torsdag 8 oktober 2015

Pre Qualitative and case study research, or the Arab Spring was really interesting from a journalistic standpoint.

“What is, Becomes What is Right”: A Conceptual Framework of Newcomer Legitimacy for Online Discussion Communities by Silvia Elena Gallagher and Timothy Savage

The paper uses using what is referred to as Qualitative directed content analysis. Content analysis is used to find commonalities within communication, be it written or verbal. It can be used to describe phenomena within the context of that, for example, specific communication platform. A directed approach is based on previous theory and research, and is therefore used to confirm or add to preexisting theory.  In this paper they are using it to find what textual codes newcomers to online communities use and what makes these codings stand out. This was done through using literature to establish a base a coding framework on, and then by viewing four different online, active communities and analyzing posts made by newcomers to these communities. They then looked at what strategies these posters used to gain legitimacy within the community and managed to establish common themes and way to do this.

Talking about “codes” like this might make the method seem quantitative, but this is not the case. The authors did not look at amount of times a word was used or anything like that, instead they looked at the whole context of the post. They look for commonalities that might not be explicit, but can still be determined from the text and subtext of the post.

I think this a good approach for this kind of study. It requires going through a lot of written material at a rather fast pace, and so the “simplifying” if using a predetermined code system speeds up the process significantly. The use of four different communities helped avoid findings that would not be useable in a general context, as it meant that local behaviours were accounted for. They established a framework that can be used in future studies, but one that as they say is not a static tool and could be adapted to a new context.


Sourcing the Arab Spring: A Case Study of Andy Carvin’s Sources on Twitter During the Tunisian and Egyptian Revolutions by Alfred Hermida, Seth C. Lewis, Rodrigo Zamith

A case study is the process of looking at a specific case to develop and prove a hypothesis. When doing a case study you start out with a basic idea of the topic you want to research, and then go and find a person, a place, or an organization of interest to this topic. It differs from other research methods in that you don’t start out with a well developed hypothesis, but instead create one during the study itself, based on what you find. You also do data collection and analysis at similar times during the course of the study, instead of the traditional approach to first gather all the data and then analyze it all. A case study also usually looks a limited number of cases, although the amount varies. It can be a single case as in the paper I chose for this week, or it can be up to eight, as in the Harris and Sutton (1986) study referred to in the Eisenhardt paper.

The paper I chose seems to be, based on the criteria from Eisenhardt, to be fairly solid. It seems the case was chosen early on, since the research questions as presented in the paper are very specific to the single case in question. It is however possible that they have been reworked in the process of doing the study, I have no way of knowing that.

The fact that the study is done on past events makes it harder for me to analyze their data collection process, since the study is based on stuff that is readily available online, as archived twitter posts. However, I can discern that the process seems to have been made with consideration to avoid bias and errors. The literature comparisons are perhaps a little limited in scope, since it seems to focus on the one viewpoint about “old media” as the gatekeepers of information, and the paper wants to prove that this barrier is being torn down in this case. However there are numerous sources on relevant material, and it helps prove their hypothesis. The conclusion is clear and easily understandable, but the conclusion of the case is not decided by the researchers as much as by the context.
Upplagd av kl. Inga kommentarer:

måndag 5 oktober 2015

Comments on Theme 3





Upplagd av kl. Inga kommentarer:

Post Quantitative research, or there is no such thing as objectivity

I think it’s easy to believe that if you just collect enough quantitative data you can get the answer to any question, and to view that data as being objective. However, there will always be questions that can’t be answered through quantitative means, and data needs to be regarded as dependant upon a context. I talked about this in my pre-seminar text, and nothing we’ve learned during the week has changed my mind.

People who are used to natural science seems to have a real problem grasping issues that can’t be quantified. This is what is called scientism. Since some fields deals with questions that often have fairly easy answers (Did it change, how many did, what was the cause) it’s easy to apply the same logic on other field that doesn’t deal in those easy answers. While a biological study on frogs can have a conclusion of about a paragraph, an anthropological study might have an entire book as the answer, and even that is the short version. That was the main thing I learned during the seminar. In our group we spent a lot of time discussing when to use quantitative and qualitative methods, and it felt like that question dominated the whole seminar.

An example of the different questioned that can be answered was one I posed during the seminar, about tastes in food. If we, in the classroom, want to learn which food is the most popular, that is easily answered by a poll. Simple, quantitative data. However, if we want to know why that food is the most popular, we need to ask about people's opinions, As soon as you get into an area where opinions are interesting you need to use a qualitative method and analysis.

So how about the drum study? Ilias explained that they wanted to know if people moved differently, but you can’t ask that type of question and get a usable answer. The asking of the question to the subject would probably make them change their behaviour, and therefore affect the end result. However using quantitative data that can be gathered without the subject being aware of the data gathered removes the risk of unconscious influence. Here it wouldn’t have worked with qualitative data.

