Jump to content

helen marshall

  • Content count

  • Joined

  • Last visited

Community Reputation

5 Neutral

About helen marshall

  • Rank
    Casual Member

Contact Methods

  • Website URL
  • ICQ

Profile Information

  • Gender
  • Location
    Centre for Applied Social Research , RMIT Melbourne Australia
  • Interests
    My background is Sociology of the family.
    I am currently working on issues around analysing qualitative data.
    I teach and consult in qualitative research methods. I convene the Qualitative Interest Group which meets monthly at RMIT city campus
  1. I went to open the sample project today and got the database component error message that (I think) usually indicates some system conflict. The problem was not webcompanion - uninstalling it made no difference. Since the only change I had made between last working with NVivo and today was upgrading Windows 10, I uninstalled the upgrade and was once again able to access the sample project and create new projects. Just reporting it in case anyone else finds the same thing .
  2. I am thinking about using the Mac version and working with survey data. If I can’t import an Excel spreadsheet as a classification sheet in NVivo for Mac, can I import it as a dataset and use autocode and 'classify a dataset at cases for each value in a column' to create the classification sheet? Will this give the same capacity to run queries (especially coding and matrix queries) as the windows version?
  3. Nodes and Classifications

    Doesn't the answer to this depend on your research question? If it is sufficient to take the current position of each interviewee and compare what each person says, you create nodes for each interviewee and record their current company and department as node attributes . So you can compare what interviewee 1 (currently in PR at X) says about the functions of marketing with the comment of interviewee 2 ( marketing at Y) If you are focussed more on what is said than on who said it, you might simply code for companies X and Y, departments PR and Marketing and use coding queries to pull togethe material that is about the functions of marketing at company X. If you need both, use both coding and attirbutes. You can create classificaitons and attributes at any stage.
  4. Autocoding using coding patterns

    Thanks for the suggestion Rahul. I think your implied calculation that it is less trouble to do a lot of recoding by hand than to fill in a HUGE classification sheet (several thousand cases) by hand is right. Pondering the issue, I've come up with three possible ways to speed up replicating the coding. First, use two people and have one read aloud the contents of each node while the other codes the new database. (This means that after the first reference has been coded, they can just the relevant text and check the tick box in the quick coding bar). Advantage, you can have some discussion about the coding as you go, disadvantage, you need two people. Second, working solo, create copies of each node and then clear the contents of the original nodes . Working one node at a time, set up a text search query with the results to be coded at the original node. With the copied node open and using key board shortcuts select whole the text of the first reference open the query properties insert the text between double quotes, hit run and wait till you see that the query has succeeded. Then delete the ref from the node copy and repeat the whole thing for the next ref and the one after and the one after and so on. Advantage, you can exactly repllciate the codng without thinking about it. Disadvantage,you are bored and lonely! Third, with two people , do an auto code using patterns one ndoe at a time and check the results against the oroginal node, amending as needed. Disadvantage, you may need to do a lot of tweaking, advantage, the auto coding may throw up interesting things you hadn't thought of.
  5. I'm trying to help a colleague who forgot to add a column with identifiers before importing a very very large dataset (over 4000 cases), then did lots of interactive coding. Because there is no classifying column that she can use to identify, she can't create cases and do any queries using the attributes she imported. I imported a new dataset identical to the one she had coded except that it contained a column with identifiers. I tried autocoding the new dataset using the coding pattern from the old set. I wanted an exact replication since the codable columns in the two sets are identical. Whichever way I ran the autocode, I only got about half the references for each node and the preparation phase always reported that nineteen of the twenty nodes were problematic. From reading the help topics, I guess this is about the calculations of similarity that underlie autocoding, (those mysterious tf-idf thingies that give you the vectors). Removing all stop words from the project made no great difference neither did choice of unit to code. I played about a bit with the less-more slider and got varied results but nothing gave me identical numbers of references or identical content in the nodes. Does anyone have ideas about how I might improve the autocoding? If it can't be done, will we have to code each row as a case, create the attributes, classify the case nodes and then fill in the classification sheet ourselves because we don't have the identifying column to use to classify nodes from the dataset? Any advice would be much appreciated!
  6. I'm playing with the new framework tool. I can build the table using nodes for case nodes and theme nodes. There is content in all the nodes I have used, and I can use the autosummarise button to enter the text from a thematic node in order to create a summary. In the view menu I selected row as the associated display, and I selected the reference tab on the matrix. But clicking in the cell, or on the row does not produce an associated view as shown in help. What silly mistake am I making please? (I'm sure when you put up a viodeo tutorial for frameworks I'll see what it was, but at the moment I just can't get it!)
  7. SEMINAR Meet NVivo 7, the latest qualitative software with Lyn Richards, QSR’s founder. When? Wednesday 24th May, 2006, 5.30pm – 7.00pm Where? Australian Council for Educational Research, 19 Prospect Hill Road, Camberwell Seminar free to AQR members, $10 non-members In this seminar, Lyn Richards will show and discuss NVivo 7, which was launched last month, and which she has now taught in 20 events in three countries. This is not a hands-on workshop but an interactive demonstration and discussion. In the first hour Lyn will be showing the new interface and functions and debating the changes in approach to qualitative projects, via setting up a real one. BYO questions, challenges, discussion topics and issues about qualitative computing! And be sure to stay for coffee and networking. About the software: http://www.qsrinternational.com/ About the presenter: http://www.lynrichards.org/ about AQR: http://www.latrobe.edu.au/aqr/
  8. I've just finished a session with a post grad student who was worried that after coding her first three interviews she had too many tree nodes in her NVIVO project. She certainly did because she’d made the mistake of setting up trees combining themes and characteristics. She has now gone away to merge a lot of nodes and set up document attributes so that she has themes her respondents raise in free and tree nodes, and the characteristics of the respondents in an attribute table. As we talked about what she expected to be doing later, I began to wonder if she should also use case nodes. Her project involves interviews with women at three stages and she has people from varied workplaces, in varied maternal status etc. There will thus be 3 documents per respondent and respondents with different attributes. It might be very useful to compare across time (either the same respondent at three times or all the respondents in one occupation across threes times). It would be easy enough to have time/phase as an attribute. It would also be easy to create case nodes for phase with each respondent as a case. I can’t work out of there is anything case nodes can do that attributes can’t. I’d love to hear from others where they have found case nodes particularly useful.
  9. In response to Chris Francovich's question to the old list - like Gonzalo I'd use a web system for communicating with students rather than trying to put everything about a methods course into NVIVO. My university has been using one for a while now, so students usually know how to use it. I've just finished my first semester of teaching qual method, and am firmly convinced that most learning occurred when students were 'doing' data collection and analysis, so that this principle should guide choices about communicaitng with students. Congratulations on the new list. Helen