This is what I was just posting about. :)

I saw a TEDTalk about this, but in HR versus higher ed. It discussed the issues with some AI/machine learning for choosing applicants, but that there could be unintended bias that is in the AI that weeds out groups based on gender, race, etc. I could easily see how something similar could be problematic in an algorithm for LA in higher ed.

I like this idea, but I wonder how many students would actually go through the trouble of doing this.

I agree that it's a good idea to have this before collecting data, but, depending on what you're doing, you might need to be more generic about the details. If you have an experimental study, for example, you don't want full knowledge of the study to affect behavior, which would taint results.

This is related to something that I was thinking, which is the difficulty with student buy-in from a logistical standpoint, where someone may not check the box for buy-in because they didn't read your email asking for it. So they may be OK with your using their data, but they haven't let you know that. So it is considered "sneaky" to do an opt-out versus an opt-in? Can you assume that they're OK with your using their data unless they submit something telling you not to, or is that considered "sneaky"?

This gets to the point I was making before. Demographics can matter. Looking at the LA of them is not always so unethical.

All good points. Of particular note is benefits and to whom and also definition of success.

This last point is an interesting one. I will be interested in the continued conversation of what sorts of data are acceptable. A few members of the class have expressed concern about demographic data being a part of this. Is there a case in which demographic data is ethical and helpful? I would argue yes.

Unless, of course, no IRB is needed and no one knows they're being studied. Which holds its own ethical issues, of course.

Which I'm already doing on my own through weekly data collection.

OK. So maybe this could be a useful package in the future. Except my studies will likely just stick to the semester, given I have no way of tracking students beyond that.

No kidding!

Another one to keep tabs on.

It occurs to me that I might've been able to use one of these instead of what I did (Google Forms) to create a survey for my students and then could've already had the data ready to export rather than needing to do the entry myself. Next time...

Another one to keep in mind for the future.

Wait--does this mean it takes in the data from the surveys which can then be exported into a visualization tool without the researcher needing to import data separately?? That sounds awesome!

This one has come up in a few studies I've read as well. Probably not as useful for my particular research.

Sounds like what I would've been using a few years ago. Maybe not as relevant anymore.

Would need to have a firmer grasp on stats to understand which elements are most applicable to me. p tests might be, but I think I recall from that week they seemed less likely something I'd be interested in, so this program may not be for me.

I don't have a large network, but am I working with more complex data. Not super complex, but I like the idea of the integration it describes.

I haven't seen this one come up as much, but it sounds more powerful than UCINET and still has the phrase "user friendly" in it. May be a better option to look into.

I like how often the word "easy" is used here. :)

This one comes up a lot in the studies I've read. I guess I was under the impression it was more basic (like the SNA equivalent of web 1.0) but could be wrong.

I had some difficulty in this area with my students because they're basically meeting minimum requirements for how many reply posts they do. I could ask questions around degree of my rockstar students or maybe look at repeated ties (suggesting the formation of a relationship of sorts rather than just a "reply to random person"). I need to see how many data shook out to start forming more of those "ah-ha!"s.

Getting better at this in R. I get the basic functions (more or less). Moving onto the fancier stuff! The Michael Marin videos on YouTube were really helpful for solidifying the basics, btw.

Mine ended up looking more like this (but with more vectors). Much better for R. :)

This was NOT a good route for me. I made the first few Excel sheets as adjacency matrices, and R was not happy. I will NOT be using this method in the future.

It took a long time to get my data all tidy, but now that I know what that looks like, it would be much easier the second time around.

I used something similar to this for coding SP and CP, which was much more meaningful for me than just binary ties. However, I'm running into issues in graphing these because SP has 3 valued elements to graph.

This was an easy question for me to answer. Directional all the way. I'm interested in both directions too, as a student's sense of belonging is potentially affected by both incoming and outgoing posts.

I collected all the data! Because I may be using this for multiple studies, I collected everything on a weekly basis (after the close the discussion forum for the week). It made grading much easier because I was able to code at the same time I was grading, and it helped me give more detailed feedback to my students in the grading.

More in next section.

Continuation from previous post...

I can see, however, that I either need to be doing the incredibly tedious thing of coding in one session, double checking coding, and/or having a more detailed rubric to follow. Because what I was applying was not totally clearcut, I can see where I may have (and probably did) code similar things differently when it came to levels of CP and SP. Not ideal, but also helpful for ensuring better consistency in future studies

I was beginning to think my answers to everything was going to be "Could've done that better...," but this was one area I did pretty well. It was more of a given, since I was teaching the class. In future studies, I could look at pairing with another instructor. That might make my coding of things like CP and SP more objective. I found myself sometimes downgrading one student whose posts were at a higher CP level because I wasn't used to them doing that level of CP in their work. I had to go back a few times to compare coding. Triangulation in the form of another coder would be helpful, but I think objectivity would help too. It's hard to do study not your own class though, given access issues.

