QDA Miner - Science topic

هذا قريب جدا من مفهوم الحقول الدلالية القائمة على مبدأ التقارب الدلالي... هذا المسلك مفيد في اختزال الخطاب في رواق عام مهيمن، وفي مسالك عامة مساعدة في بناء الدلالة.

Don't you run into the same problem in other CAQDAS programs? There's gotta be a way to do this in NVivo. If NVivo can't do it, I don't know if other programs can. You can certainly check the tutorials. See if the following helps:

Dear Ankur,

As you asked us to assume you are a layman, I'll do so (and maybe exaggerate some parts to make my points clearer).

In its essence, a QDA software (CAQDAS such as QDA Miner, MaxQDA, R's QDA package, Dedoose, nVivo, Atlas.ti etc.) is the digital equivalent of a highlighter and a blank paper. They help you carry out content analysis. Let it be quantitative or qualitative content analysis (and it is mostly quantitative), you basically 'count' how often a certain word or a concept is repeated. In the good old non-computer days, the process was literally done with printed spreadsheets. Coders would just count. Softwares help you keep your codebook, documents, and other variables together. They also help you count. That is pretty much it. CAQDAS do not necessarily move beyond this.

However, there are "smarter" software out there that do text-mining. R's tm package, or QDA Miner's sibling WordStat actually help you find the most repeated terms / concepts. They also help you find out concepts associated with each other. I am aware that there are even smarter software used by natural language processing people but those are way too smart for me. I have never used any.

Going back to your questions:

1- I have two answers for this one:

1a) You can use a CAQDAS to create your codebook, dump all your company reports, and start coding. Once you are done with coding, you can use case/document variables to create a model. For instance (I am just making things up), you might see that "mortgage" was referred as a reliable source of income only four times in 2011, six times in 2012, sixteen times in 2013. You can argue that there is a trend towards seeing mortgage as a reliable source of income (year would be your case variable here). Or maybe you will find that different parts of the company have different ideas about performance ( parts of the company would be your case variable here).

1b) If you can get your hands on one of those smarter software, you might do keyword and phrase frequency counts. So, without manual coding, you will know how many times a word / phrase was used in a given document.

2- Yes you can. It is even possible to do an automated sentiment analysis.

Here is my personal recommendation to you. As far as I can see, you have only five annual reports. I assume we are looking at 500-800 pages of text or so. I would recommend you to do an inductive manual coding, using any software you want or just an excel spreadsheet. Go through the texts, and mark everything you think signifies growth or degrowth sentiment. Once you are done, try to see whether they are recurring themes across years. Classify your marked text under categories based on these recurring themes. Use case variables (such as year, company division, function, whatever is interesting for you) to measure whether there are any significant difference.

If you are willing to invest in a software package, I'd recommend Provalis Research Suite (QDA Miner and Wordstat). If you have a lower budget, Dedoose might be a cheaper alternative. If you are familiar with the R environment, you can use rQDA (not really great but it works) and tm packages.

I hope this answer is helpful. Let me know if you have any other questions.

Efe

@Mani...I agree with you Mani, but I have already started the work and I can`t quit in between..Let's see how it gets configured.

@Ramakrushna..Thanks for the article.

Sarah,

String variables cannot be used as independent variables in Wordstat.  The trick is to convert them to nominal or ordinal variables, or into dates (if they contain a date)

QDA looks great, but if your inquiry is the use of one analytic application that could easily be used to analyze both qua and quan data inputs, I would rather suggest that you look into NVivo. The newer version of NVivo has new added features previous platforms that allows you to analyze both quantitative and qualitative and even geospatial datasets.