patterns of missingness
I need to look more into how to address the absence of certain data points. If I recall correctly, 133 of 144 students in the 9th grade at my school participated. Reasons for absences vary, and I need to look into what to do about the missing nodes. Any suggestions?
network exchange theory
So this is a sub theory of network theory that focuses on concepts such as structural holes and brokers?
standard, basic statistical software (e.g., SPSS, Stata, or SAS) will not give correct estimates
So dose the basic statistical package in R work, If I am not mistaken, I think we also can't use statistical package to make estimates for relational data.
How well does a leader's perceived level of trust in his or her colleagues predict the number of alters to whom the person sent a collaboration tie in year 1, controlling for gender and the level at which the person works (district vs. school)? This question requires three vectors of independent variables (trust score, an indicator for gender, and an indicator for level) and one dependent variable vector (collaboration year 1 out-degree).
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?
The question, therefore, is whether school leaders prefer to collaborate with those with whom they have collaborated in the past or with those that they have turned to discuss confidential issues.
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??
The average of the simulated distribution of reciprocal ties is calculated and then compared to the value in the empirical (observed) network
Can ERGMs be used for nonreciprocal networks?
except, of course, ego-level network studies from which egos have been randomly drawn from some target population)
I am using ego-level networks for my final project, but how would I go about ensuring that the ego in question was chosen randomly?
artifacts
Maybe I missed something in our previous readings/videos, but can someone explain to me what is meant by the term "artifacts"?
Even the tools of predictive modeling are commonly applied to network data (e.g. correlation and regression)
Would running such tests require a need for assessing latent variables that emerge from network analysis? I will keep reading, but from what I know of SNA, it seems like you are only analyzing observable variables and it would be difficult to obtain a correlation from such unique variables. Am I way off here?
A model is a simplification or approximation of reality and hence will not reflect all of reality
when reading this, I don't know why but a question suddenly came into my mind, why do we need so complicated/fancy models in social science research, specifically, except core independent variables and outcome variables, why do we use covariate/ control variables in a given model. I had an insight from a professor's explanation: for natural science, most objects of study are homogeneous and scientists can have a good control of interference in lab environment with careful experimental design. However, in terms of social science phenomenons, they are so complicated and are impacted by so many factors, including which we already know, and also a lot of which we don't know yet, let alone the subjects of social science study are so unique and heterogeneous. So we have to use advanced model to get closer to understanding those complex phenomenons, and we have to try our best to control the covariates we already know to carefully test the real relationship between independent variables and dependent variable. In addition, because we can not know or measure all factors that will impact a certain complex phenomenon, this is one of the reasons that a model is a simplification or approximation of reality and hence will not reflect all of reality.
So, let's say you are interested in the number of collaborative exchanges that occur between teachers from two different grade levels in a complete network of teachers within one elementary school. First, you count the number of times these types of exchanges occur in the observed network and then permute these relational data lots and lots of times. With each permutation, you calculate the number of times this type of tie (collaborative exchanges between teachers from two different grade levels) occurs and compare this result to the original observed network. After this process of permuting and comparing, you can see how often the results of these permutations are the same as the original observed results: The more often the results of the permutations are the same as your observed data, the more likely that the pattern of exchanges in the observed data was due to chance. If, however, the results from the observed data are so unlikely when compared to the results of the permutations, then you are to conclude that your results are not the byproduct of chance. Therefore, this result would be considered statistically significant.
In terms of my project (looking at racial and gender-based biases in communication between undergarduate students in an online class), then I could use this same rationale and process in order to make generalizations to a broader population?
the difference between the mathematical and statistical approaches to social network analysis
What are the differences?
Egocentric analysis shifts the analytical lens onto a sole ego actor and concentrates on the local pattern of relations in which that ego is embedded as well as the types of resources to which those relations provide access.
Given the nature of my data (Forbes top companies), I think it would be appropriate to look at specific countries as the Ego and the job categories as the alters. Am I correct in assuming that the local pattern of relations would be how my selected county (the ego) is connected to other countries through job category?
The size of ego networks typically ranges from 0 to 6, since a name generator typically limits the number of alters that ego can list.
Is it not valualbe to analyze the size of ego networks for social media? This likely wouldn't limit the size of the networks to 6, right?
Table 7.2
There seem to be two types of questions here: relational questions and attribute questions. Might it be worth categorizing these questions about eliciting egocentric network data into two seperate categoies?