Which is something Tasha pointed out to me once for my own study. It was a good point.

As Bodong knows all too well, my issue is in narrowing. I have way too many research questions currently because I'm trying to work too many elements together. I could do a better job of isolating elements rather than trying to find some grand way they all fit together. Just because they're related doesn't mean they belong in the same study.

When I first read this, I thought it was going to be about the 30K level stuff, which I am also still wrestling with a bit--not like stats. Different sort of brain workout.

This one might be particularly useful for my research. While my study is set up as being causal, it's less causal on the SNA elements themselves. I would need a more comprehensive understanding of the stats before really delving into more deductive elements.

I had done some review of the lit before taking this course, but I didn't understand SNA well enough to "get" the studies. Other studies I've pulled have happened after my own research began. If I were doing this process again, I would do a better job in this first step of reviewing before formulating. I would also be a little more inclusive--there are not as many studies for SNA and CoI together (so great niche to publish in, but for lit review, I should cast a wider net).

Ooh! This seems like it would be very useful for my divvided up SP and CP, which I have categorized at L, M, and H as well. So, "Do students' ties vary by their level of SP?" would be the question I would ask.

Oy. This (multiple linear regression) is much harder for me to wrap my brain around, but I'm going to give it a go (especially since I have so many variables and it might be useful for tying them together). Again, I have SP, CP, and discussion forum type. If I did the surveys, I'd also have perceived sense of belonging. How well does a student's perceived sense of belonging predict his/her level of CP in discussion posts, controlling for discussion forum type (for example, only looking at large group discussions)? That doesn't get the SP in there though, and may not actually be a great model of regression. :/ Help?

This is very similar to what my research is doing, where engagement is defined as SP and academic achievement as CP. :)

Again, if I were to try to apply this (p) to my own data, I could look at things like gender, race, etc, but it would make more sense to see if SP was tied in--in other words, take the same scenario from p1 (if a student respond to another (say, in a large group discussion), is that student more likely to respond back when higher levels of SP are involved? I think* I applied that correctly.

Again, attempting to apply to my own study. This one (p1)seems straightforward. If one student responds to another (in a large group discussion), is the other student more likely to reciprocate by responding to that student in future discussions?

I have next to no stats knowledge, so I'm going to try to extrapolate this out in regards to my own research to try to better understand it (hopefully!). In using my own research with SP (social presence) and CP (cognitive presence). I'm going to start with the varibles: levels of SP, levels of CP, discussion forum type. A question I have been asking is whether discussion forum type affects SP and/or CP. Modeling the question the same way as this one, it might be whether students are more likely to show higher CP with students they were in a programmatic small group discussion with versus just large group. I think this models this line of questioning, at least. This probably doesn't get the SNA part in. So, trying again... Are students more likely to respond to a student in a large group discussion that they formed a connection with in a programmatic small group discussion or a random small group discussion? This doesn't get the SP or CP working in there, but it gets the SNA. So part of what I'm studying. But, since I'm graphing SP as a weighted measure for SNA, maybe it could be whether students are more likely to demonstrate higher SP in an ensuing large-group discussion with students they were in small programmatic group discussion with in a previous module. Does that get all the parts working approporiately in a MR-QAP-procedure question??

Since my own study takes these measures weekly, I'm curious to see how the statistical aspect may help draw connections between weeks and predict what could happen in future weeks.

Although, I will say I think this is only true in a broad sense. For example, I think I may be able to predict from a couple weeks of SNA data who the central players in future weeks will be (rock star students tend to retain that status throughout), but it may not be able to predict who responds to whom. Though such a measure may not be all that useful anyway.

Coming back to this after reading the chapter, I'm having a difficult time deciding whether statistical approaches make sense for me. I am trying to make generalizations, but it's not around categories (like gender, race, etc.). Unless maybe my category is SP (social presence)? To make a prediction about students with high SP having high CP (cognitive presence)?

Mathematical is descriptive (centrality, degree, etc), whereas statistical is inferative (allowing to make a prediction about the network). A researcher would be interested in the former if they do not need to make generlatizations about their network--if they are just interested in their network for its own sake. If the researcher wanted to say something about the population in general based on their network, then they would want to consider integrating statistical approaches.

If your network differs from a typical random creation, then you can perhaps make a claim about unique properties.

Simulations make things a lot easier. In even a modestly sized network, there could be thousands of permutations. Having a tool to create randomizations makes the process much less cumbersome.

For my own study, this would be making a claim about the connection of SP to CP in certain discussion forum types to suggest that you can get the biggest bang for your SP and CP buck by adopting a certain discussion forum type.

Thinking about direction makes sense for my study. It could be interesting to compare an ego-centric network to look at one student (Student A) to see if Student A's out and in neighborhoods are the same or mostly overlap. From what I've seen in my data so far, this has only been the case when students are placed into small groups. When it's the discussion forum as a full class, there is significantly less overlap.

