  3. Hi Robert, My suggested approach combines Search Folders and a Coding Query. This approach assumes that you have already created a case for each respondent (coding the entire row of their response to each case). It also assumes that you have created a separate user folder for your cases. First Step: Create a Search Folder for Red, Blue and Green You can create 3 Search Folders to find the cases coded at each of the 3 different codes created for Q1 responses. These search folders would have criteria like: Look for : Nodes That code: Red In folders or Sets: (select your cases folder) Second step: Create a node for Q2 If you haven't already done this, code the Q2 column of your survey to a node (for example 'Q2') Third step: Create a coding query to find Q2 responses for each of Red, Blue and Green You can use the 'Simple' tab in the coding query options, with criteria like: Search for content coded at Node: Q2 In: Selected Items ( <- click select and then choose the 'Red' search folder you created in the first step) I hope this helps. Regards, Kate
  4. Hi Christine, It's a bit difficult to know what is going wrong without more information. Perhaps you could copy your query criteria into a post and describe what you think you are searching for? Assuming (from your message) that you are looking for intersecting content between two nodes, for cases assigned to a particular attribute value, your coding query criteria should look something like this: Content Coded at all of these nodes: {Node A, Node B} AND Coded at any node where Person:Gender = Female This query would return any content from my female cases that is coded at both Node A and Node B. (Note: If I have a Female case and I have coded one paragraph to Node A and a separate paragraph to node B, I would not get any result for that case. The coded content must overlap, and only that part that is overlapping will be returned in the query results.) Thanks, Kate
  5. Hi jrc, I'm assuming the cells you are asking about are part of your classification data? That is, when you've created cases for each respondent, this data would be recorded in NVivo as attribute values? If so, I would leave the cells blank. You can then use the 'Classify from Datasets' function to create cases and assign their attribute values. Any cells that are blank will be recorded as 'unassigned'. Regards, Kate
  6. Hi maumassaro, It isn't possible to use text search query to find instances of ( . You could try using the Find function as an alternative - however, it works on a single document at a time. Regards, Kate
  7. Hi JSG I think I can offer a work-around for you. First you need to do a bit of copying and pasting to create a consolidated source document for each country: 1. Create a new document called 'Interviews from Country A' 2. Open each interview from Country A and copy all of the content (You can use the CTRL+A shortcut to select all content then CTRL+C to copy) 3. Paste the content into the new document 'Interviews from Country A' Note: When you copy and paste the content from one document to another, NVivo also copies the coding. 4. Repeat for Country B Now you can use these two consolidated documents in a model, showing the Nodes that code them. One thing to watch out for - if you have queries in your project that are searching all sources, these two new documents and all of their duplicated coding will be included. One way to address this is to edit your queries to specifically select only the interview documents themselves, excluding the consolidated documents. I hope this helps thanks Kate
  8. Hi JSG, There is a way to create log entries automatically for a given time interval. Open your video and then follow these steps: 1. Click on the 'Layout' ribbon 2. Under the 'Insert' button, choose 'Insert Rows…' 3. Next to the prompt 'Add transcript rows of duration:' enter '10' then click OK This will add a blank log entry for each 10 second interval from the start to the end of your video Thanks Kate
  9. Hi Eleanor, How about this approach? Step 1: Create a Coding Query Build a coding query to return all content coded at : any of {A,B,C,D,1,2,3} AND NOT coded at all of {X,Y} This query returns all the references excluding any reference that is coded at X and/or Y Save the results of this query as a node in the Nodes location - let's call it 'ABCD123 excluding X or Y' TIP: Save the Query. That way when you code more content at A,B.C,D,1,2,3,X or Y, you can rerun the query to update the node. Step 2: Create the Matrix Coding Query Rows: A,B,C,D Columns: 1,2,3 Instead of searching in 'All Sources', change the scope to 'Selected Items' and choose the node 'ABCD123 excluding X or Y' Let me know how you go with this approach :-) cheers Kate Edwards-Davis
  10. Hi Kim, In NVivo, you cannot search for Sources that have attribute values containing a string of text as you describe. Your problem is not unique - there are many users who want to record multiple values for a similar attribute. The most frequently used solution is to create multiple attributes (in your case, one for each type of microbe) and set them up as 'Boolean' type attributes. Then for each source you would record a Yes or No against each attribute. Thanks Kate
  11. Hi reeck009, I wonder if your search phrase includes words that are eliminated from the search - because they are Stop Words? You can see a list of the Stop Words for your project by opening File> Info> Project Properties. On the 'General' tab, click on the Stop Words button. If your search phrase is eliminated because of Stop Words, you can edit your Stop Words list to remove the terms you need. When you do this, NVivo will need to reindex all of the text in your project, so it make take a few minutes. Once you've edited your Stop Words list, the new list will be used for all future Text Search Queries and Word Frequency Queries. thanks Kate
  12. HI jamie10, If you put the two terms within double quotes, the search will only return instances of the specific phrase. Try: "trained assessor" I suggest you take a quick look through the NVivo online help page for Special Characters and Operators in Text Search Queries for more tips. Thanks Kate
  13. Hi Ireiten I don't think you can create a report that includes the information you are after. The best option I can think of at the moment would be to open the initial node and turn on coding stripes for the other nodes. You can then print the 'Reference' tab of the initial node including the coding stripes. thanks Kate