Challenges when working with quantitative research can lie in analyzing your data. It’s important to remember that data isn’t necessarily objective or true in all contexts, and to keep that in mind when you look at it.
Upplagd av kl. 6 kommentarer:

fredag 2 oktober 2015

Design Research, or the power of change(ing tiny details)

Réhman, S., Sun, J., Liu, L., & Li, H. (2008). Turn Your Mobile Into the Ball: Rendering Live Football Game Using Vibration

Media technologies is in my opinion best evaluated through user testing. Interface design and usability can be clinically tested in and evaluated in lab, but that controlled environment will not tell you anything about how the end user will view your product. Users see your work differently than you do, since they have no preconceived ideas about what will work (if you do your study correctly and manage not to influence them). As a researcher it’s easy to get stuck on thinking that the idea you have developed is the best way to achieve your goal, and it can be hard to see your own bias. It is always hard to predict how and what a person gets from an interaction until it’s done. This makes prototypes invaluable in our field of research, since you need something to evaluate and iterate upon. Especially when you are dealing with physical products there is no way to know how it will work until you have a prototype that can be tested. In the Réhman et al paper they show how vital it is to have a prototype, since before they developed one all their theory was just that, theory. Prototyping is the best way to convert theory into practice, and to get confirmation that the theory works in your context.

This also applies to a proof of concept prototype. If the goal of your research is to develop a product that will be operated by users, in the end you have to make something that can actually be used.  Before that point, no matter how many low fidelity prototypes you have, it is just speculation that your design will work as intended.

It might be tempting to work directly on your high fidelity prototype then, since that is in a way the end goal. However, that is mostly not the right way to go. This sort of research is dependent in iteration, and to iterate on a nearly finished product if often a waste of resources. Low fidelity prototypes allow for quicker changes with less effort. They are not a replacement for a proof of concept though, because as I said previously, you need to prove that your concept is valid and working as intended at some point.

A proof of concept is also in most cases the best way to communicate your results. User based research is hard to present in graphs if you don’t have a functional or nearly functional prototype
to test, since it’s by it’s very nature subjective and qualitative.


Finding design qualities in a tangible programming space - Fernaeus & Tholander
Differentiated Driving Range - Lundström

The empirical data in both of the concerning papers is test based. Both Fernaeus & Tholander and Lundström did their research by looking at existing technologies, and then developing their own and testing that in an environment similar to where it would be used in an actual case.

How then is this contributing to the pool of knowledge? I would argue that finding different and perhaps better ways of doing things is in itself a goal to be strived for, since otherwise we might still use the text based user interfaces of the computers early days. If iteration wasn’t considered to be a knowledge contribution the only things that would be counted would be to make new things. There would be no improvement on existing stuff, and that would be a shame.

There is a difference between designing for research and designing for production. When you design for production you need something that works. That is the end goal, and it is quite possible that when you find that something that achieves your goals you happily stop there. Designing for research on the other hand might mean iterating for the sake of iterating, to learn new things. You care about not just that it works, but how and why and why something else works better or worse. The end goal is not a well functioning design, although that might be a side goal, but to learn things about what you are studying.

All research is dependent on the context in which it has been made. While it might be easier to view a quantitative study on the mating habits of frogs as a more replicable study there must be space for the context to matter. Sure, design research is by it’s nature narrower in its applicable context than other types might be, but I think it’s a mistake to view it as therefore being always worth less.
Upplagd av kl. Inga kommentarer:

måndag 28 september 2015

Comments on theme 2

Upplagd av kl. Inga kommentarer:

Post Research and Theory, or what maketh the man.

This week felt different from the previous ones. It was to me less philosophical and a litte more practical, in that we what we read had to be applied straight away. It was still quite theoretical (pun intended), however.

I had a tough time grasping what theory is, but an easier time understanding what it is not. The text by Sutton & Staw was really helpful in telling me what it wasn’t, but the other text from MIS Quarterly was a little harder to grasp. What I chose to hang my understanding on was the statement “Theory is the question why”. So in that vein, theory would be trying to figure out why things happened. Say that you have observed a bird flying backwards. Facts about how it works, diagrams detailing how often it happens, none of these things is theory. Trying to figure out and explain why the bird does this, that would be theory.

At the seminar we discussed how it is possible to build a theory in the scientific sense when you can’t be sure that you will never be proven wrong. That, in a way, makes the question itself mute. You can never be sure to never be wrong, because you can only work with what you know and have at the time. It is Kantian, in a way, realizing that as he decided to work with and not through his senses, we have to work with the state of things as they are right now. You have to make assumptions, but make them sound and as good as you can. Build on those that came before, but be open to the idea of being proven wrong at some point, when someone makes a new discovery. You can only work within the paradigm, but don’t be afraid of it shifting.