It could be interesting to compare students' sense of belonging/community on the basis of whether they have a strong in or out neighborhood. I could see students who have high out but low in as feeling disconnected, and students with high in but low out maybe not even paying attention/not really caring about sense of belonging. Methinks I need to add a survey to my research...

This may or may not make sense for my research. Since my network is small (19 students; 20 if you count me, the instructor), it's easy enough to look at the whole network (socio-centric) and then decide if it makes sense to zero in on particular students. Reasons I may want to use ego-centric: To study high-performing (vs low-performing) students' networks; to isolate students showing high cognitive presence (one of the aspects I'm researching) to see what their individual networks look like; to explore students with high social presence to see what their networks look like--does quality of SP go along with quantity of SP?

If I were administering surveys (which I do not believe I'll have time to do), then I might want to look at egos based on survey results, for example, to a ask like "Do you feel a sense of belonging in this online class?" to study how students' perceptions of belonging align with their actual quantity and quality of ties.

It may be interesting to apply this after I've had students in the same small groups for a couple weeks for their discussions and then return to a whole-class discussion (to see if the students respond more so to the students they had been in a group with previously).

I could see this being useful for looking at my own research. Not many of the students are interacting beyond the minimum requirements for responses (typically, 2), so I could use this to separate out those who are a bit more easily (so doing a 3K-core).

Centralization interests me for analyzing discussion forums--are there key players, and do these key players show higher degrees of cognitive presence?

Calculating for centralization by number of connections seems quite straightforward in R: centralization.degree

Foundational math question: Does this CDi mean all other scores added together, or is this calculation meant to be applied between the actor with the highest degree against every other actor? So, in the scenario with the 17 frat brothers, is calculating the centrality a matter of doing Freeman's formula once, or 16 times?

Is there a typo in this equation? I would think that instead of having Aij in all 4 positions, of the equation, that Aji would be in 2 of them (as that would indicate the relationship of j to i instead of just i to j). Am I thinking about this wrong? Why are all 4 of the equation factors with a focus on actor i to j?

This is interesting! Didn't realize cognitive dissonance had SNA roots.

I don't think I adequately understand the difference between these two. They are both about connecting separate networks, and I thought at first the terms were interchangeable, but they're used together here, so clearly they're not. What's the difference?

This is such an interesting question! My area is in college students, so I don't have to worry so much about parental involvement, but for those in K-12, the assumption that this sort of close communication is good. I wonder if it's like the stroke victims where having more weak ties among the network is a good thing. Though I can't fathom why that would be the case in this situation.

OK. Now I'm confused. Can anyone help clarify the difference between network and contact resources? It sounds like contact resources are an attribute of one's network resources--is that accurate?

This would relate to the talk we watched on stroke victims. In that, there was more constraint because of how close-knit the stroke victims who came in late were to their networks (presumably, families). If I understand it correctly, there would be few bridges in these networks because the nodes are already connected to each other, so there's not a "bridge" to a non-centralized group.

I believe this is also referred to as a bridge. Is that accurate?

I find it interesting that there isn't an agreed-upon definition. Is there one theorist's definition considered the most valid in educational SNA?

This has come up a couple times now with no definition (that I've seen), so I looked it up on dictionary.com: the measurement of attitudes of social acceptance or rejection through expressed preferences among members of a social grouping

I can see where the inductive may align more with qualitative whereas deductive may be more quantitative, but SNA still strikes me as much more quant than qual. I'm still new to research ideas, but it seems that to move SNA from quant to qual, you'd need to include more than just the mathematical modeling, such as having interviews to add some color to the SNA. Without those interviews, though (or something like that), it seems despite being more inductive, it would still be considered quant, not qual. As I said, though, I'm newer to this, so if someone wants to provide opposing reasoning, I'm very open to it!

By "defy," does McFarland mean to misbehave? If it's also about not completing work, I wonder how this same scenario would play out in the online realm.

I hadn't really thought of SNA in these terms, though it does make sense. This reminds me of our current political situation. I wonder what work SNA has done to analyze this. I'm sure I could find some things, but if anyone's run across something they'd like to share, I know I'd be interested.

This seems to somewhat reflect a similar evolution in education theory (not directly, but somewhat), tying into the newer theory of connectivism, as I see it. Though this doesn't necessary indicate a progression (maybe though?)

Can't help but think of the division of tasks within my own household. I find it interesting that stronger ties suggest more segregation between husband and wife. Maybe the thought is that when there are less strong ties in the family structure, they rely on each other more and hence work together more?

Diffusion--as I understand it--is tracking how a particular phenomenon moves through a social network. This reminds me of a TEDTalk on how to start a movement: https://www.youtube.com/watch?v=RXMnDG3QzxE

This makes an interesting point, one which is important to keep in mind when analyzing a network--you know only the bounds of that particular network and thus any hypothesized implications of, for example, peripheral members can only take you so far. Being able to fully analyze would mean access to information from other classes (in the context of education) and beyond. This already got more complex...