 Qualitative data analysis differs from quantitative data analysis. So when you do qualitative analysis, the key problem is that it is more of an artistic challenge than an engineering challenge. So in quantitative research when you want to make a causal claim after you have collected your data on your association on your contra variables, then you simply apply a regression analysis or one of its variants. Then the choice of which variant to apply and how to do it basically boils down to understanding what these different variants do and then being able to execute that on a computer. So it's basically an engineering problem. So you have a well-defined analysis problem, you have a set of tools and then you need to understand what those tools do and be able to use those tools then you pick the right tool and you apply the tool according to the instructions. So this is what engineers do. In qualitative data analysis on the hand, it's much more about making sense of the data. So you have text, video, photographs, maybe some numbers and you need to make sense of that and try to get to the causal processes that operate underline your data and then explain those processes in a way that is clear and convincing to your readers. This is much more of an artistic challenge and there are no ways that you can go follow stepwise and then arrive into some kind of good result. In contrast to quantitative research where generally no one is going to criticize you heavily if you just apply regression analysis and do it correctly. So how do we actually do qualitative analysis? There first we need to take a look at the research process. So this is the normal process according to singlet on a straight which is used for quantitative data analysis or quantitative studies. So you start with research topic, then you have a question that is a more specific thing than a topic. Then you prepare your research design, you design the study, then you and here you have two important things. You need to decide what to measure and you need to decide what you need to measure. So this is a variable, this is our cases, we need to think what variables and what cases, where do we get those cases. Then you collect the data, you process the data and then you analyze the data and interpret the result. So that's fairly straightforward. This is if you ever have read about software engineering, this is kind of like a waterfall process that just goes from up to down. Qualitative data analysis or qualitative research is quite different because in qualitative research we typically observe processes over time or if we do retrospective study we have a chance of going back to those companies and ask for more information or go back to people and ask for more information. Qualitative data analysis looks more like this. So we have still research topic, we have research questions and we have research design and then we have some initial idea of where do we go for data, do we study organizations, which organization we study and we also have some idea on what we are going to measure. So let's say that we do a qualitative data study based on interviews. To get started with an interview, for the first interview you need to know who you interview and then you need to have some questions that you ask from your interview. Then we go and collect data, but here things differ. So whereas in quantitative data analysis and quantitative projects the measurement and sampling are basically decided here and then they are set to stone so you can't really change them afterwards. So if you do a survey study for example after you have sent out the questionnaires you can't add more to those easily. Particularly if you do paper and pencil kind of surveys. In qualitative data analysis we typically have a rough idea of what we want to study but we don't have the specific theory in mind. So in qualitative data analysis we might have a rough idea that for example naming a woman as a CEO the course is company to become more profitable. So that would be like an initial hypothesis that we have and then we go and study what do the women do differently from men to cause the profitability difference and typically you need to iterate. So we collect some data, then we analyze the data, we come up with some initial theory on what might be going on. Then we go around, we realize that well we saw that women are actually socially more capable than men. Then we realize that we need more data about the social capability of the women CEOs and the male CEOs. Then we go back to the field and we collect more data. So we have this iteration of analysis and data collection. So typically when we have multiple cases let's say we have six cases in our multiple case study we start the analysis after the first case. So after the first interview we analyze the data, we start to think what could explain what this person tells me and then we refine the interview protocol, we may add more cases to the study and we iterate. So we go and we have more measures, we have more cases and we could start from for example four cases and then in the final study we could have eight cases. So where do we stop? Because you can always find more companies or more people or more whatever you're observing. We stop when we realize that the final case that we added to our study did not really give us any more information. So our idea of a theory did not update any more after adding a case. Then we know that we have obtained what is called theoretical saturation and then we finish the data collection and data analysis and then we write our report. So to understand qualitative data analysis you need to understand first that the research process differs. Whereas in quantitative data analysis you start with the data collection and then you proceed down to data analysis. You hardly ever go back to data collection. Sometimes you do but that's very common and then you work with what you have, you write your report. In qualitative data analysis the data analysis and the data collection they go hand by hand and your data analysis guides your future data collection efforts so that both the cases and the interview protocol or observation protocol are updated as you get more insights from the data that you analyze. So how do you then actually analyze the data? There are different ways to do that. Typically we pick one of the leading scholars for example Dennis Gioja, Kathleen Eisenhardt, and Langley and we follow what those do. All these approaches have something in common though and it is called qualitative coding. So when we have 500, let's say 500 pages of interview transcripts and 100 pages of field notes from our interviews and our observations from the field we can't publish that for two reasons. No one is going to read that, no one is going to make sense of that and the second reason is that it's typically confidence. So we need to summarize those 500 pages into some insights that our readers can then use or whoever is the consumer of our research results. And qualitative coding is the way how we do that. Understanding qualitative coding is perhaps easiest to do by looking at an example. So this is a random interview of a random CEO found from the internet and I will demonstrate the principles of qualitative coding and by just going through how I could code this particular small interview transcript. So this is, if I remember correctly, this is the old CEO of Caterpillar and we are going to code and extract meanings from what he says. So typically you go paragraph by paragraph or sentence by sentence and then you start to think what does this guy mean? So first of all the person says that this is a great company so we can code that as pride. So the person is proud of his company. Then he explains that the company has been around for a long time, so there's long history, 140 years. We don't need to know about the specifics. So we kind of like increase the level of abstraction by coding and we seek to extract meaning from these sentences. The next thing that we note is that there are bad times. Then we go on and we note that there is trust for the image that this company has and there is an American company. The person is proud of the American history of this company and then the company wants to grow internationally. So we go through the data and we mark text, then we extract meaning from that text. And after we have gone through this initial, we call this open coding to get the first order categories from our data. We start to think like okay so do these things have anything in common? So are there any second order categories? So for example can we abstract these first order codes? We could for example say that the heritage here, American heritage and long history, they could share something in common. So they could indicate that this company gains its legitimacy through this long history of operating in the American markets. Then we could also code abstract categories, for example that this company has a brand advantage because of trust for the image. So trust for the image is a more specific code than this more general that you have brand advantage. Once we have coded or combined these codes into more abstract categories, this is sometimes referred to axial coding. Then we start to think how do these categories relate and that's part of what we call theorizing. So we could for example say that we have this initial theory that if you have this long legitimacy through history, then that leads to brand advantage. So this company has a brand advantage because of their long history and then we could say that well these works in the context of international markets. So this is a bit of a silly example but that's a general idea. You extract meaning from text on these first order categories, then you have these more abstract second order categories and then you start to look at relationships. Of course because this is a qualitative study, we have to pay attention to two different things. One is that this is just an initial idea and then we have to iterate many, many times. If we have other cases that we study, if we have multiple case study, we have to check if the other cases support this interpretation or if this is something that is just specific to this case or if this is something that we just can't really support empirically. Another thing that we need to look at is explanation of this process, so how exactly would this legitimacy lead to brand advantage. Then we iterate many, many, many times. Nowadays this process is typically supported by computer software and the computer software has a couple of advantages over the older style of printing things out and then coding with pencil and notebook. First of all, you can automate something. So if you find that, for example, innovativeness is useful construct for your study, you can automatically search for innovation related terms from the data, then it also keeps a track record of how you infer different things. So you can look at your coding history and then if someone asks you how did you come up with this theory, then you can point to specific instances in your code book. And then finally, quite often we have in quality studies, the results are presented as tables and quotes or we support the results with tables and quotes that illustrate our data. Then qualitative data analysis software makes that easy to do and easier than if you were using just the old fashioned pen and notebook coding.