Dr Corrie Uys, Dr Pat Harpur and I are working on a manuscript that explores the research implications of differences in Qualitative data analysis software (QDAS) packages’ support for live Twitter data imports. This paper's software comparison contrasts the four prominent QDAS tools that support such imports, namely ATLAS.ti™, NVivo™, MAXQDA™ and QDA Miner™. We discuss key discrepancies in their use during the organisational phase of qualitative research and address related methodological issues.

Outside the paper's scope, our software comparison also uncovered several suggestions that developers of these QDAS tools might follow to improve the user experience for Twitter researchers:

Written mostly for MacBook Air users in Cape Town and NVivo for Mac 10 users anywhere.

After replacing my MacBook Air's hard-drive, the next project to support its longevity was battery replacement. Mine was lasting just two hours, a far cry from the initial nine. Apple recommends that battery replacement be done via an authorised service provider. Although online tutorials suggest that is perfectly feasible to do a MacBook Air battery replacement oneself, EveryMac has identified difficulties in sourcing batteries of sufficient quality for post-2009 models.

Since I required speedy and safe replacement in a later model, I followed Apple's advice: Digicape Cape Town were prompt to respond with a quote after I provided them my MacBook Air's serial number. A battery (from Europe) plus installation came to just over R 3,100, including VAT. Although not cheap, the peace-of-mind this provided seemed well worth such expense: Digicape did an MRI/Apple system diagnostic to confirm the fault and to validate the order. After confirming the quote, it took just three days for the battery to arrive. I then dropped off my laptop and the replacement took just over an hour.

A surprising benefit of this replacement was fixing an NVivo for Mac 10 software issue: I was using this qualitative research software to code 25 journalist transcriptions for South Africa's contribution to the 'Journalistic ethics and work practices in conflict societies’ MeCoDEM project teamwork. After installing NVivo 10.2 and running the software, it showed the error message: 'Your NVivo license has been cancelled', then 'Your NVivo license has expired'. After QSR Support fixed this via a remote help session, I was surprised to experience the same error a few days later. The options under the licensing menu were all low-lighted, so there was no way to enter my institution's registration key to authorise the software. With an NVivo project file submission deadline looming, this repeat problem was a potential showstopper...

In messaging a QSR technician during the second fix, he recommended that I get my laptop's CMOS battery checked. When the laptop runs out of battery, CMOS behaves as a backup of the system's date and time. If the CMOS is faulty or not working the time gets set to default which was likely to cause the expiration and cancellation of one's license. Post-battery replacement, I am pleased to say that I have not experienced this problem again, so I'm hoping his explanation is optimal!

Written for NVivo 9's developers, QSR International, and their future NVivo users.

In 2011, I wrote Six key changes I'd like to see in NVivo 10. As NVivo 10's development nears completion, here's a further nine improvements for QSR International to consider implementing:

7. Provide a program execution status
It is useful for the user to know whether NVivo is initializing, running or shutting down. If the software seems to be unresponsive, seeing a status message, like: 'initializing', 'running', 'paused' or 'shutting down' would be helpful to the user and a better alternative than using Windows Task Manager to double-check.

8. Show closing down status messaging
Further, if NVivo is slow in closing, it would be useful to show how it is progressing. This is very important; if the user thinks that it has stalled, he or she may "force quit" the program when it prevents Windows shut down. As failing to close NVivo down properly can corrupt the project file, thereby costing the user much more time, the value of accurate status messaging (and even warning the user 'do not force quit') should not be underestimated.

9. Provide better communication when there are errors
Just as Microsoft impresses with problem messaging linked to solutions via its Solutions Centre, so too should NVivo consider linking its error messages to online help. For example when I was trying to fix an NVivo installation,  the message "Database component did not initialise" could have been made far more helpful by linking straight to an FAQ page with possible solutions.

Nvivo error message: 'Database component did not initialise' (10 April, 2012)

Given that NVivo users are qualitative researchers, it is unlikely that many of them have experience troubleshooting software. Providing us with more thorough and helpful error messaging will provide an improved user experience just where we experience most difficulty.

10. Report on third-party interference
I have noticed that NVivo's timebar may be paused (at 67%) for several minutes when an important Windows 7 process runs (such as deleting a backup snapshot, running a backup, et al.).

NVivo stalled at 67 percent progress (6 June 2012)

It would be helpful if the user could be shown that third-party software has interrupted NVivo, ideally in the software itself. Even a notification from Windows OS would be helpful.