  14. Hi hewhitty, Did you know that you can add a transcript to an audio file in NVivo - allowing you to code and search the transcript text as well as coding the digital audio? If instances of silence, humming, etc are important in your analysis, it would be best to import the audio files into NVivo so that you can code directly from the media. Once imported, you could simply code the relevant sections directly from the media timeline to a node for 'Silence', for example. Each reference coded to the 'Silence' node will show the timespan from the audio file. If your analysis would benefit from differentiating the references by duration, you could then code each of the references in this 'Silence' node to nodes created as children of your 'Silence' node - much as you suggested above. You could perhaps also ask yourself if it's apparent why the interviewee was silent, and what was going on in the interview at that stage. If you are able to determine why, then it would probably help to code on to nodes accordingly. For example - 'Considering response', 'Not willing to respond', 'Reading Museum Signage', 'Distracted' etc I hope this is helpful! Thanks Kate
  15. Hi Kari, I'm guessing you are interested in finding which sources are coded to a particular combination of nodes? (As differentiated from finding out which sources are simply coded at any 2 or more nodes?) If there is one particular coding combination you are looking for, I would use a coding query. The criteria would look for sources that are coded at Node A and Node B and Node C. When the query results are returned, the reference tab will show all of the text. As the text references do not really hold any significance for you, switch to the 'Summary' tab. The Summary tab lists the sources that match the coding combination you are searching for. If you are interested in combinations in general, then I would use a matrix coding query. One possibility is to select the sources as rows and nodes as columns. The results will show the number of references by default - you can change this to the number of coded sources if you prefer. To find sources coded at any particular combination, you can filter by each of the columns representing the relevant nodes. I hope this helps thanks Kate
  16. HI hewhitty, Can I ask - are you coding your interviews from a document or from an imported audio file? Thanks Kate
  17. Hi Ireiten, I'd just like to correct this - NVivo does not contain any limit to the number of attributes you can record for a classification. Similarly the software doesn't limit the number of attribute values recorded for each attribute. When you import data from a survey as a dataset, the maximum number of columns that can be imported is limited to 255. There is also a limit of 1,048,576 rows. Thanks, Kate
  18. is watching NVivo 11 take shape - it's exciting :-)

  19. Hi Orionspants The limit of 50 attribute values only applies to the chart - there is a realistic limit to the number of columns that can be drawn in the chart, after which adding more columns makes the chart less legible. The display order of attribute values in the chart is optional. You can choose from: Order of values as defined in the Attribute (this is the default) Descending Y-axis value (most to least) Ascending Y-axis value (least to most) You can set your preference on page 3 of the chart wizard. For a chart that you've already created, you can use the 'Select Data' option on the 'Chart Tools' ribbon to change the display order. If you want to analyse the number and distribution of responses for more than 50 attribute values at a time, you should use the Matrix Coding Query. Thanks, Kate
  20. Hi Orionspants, NETHRS problem Taking a look at your file in Excel, the following rows have what might look like a number in them - however, there is a space before the digits, and it therefore reads these fields as text. If you trim out the spaces, your file should load correctly. Affected rows are: 66 67 73 77 81 84 86 96 97 102 110 114 116 124 130 134 136 138 142 144 146 149 157 159 167 176 198 199 201 207 214 Country = OTHER When you choose to chart a Node by Attribute Value, NVivo can display a maximum of 50 attribute values in the chart. In situations where you have more than 50 discreet attribute values, NVivo will display a bar for the top 49 attribute values, and include the remaining values in the bar labelled as 'Other'. You might find it more useful to create a Matrix Coding Query using the Nodes you would like to visualise as the Row Items and the Attribute Values as the Column Items. You could then export the results to a charting tool such as MS Excel. I hope this helps you resolve your problems! Kate
  21. is excited to hear what people think about NVivo for Mac

  22. Hi, I'd recommend you use a Matrix Coding Query to see how many participants have coding at a particular theme. A Matrix Coding Query produces a grid - with each cell in the grid showing the number of coding references that are common to the intersecting Row and Column items. You can set up a simple Matrix Coding Query using the Query Wizard: On the opening page of the Query Wizard, choose the option: "Cross-tabulate how content is coded" For the rows, select your nodes representing each participant or Case For the columns, select your nodes representing the Themes Search in All Sources. On the final page, choose to "Add this Query to Project" (so that you can revisit the Query as coding changes or whenever you need). The results are presented in a grid - a number is shown at the intersection of each row and column. By default the number will show the number of coding references from each Case at each Theme. Double-click in any cell of the results and NVivo will open up a new tab showing all of the references that are coded to both the row and the column. You can learn more about Matrix Coding Query in the online help at http://help-nv10.qsrinternational.com/desktop/procedures/run_a_matrix_coding_query.htm Hope that helps! Regards, Kate
  23. Dear Nicole, Thank you for your suggestions. From NVivo 9 onwards, 'Case' nodes are no longer treated any differently from other nodes in an NVivo project. Your examples show two scenarios within the Model tool where you would like to able to view the linkages between Cases (or their attribute values) and Themes. Are there any other areas in NVivo where you would like to see more direct functionality for Cases? By the way, you might like to try using a Group Query to visualise the links between Themes and Cases. In the Group Query, use settings as follows: Look for: Items coding Scope: Selected Items (select the Theme nodes of interest) Range: Selected Items (select your Case nodes) Run the query. Once NVivo has calculated the Query results, you can click on the 'Connection Map' tab to get a visualisation of the connections between the Themes and Cases you selected. Regards, Kate
  24. Hi Natalie, You can code the text in the Content column. You can't code text in the Custom columns of a transcript. You can use text in the Custom columns to filter your transcript, or to create nodes using auto coding. When you use auto coding, NVivo creates nodes for you based on your transcript data. The text in the Custom columns become node names, with the text in the corresponding Content column coded as a reference. You can read more about using Transcripts in our online help - see http://help-nv10.qsrinternational.com/desktop/procedures/basic_coding_in_audio_and_video_sources.htm Thanks Kate
  • Create New...