For me, the most important good practice in qualitative research is looking for contradiction. If you do not look for contradiction, you end up creating a Paul is dead story. A conspiracy theory. And here it says, do not only look for evidence. For instance, look at this scatter plot. And there we see all these scatter points. And it certainly points at one direction. For instance, the more lectures you watch, the more you know about qualitative research. Now this is more or less what we expect. But some people watch many lectures, but do not get to know anything about qualitative research. Now for a qualitative researcher that would be a very interesting group because they do not follow the theory. The theory must be something like this, but in reality the regression line works more or less like this. And in qualitative research, this is the group we focus at, not only on this group. If you understand qualitative research pretty well after following all of this lecture, it's very good. But if you're not following at all you are really interesting for us, because something went wrong there. This is what you should do as a qualitative researcher, look for contradictory evidence, look for counter evidence. This is what you should do when you study something like the Paul is dead hoax. Look for counter evidence. Paul was in France. He wasn't dead. So we looked for deviant cases. And according to Perakyla, we are looking at three different types of deviant cases. The first is deviant cases that provide additional support. For instance, we've been observing outside, just like Len Laughlin did, at how people privatize public space. They create a bubble around them. Now, a deviant case would be someone who steps up to someone else and then starts talking. Len Laughlin calls them adventurers, explorers. And if we look at the reaction people give to this adventurer, to someone who starts talking all of a sudden. You are nestling nicely or reading your book nicely, and then all of a sudden someone starts talking to you. It shows that that behavior is deviant by how you deal with that person. At least that's what Laughlin says, deviant cases sell provide additional support. Look, people are a bit scared. If you're sitting in an empty train, and then someone comes and sits right next to you, in essence it's kind of weird. So we tend to flee or something. So that's a deviant case. That's the first type of deviant case. The second type of deviant case is a deviant cases that require modification of a theory. The most famous example of this is the PhD thesis of Emanuel Schegloff. And Schegloff was looking at openings in telephone calls, callers and called people. And the called people, they tended to answer first. They tended to talk first. So after analyzing about 500 cases he came to the conclusion, well, the called talks first and he or she is the one opening the conversation. And turn taking takes place. But then someone, or some case came across a deviant case, because this call did not answer properly and the caller started talking. Something weird, but it showed something for Emanuel Schegloff because what it showed was that This theory was actually not good enough. It wasn't the called who was talking first, it was a summon that was not verbalized but it was a telephone ring by the caller. And then the called had to react to that summon, and if something went wrong with the summon, as in that deviant case, it showed that his theory didn't fit. So he reversed his theory. Then the third deviant case is a deviant case that is simply exceptional, and we have to accept it. Either because we can explain it easily, because something else is going on or we cannot explain. But at least we have to describe it then. We have to account for this contradiction. Now, how do we look for contradicting evidence? We will discuss that later, but part of it is to do comprehensive data treatment, as David Silverman and Paul Ten Have call it. And what do you do in comprehensive data treatment is you fight anecdotalism. You fight cherry-picking. And how? You do not only pick the nitty gritty details and you describe those details now you also describe the deviant cases. And you show how this is either something like an average or generalizable to all your data or not. But you do not only pick the cool stories. So you deal with all data. And Ten Have suggests not to sample only the cases that fit your argument. Do not cherry pick. No, also try to deal with the cases that do not fit your argument. Because they show more depth. They show that you might be wrong and you have to adjust your theory or these are simply deviant cases.