11. Offer a back-up file option
Just as Apple's OS X Lion introduced Versions to protect users' files, so should NVivo consider adding a 'backup' option in addition to 'auto-save'. For users whose files became corrupted and had failed to create a back-up, this prompter could be very beneficial. 

12. Provide a video rewind shortcut
For researchers working with many video interviews, a rewind F button shortcut would be great. I recently started work on a PC laptop and simply having the F7 (play/pause) shortcut available is a real timesaver {after moving from Mac, where I did not have this option}.

13. Allow the user to set a volume limit
The user should be given an option that sets a limit to NVivo's volume. In working with listening to mobile phone videos of varied audio-quality through headphones, this would be useful to protect my hearing as I move from soft, low-quality files to very loud, high-quality ones and forget to change volume.

14. Offer a pseudonym generator
As qualitative research invariably involves the ethical requirement to protect the privacy of one's subjects, it would be a nifty value-add if NVivo added a pseudonym generator.

Written for those struggling to install the embedded SQL software in an NVivo 9 installation.

NVivo 9 installer "File is corrupt." error message. 29 April 2012.

NVivo 9 standalone local connection error message. 29 April 2012.

Written for researchers interested in the ICT Access and Use Project's coding indices' development.

Three months after making the initial coding indices for the fourth phase of the ICT Access and Use project publically available, I have just updated indices for "Past -", "Current -" and "Future ICT Use" to reflect the subsequent additions to our NVivo 9 project file's nodes (as illustrated below). These coding additions were necessary to reflect new insights from our research subjects' first and second interviews. They are described below under their relevant index:

Personal ICT Use coding index changes

Scribbled changes to the previous coding index for Past ICT Use 1 (30 March, 2012)

Nodes were added for "Other ICT exposure 1.15" and "Self-taught 1.16" as not every student received formal schooling in ICT: for an example of 1.15; A Fort Hare university student had never had professional or formal classes in computing. For 1.16, the only training a Rhodes University student had was teaching himself; he fixed his family's computers himself. And a University of the Free State student "used to touch the home computer" to teach herself.

I thought it was a bit old school for freshers in 2011, but was suprised when MS Dos was mentioned as a first operating system! So, the node "MS Dos 1.26.4" was added.

One student accessed computers outside school and home at secondary school (hence "Internet cafe 1.43"), two students volunteered information on why they got mobile phones ("Why get a mobile phone 1.21.3"), another spoke about one's first computer tablet ("First tablet 1.28").

Current ICT Use coding index changes

Marked changes to the initial coding index for Current ICT Use 2 (30 March, 2012)

The largest index saw the most revisions. Given the research project's interest in the use of mobile phones, the most important change was adding nodes that would allow the research team to track the most common platforms used to access online services and apps ("Platforms used in access 2.26" included "Computer 2.26.1", "Laptop 2.26.1.1", "Desktop 2.26.1.2", "Mobile 2.26.2" and "Tablet 2.26.3").

I also added the online services ("Webmail 2.21.13", "Music downloads 2.21.14",  "Whatsapp 2.21.15", "Online forums 2.21.16", "YouTube 2.21.17", "Snap2 2.21.18", "Shazaam 2.21.19", "Online radio stream 2.21.20") and mobile phone apps ("Cognician 2.22.10" and "ToGo 2.22.11")recently mentioned by students. I also added the node "Does not use apps on mobile 2.22.9" to cover those students who explicitly stated that they did not download and/or use mobile apps.

Given the prevalence of iPods, I had to add "Music players 2.15.4" under the category "Other ICT owned"! Some students mentioned that they had "Access to ICT at home 2.17", while "Multitasking 2.23.9" was added as a theme that kept cropping up in students' use, for example of social media in the background as they worked on assignments.

The line between the use of social media for social and academic use is becoming blurred, as students report using social networking tools for both formal and informal academic work. This seems largely to be due to social media's affordances for ease of communication, efficient creation of academic groups and even saving money on MXit. This was reflected in student interviews and nodes for "Status update 2.24.10", "Messaging 2.24.11" and "Friending 2.24.12" were added, which could be applied to tracking affordances typical of social media.

Some first-year students also volunteered a fair amount of information on their preferences for response times ("Availability and feedback speed 2.25.11"), what they preferred to access services on ("Service preference 2.25.12"), how they felt about losing network access ("Mobile network access 2.25.13") and whether they experienced any limitations from their cellular provider ("Service limits 2.25.14").

Several students described having problems when using photocopying equipment for the first time ("Photocopies 2.31.7") and the importance of formal provision of either analogue and/or digital course material ("Textbooks and course reader 2.31.6").