Table 7.1
I notice that Leader 35 has the smallest number of alters and the highest efficacy score. Did I read that correctly? I am sure this is too tiny of a data set to conclude anything, so I'm curious what would happen to that pattern with broader analysis.
here are several standard measures that can be calculated from egocentric network data, including size, strength, diversity, centrality, constraint, and brokerage.
I believe SIZE and DENSITY are very intuitive measures that enable readers to understand the characteristics of eco-centric networks, especially, when we try to compare the different eco-centric networks. My related question is "how we can statistically compare those values from different networks?". To expand the qustion, "how we can statistically test the differences among different networks?"
the density
I can't insert permutation and combination formula here, I will try to explain my confusion. I don't understand why use C 5 3 instead of C 5 2, although C 5 3 equals to C 5 2, I still think only using C 5 2 can make sense in this setting. Maybe I get something wrong, please correct me if so.
as they each capture a different view or purpose of centrality.
Based on what I have learnt, betweenness captures brokerage, and closeness captures reachability (it is about how far away the rest of the network is from a certain actor). However, we also talked about eigenvector centrality a few weeks ago (the thought of eigenvector centrality is that your importance is determined by your neighborhoods’ importance) , I was wondering why the author didn't introduce eigenvector centrality of egocentric network in this chapter. I think one actor's eigenvector can indicate this actor's potential value or importance, it can be very useful in some specific settings.
Finally, while most analyses of ego networks use simple graphs—binary data that simply indicate whether an undirected tie is present between two actors—it is possible to incorporate directed relations into ego network analysis.
Could one use bidrectionality of connections as a emasure of density?
two types of measures
These two types of measures sound like the traditional statistical methods analyzing independent variables(to get some descriptive results of central tendency and tendency of dispersion). However, I think egocentric data is relational data which violates independence assumption. What are the differences between relational data analysis and independent data analysis. I am not sure whether I propose my question appropriately and correctly, and I remember I had read some detailed information relating to this question before, but I am still confused about it. (According to these examples listing in this paragraph, maybe the author talks about attribute data, not relational data?)
By charting this process, you are able to identify whether there is a “core” group of actors at the center of the network, while others are on the periphery
To identify "core" actor, what is a difference between a K-core collapse and centrality. I understand that there are difference logics between them. However, considering the expected results, I am not sure how we differenciate those different techniques.
A weak component ignores the direction of a tie; strong components do not. Stated differently, strong components consist of nodes that are connected to one another via both directions along the path that connects them. Weak components consist of a set of nodes that are connected regardless of the direction of the ties.
Which means, what I understand that, there is no diffence between a strong and a weak component if a network is undirected. Is it right?
Though each procedure will provide you with a different take on the network's substructures, they are all based on the ways in which actors are interconnecte
What would result from analyzing the similarities and differences between the results from each of these methods of group analysis? Each is distinct, for good reason. Nevertheless, overlap may rightly be expected. Might there be a way to identify that overlap and then make mearning from the overlap?
By relaxing the criteria for group membership, the number of cliques has increased from 11 to 20, with Student 1, for example, being a member of 16 clique
This seems to be an unweildy number of cliques for a group of 17 people. How can this bottom-up analysis conclude similar findings as a simple ethnographic inquiry? Would more analysis be necessary? If so, what analysis?
The number of “blocks” to enter is up to you, and after working upward from two, it was decided that three was a meaningful number of blocks.
I trust that there was a theoretical framework for this choice. Nevertheless, I am at a loss to understand that framework. Why were three blocks chosen?
ou can also increase the value of n, but this is not advisable, as it seems odd for actors to be in the same clique if they are three steps from one another.
While I certainly understand this in a real-world sense, I do interact with friends-of-friends in social media, either directly or through group membership. As such, is it acceptable to increase n when doing research, if one's research questions/interest require?
By charting this process, you are able to identify whether there is a “core” group of actors at the center of the network, while others are on the periphery.
Does anyone know of a way to illustrate this in a GIF or similar? I guess maybe doing it manually could be workable, but it sure seems to me that automating the charting of this process would make for increased ease when discussing research online.
bottom-up approach starts first with the dyad and extends upward
I 'm confused to tell the difference between "top-down" approach and "bottom-up" approach, it seems like they begin with different unit, "top-down" approach starts first with the whole network and then try to find its subgroups, and "bottom-up" approach starts first with more micro unit like dyad. Do I understand correctly?
“keep things together.”
so is this directly correlated to structural cohesiveness?
Which structural properties of the complete network might be of interest to you?
before the participants got to know each other
If they were asked at a later point to similarly identify three friends, if reciprocity increases, would that leave more structural holes?
cohesive network with minimal clustering.