During the lecture on monday we talked about theory in a historical context. In the western tradition seeing is the way to gather knowledge. That means that theory is taking a step back, to view things from the outside. Practice would then be the opposite, since it means actually doing things. Leif mentioned something I found quite interesting on that topic, and that was the different ways to view a subject. KTH, as a technical university, has a practical approach to things. We learn how to do math, and physics, and coding. Stockholm university, on the other hand, teach theory. Their students, even in subjects such as math, learn the theory behind things, and how to apply it. My friend from SU might not be able to solve a third degree equation as fast as I can, but she can do a proof of things I can’t explain. I think you need a combination of both to be a well rounded scientist, both theory and practice.

Although the lecture this week was very interesting (I found myself disagreeing about a lot of things, but in a very passionate way), I didn’t really see the connection between it and the seminar. The lecture talked about theory, sure, but on a different plane than what we saw in the literature and seminar. The most interesting thing to me was the discussion on what made a human. I’d like to formulate some of my thoughts about that, just briefly.

I believe, and stated during the lecture, that a human is made up of a human consciousness and a human body. A dead body is in a way still human, but it’s not a person because it lacks consciousness. A mind without a body, while very sci fi, like in an android might also not be considered fully human. I don’t really want to get more specific than that, because it feels like it leaves room for acceptable abnormalities. If you argue for something like language, how would you then view someone who is nonverbal (something that can occur in people on the autism spectrum, for example)? If lying is intrinsically human, is a bad liar then not a person? Someone mentioned being highly evolved, but that to me is meaningless. The mantis shrimp have 16 types of colour receptive cones is their eyes, compared to the human three. They can probably see colours we can’t even imagine. That seems highly evolved to me, but that doesn’t make them human.
Upplagd av kl. 6 kommentarer:

fredag 25 september 2015

Quantitative research, or how many people is enough?

Analyzing quantitative methodology

The paper I’m looking at is Social Networking Sites: Their Users and Social
Implications — A Longitudinal Study by Petter Bae Brandtzæg. The paper looks at how people are affected by the use of so called SNS’s (Social Networking Sites). It tries to examine whether users of social media are more or less antisocial and lonely than people who do not use social media. The paper uses a longitudinal study, that is, a study that looks at how the data changes over time, in conjunction with a survey that gathered quantitative data from the respondents.

What stood out to me when I read the paper was the drastic lessening of responses they got over time. From about 2000 participants in the beginning of the study it had dropped to 708 or about 35% of the original respondents (the study was conducted three waves over the course of three years). While I was aware that a drop off in participation is expected, I wasn’t expecting it to be quite so steep. This, to me, stresses the importance to make sure to start with a large research material. I would also suspect that this makes it important to control that the dropout haven’t shifted the demographics significantly.

I fail to see any major flaws in the methodology at work here. However, I’m actually unsure if this means that are none, or is just an indication of my own lack of comprehension of what the author is doing. I understand the data collection process, but I’m having trouble following what they’re doing to analyze the data. Similar to a discussion we had during last themes seminar on not finding theory in a paper because of being overwhelmed by data, I’m wondering if I’m blinded by the sheer amount of numbers and diagrams here. The one possible flaw I can see is in the selection of the participants. The author have chosen data from Norway, a very computer heavy and internet savvy country. This might make the conclusions not applicable to a less computer dominated society.


Reflections on Drumming in Immersive Virtual Reality

Do you move differently in someone else's body? That is the question the researchers are trying to find the answer to. They did this through motion recording of people drumming in a virtual reality setting, where the all white participants were given either a white or a black avatar.

What did I learn? That it is really difficult to read someone’s quantitative research methodology and understand what it is they did. Although I had no trouble grasping how the study was done or what the purpose was, I really didn’t understand how they had analyzed their results. I am not a statistician, which is good since the numbers and deviations just confused me. This is quite similar to the issues I had with the other paper, which I have detailed above.

This leads me to believe that one of the downsides of using quantitative methods is that it’s harder for someone not in the know to understand why the conclusions were drawn. Other limitations may include a difficulty in distinguishing anomalies from rare behaviour. This is not based on the essay, but instead just a speculation on my part. I would argue that when using quantitative methods, if you see something out of the ordinary it’s harder to know what to deal with. If you were to use a more qualitative approach you could always ask the subject, or something along those lines, but that is not possible in this scenario. Benefits would include things such as a, in a way, more neutral result. Quantitative methods leave less room for the researcher's bias to be made apparent, since you don’t really draw conclusions in a way that leaves much room for speculation. It is perhaps also in a way more giving for future research. If the data and analytical methods are clearly presented, it would be fairly simple for another research team to use your results as a accessible jumping off point.

So what about qualitative methods? Well, some things would be what I accused quantitative of not being. It is often easier for a layman or someone not as deep in the field to view and understand what you have done. It is able to present a fuller picture than just data points can, and leave more room for interpretation and adaptation. This is not necessarily a good thing, however, since the same argument could be made as a weakness.
Upplagd av kl. Inga kommentarer:
Prenumerera på: Inlägg (Atom)