Interviews revealed a wider range of actions with university software than initially identified, so the following nodes were added: "Group practicals 2.33.10", "Lecture notes 2.33.11", "Catch up missed classes 2.33.12", "Read announcements 2.33.13" and "Question and answer 2.33.14". And the following nodes were added under phases and types of academic use; "Frequency and duration of use 2.35.7", "Desired use 2.35.8" and "Self taught 2.35.9".

Future ICT Use coding index

Changes drawn on the coding index for Future ICT Use 3 (30 March, 2012)

Due to having recently upgraded mobile, laptop or other ICT technology and/or experiencing financial constraints, some students indicated that they did not want an upgrade ("No new mobile phone upgrade desired 3.11.2", "No new laptop upgrade desired 3.12.2" and "No other ICT desired 3.13.2").

Some students spoke about their ambitions at university ("University ambitions 3.24"), as well as the services they wanted to be on ("Academic services 3.31.1" and "Personal services 3.31.2"). Another University of Cape Town student mentioned his community involvement and promotion of ICT, so the node "Community involvement 3.4" was added to reflect this.

Lastly, a node "Play at Conferences 4" has been added to track the most interesting segments of student interviews. These are intended to be exported and formatted for playing at conferences by the project's research lead, Laura Czerniewicz, or principal investigator, Cheryl Brown.

2. If you work via a network, you can also expect big problems if you shut down your computer without closing NVivo properly.

This post was written for researchers interested in the background to the fourth phase of the ICT Access and Use qualitative research project's coding process.

The fourth phase of the IDRC-funded Centre for Educational Technology's ICT Access and Use project uses digital ethnographies to understand how twenty six, first-year students at four South African Universities used Information Communication Technology (ICT) for study and leisure purposes last year.

This research phase saw four researchers at the universities of Cape Town, Rhodes, Orange Free State and Fort Hare prepare eight sets data:

1. A series of interview videos between the university's researcher and his or her subjects;
2. Videos of focus groups;
3. Videos of ICT use at home;
4. Videos of formal and informal mobile phone video use; 
5. Videos of social media and internet use;
6. Videos University software use (such as learning management systems)
7. Screengrabs of Facebook use;
8. and documents of the researchers' reflections.

NVivo 9 software has been used to import these media files for coding and qualitative analysis. However, before either of these could start, Cheryl Brown, Laura Czerniewicz, Kelsey Wiens and I worked at preparing  classifications and a coding index that could be queried for most of the project's research questions, whilst being robust enough to answer any new questions that might arise.

Preparing this coding followed these eight steps:

1 Kelsey and I reviewed the project's documents and transcribed key points from student interviews;

2 I illustrated these points on two large cyan posters with yellow stickies (these were very useful for re-grouping concepts on the board);

ICT Access and Use phase four coding poster (9 January, 2012).

3 These points were reviewed internally and presented externally to the universities' researchers and their most engaged students;

4 Kelsey and I separated the points that were to be used for classification or coding;

5 Kelsey developed a numeric index in creating the Google document: "Past ICT Use 1". I followed this index in developing "Current ICT Use 2" and "Future ICT Use 3";

6 We reviewed these documents internally and revised them;

7 I then added these codes to the NVivo research project file and am now using them to code student interviews;

8 As I apply these codes using node shortcuts, I am also updating the "ICT Use" documents with new codes to describe ideas that may prove useful for querying later.

Below is an example of how these codings have been added to a video in the research project file:

Screengrab of coding for Ace's first interview in NVivo 9. (9 January, 2012).

I trust this post enables you understand the coding process we followed?

If you have a question, suggestion or other feedback, please type it as a comment below.

Written for users of video analysis in NVivo 9 qualitative research software. Attention: Technorati blog aggregator. The code GCJXS6STZCJQ confirms this blog as mine :) !

With your video file in the right format, size and descriptively named, it is ready to import as either external or internal source into NVivo 9. Using the analogy of an Adobe Illustrator's use in graphic design, one can either choose to embed files internally within the file (like a poster graphic with fonts and images saved in the .ai file itself) or linking to them externally (the poster file is smaller as it links to the files outside itself).