So if we have a large (size) cohesive network with minimal clustering, does it follow logically that there will be high transitivity and high reciprocity and low centralization?
teacher networks with high reciprocity would be positively associated with teachers’ perceptions of their schools’ innovative climate and trust
But who would be excluded? Or is it possible to imagine a network with high reciprocity, where there is high inclusion as well?
Table 5.1 T
Is this an acceptable format for an excel/csv file in R - or is it better to convert to weighted format, which is I think how LesMis data was presented?
structural holes (one actor connected to two others who, in turn, are not connected to each other)
From my understanding of this, structural holes differ from triads since triads generally indicate that all three actors are connected in some way. Is this correct?
The paradox, of course, is that the relative paucity of weak ties is what makes them “strong,” as they provide early access to diverse information
Is it fair to say that networks with high reciprocity and transitivity might mitigate some of the negative effects of clustering?
A high degree of reciprocity means that a network's actors choose one another. It could also mean that while some actors choose one another, they are not choosing others, which results in a high degree of clustering within the network
Hmmmm, high reciprocity is associated with network stability and social equality, but it might also be associated with clique-i-ness (clustering). What is the nature of a social network with high reciprocity and low clustering? What is the value of this type of network?
This was done simply for purposes of presentation; any manipulation of network data should have some theoretical or empirical basis.
So are 'purposes of presentation' valid basis for data manipulation? I am left yearning for more explanation.
the pattern of relations among the network's actors.
(Mental Note) The concept that I have to focus on...
CDi indicates individual actors’ centrality scores
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?
following measure of reciprocity
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?
If an individual opts out, this should mean that their name appears nowhere on the social network diagram (even if they are identified by another individual as being part of their social network). For instance, in the sample map, you can see that the map would be very disjointed if John and Holly opted out of the SNA.
Are we allowed to include nodes for John and Holly if they are identified by others, but without using their name? For example, we would refer to John as Anonymous 1 and Holly as Anonymous 2. Or would we have to exclude any data that involves them, regardless of anonymity?
This approach is either based on your knowledge about relations among a set of actors or relies on the actors themselves to nominate additional actors for inclusion.
This is interesting for my own likely research needs/goals in looking at how technology integration processes spread. That said, it might also be difficult to track in a truly large-scale project. For example, there isn't one "universal" place for teachers to go and find technology integration processes. Perhaps this will have to be bound by the PLC/PLN I have in place?
For example, Coleman claims that social capital is any “social-structural resource” that generates returns for someone in a specific action. Accordingly, social capital can be captured only by its effect; whether it is an investment depends on the [Page 221]return to a specific individual in a specific action (Lin, 2001b). Obviously, it is impossible to theorize social capital when its causes and effects are folded into a single function.
I have an issue with this as stated. Couldn't it be that the "return" of social capital, as it is considered here, is only looking for extrinsic motivation? What about intrinsic motivations, such as helping someone simply because it feels good or is seen as the "right" thing to do judged solely by one's moral compass?
he concept of social capital is said to address all of these situations (Kadushin, 2004).
I keep coming back to the "mind as rhizome" metaphor as we talk about social networks and the theories that surround it. It seems to me that we, as humans, often create things that are structurally or functionally similar to our own bodies (i.e., the ways that a computer works and it's internal hardware mimicking the structures of the brain). Perhaps the same is true with social capital and social media: are these actually extensions of our brain?
bridges, brokers
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?
For example, does a child benefit from having parents, teachers, neighborhood adults, and so on communicate with each other, or do these relations constrain that child?
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.
Network resources refer to those that the individual ego can access through direct or indirect with alters. Measures that capture network resources focus on the range, quality, variety, and/or composition of these resources. Contact resources are slightly different in that they refer to the valued resources possessed by alters (e.g., power) and applied in specific ego actions, such as bullying, choosing a college, or finding a teaching job.
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?
Brokerage opportunities are those in which an individual is located in a position in which he or she can broker the flow of information between people and control the tasks that bring different people together
I believe this is also referred to as a bridge. Is that accurate?
This idea that there is value in relations undermines the different ways in which the term social capital has been defined by a number of prominent theorists, who in the 1980s independently explored the concept in some detail (Lin, 2001a)
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?
Social network theorists and analysts go as far as to recognize that the inductive modeling strategies of social network analyses—that is, generating big ideas from small observations—are in opposition to the usual canonical assumptions of statistical methods, which prefer a deductive logic that operates from ideas to observations
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!