QSR encourages users to only import the files that are core to their research, since importing source internally increases file size and, if very large, may impact on NVivo's responsiveness. An important limitation of importing video files as external source, however, is that it does not afford you the same opportunities to code material that internal files do. See the screengrabs below:

Screengrab of an external source video in NVivo 9

Screengrab of an internal source video in NVivo 9

Our research project has 26 subjects and for each at least two video interviews and two Day Experience Media (DEM) videos. We could try to import all 104 videos internally: at a conservative estimate of an average file size at 50 MB for interviews and 30 MB for DEM, that's 
2 600 MB for interviews and 1 560 MB for DEM. This could total 4,160 GB, a big file. NVivo's Offline Help says a standalone NVivo file has a size limit of 4GB, so we may need to reduce the low settings used in our optimized encoding process to ensure that internal files are well within this limit!

In starting the initial coding, we will focus on four subjects (Ace, Edmore, Thabang & Khanya): all four were hosted at a digital storytelling workshop at UCT recently thanks to their comprehensive contribution to the ICT Access and Use research project. I will use their thoroughly documented examples to initially test file imports:

After consolidating all their first and second interview files to a shared intranet folder, I have imported these videos as internal source (Adoné will also be providing a transcribed interview shortly). The research team can now begin to code these internal videos, while the research leader can test the import of codes and classification attributes from the researchers' NVivo master copy files back into her project master.

Written for research leaders to copy-and-paste for briefing their field researchers.

If your NVivo project involves working with a team of data-gathering researchers, I would recommend you email them the following:

Written for current and prospective users of qualitative research software at the University of Cape Town.

My blog's NVivo posts aim to provide an example of the use of a qualitative data analysis software (QDAS)* in a South African (or developing world) academic context. Hopefully by posting on the experiences of using NVivo 9 in completing a PhD and working for the ICT Access and Use project, I can stimulate much needed discussion about how students and staff could (or should) be better supported in their QDAS-use.

There is considerable scope at the University of Cape Town (UCT) to improve its support for student and staff QDAS-use: currently, UCT does not do even one of the four key activities listed by David & Jacobson (2008) for QDAS institutional success as exemplified by the University of Massachusetts Lowell! 

Based on my experiences at UCT, the status of these key activities are:

1. Software is not readily available - UCT (and its Humanities Faculty) does not have an NVivo site license;
2. Training is hard to access and costly- my NVivo workshop attendance was self-funded;
3. There are no user-groups - there is no user group at UCT, such as a "Qualitative Research Software Users Network" for students and staff;
4. There are no forums for open discussion - UCT has no fora dedicated to fostering discussions on technology and qualitative analysis.

As a result, it is unsuprising that there is; limited decision making support in choosing the most appropriate QDAS software to use, little or no materials on QDAS in the library and no readily accessible online guides by UCT researchers on their QDAS-use.

This does present opportunities for UCT's staff and/or students to take the lead with:

1. Establishing a UCT QDAS Users group;
2. Setting up a regular QDAS forum;
3. Documenting how may researchers are, and intend to, use QDAS software;
4. Using usage statistics to justify:
5. - investment in training staff and students to teach about QDAS and its use;

1. - b
2. uying a site license for UCT.

If you are interested in working with me on these actions, please get in touch or add your feedback in the comment box below. 

* Also more commonly known as Computer Assisted Qualitative Data Analysis Software (CAQDAS).

Davidson, Judith and Jacobs, Cynthia. The Implications of Qualitative Research Software for Doctoral Work: Considering the Individual and Institutional Contexts [online]. Qualitative Research Journal, Vol. 8, No. 2, 2008: doi: 10.3316/QRJ0802072.
Availability:<http://search.informit.com.au/documentSummary;dn=425496252675286;res=IELHSS> ISSN: 1443-9883. [cited 05 Dec 11].

I've just learnt that a step and a process described in an earlier post are problematic: in testing file imports into NVivo 9, I learnt that (1) the MP4 format is optimal for sizing, but heavily reduced files are not importable. Also, (2) NVivo 9 does support importing files bigger than 40MB as internal source (contrary to the software's help file advice, ahem!):

Written for QSR International and NVivo 9 users on Apple Macs and NVivo 10's software developers.

Written for researchers importing video into NVivo 9.

• The choice of a first name protects the research subject's privacy; whilst not over-complicating the researcher's codings for subjects.
• Researchers conducted several interviews with each subject and these need to be distinguishable.
• Wherever possible, the date of interview should be added to ease citation for research articles.

Written for Apple iPhone, QuickVoice recorder and NVivo 9-using researchers.

1. After downloading Switch, install and launch it. 
2. Make sure your internet connection is active.
3. Once you've selected one or more .caf files, specified the .mp3 conversion and pressed "convert", Switch will automatically update itself with the code it needs to do this type of conversion.
4. The converted .mp3 will be exported to the folder you specified.