I carried out some research recently that is relevant to your post. In my study I was interested in using regression and centrality measures along with a measure of knowledge to better understanding medicinal plant remedy knowledge transmission in a rural community in Mexico. The method is detailed in my paper titled “Use of network centrality measures to explain individual levels of herbal remedy cultural competence among the Yucatec Maya in Tabi, Mexico” coming out very soon in Field Methods. See the abstract below for more information.
My findings were that in-degree correlated with the competence scores (a measure of knowledge) with an r of 0.28, p<.01. When I ran the regression and included attribute variables, such as age and gender, along with centrality measures, such as in-degree and betweenness, the model explained 26% of the variation in knowledge scores. Interestingly, age trumped all other variables, including the network variables in its explanatory power. After doing some further analysis to explain this finding I determined that age and in degree are positively associated (r = .48, p < .01), which suggests that, as generally expected, the trend in Tabi is that older people are both the most knowledgeable and the most centrally located in the network.
Abstract for Field Methods paper
Common herbal remedy knowledge varies and is transmitted among individuals who are connected through a social network. Thus, social relationships have the potential to account for some of the variation in knowledge. Cultural consensus analysis (CCA) and social network analysis (SNA) were used together to study the association between intracultural variation in botanical remedy knowledge and social relationships in Tabi, Yucatan, Mexico. CCA, a theory of culture as agreement, was used to assess the competence of individuals in a domain of herbal remedies by measuring individual competence scores within that domain. There was a weak but positive association between these competence scores and network centrality scores. This association disappeared when age was included in the model. People in Tabi, who have higher competence in herbal remedies tend to be older and more centrally located in the herbal remedy inquiry network. The larger implication of the application of CCA and SNA for understanding the acquisition and transmission of cultural knowledge is also explored.
For more information about the study see chapter 5 of my dissertation available at: http://etd.fcla.edu/UF/UFE0041175/hopkins_a.pdf
Allison Hopkins, PhD
Assistant Professor, NYU Stern School of Business.
Research Affiliate, MIT Sloan School of Management.
Personal Webpage: http://pages.stern.nyu.edu/~saral SSRN Page: http://ssrn.com/author=110270 WIN Workshop: http://www.winworkshop.net Twitter: http://twitter.com/sinanaral
[log in to unmask]">***** To join INSNA, visit http://www.insna.org ***** Dear Members of the SocNet List,_____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message. _____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.
I am looking for any papers that have computed correlations or regressions of popular centrality network measures like closeness, degree, eigenvector or betweeness and the actual transmission of information.
So for example I might have
a) a network of friendships of peole and additionally
b) I might have a network that has an arc if a person has forwarded information from a person.
If I compute a network regression of the actors degree in the friendship network and the number of how often his information was forwarded I can see how useful the centrality measures actually are in predicting future information diffusion for an actor.
I have tried that a couple of times on online friendship networks and the regression usually ends up having an explanation from common centrality measures are around 30%. I am wondering if there are similar attempts out there in order to compare my values and my approach?
I think this question is highly relevant because it actually asks if the centrality measures we use every day to highlight certain actors in information diffusion are really usefull.
I can also think of computing different measures for brokers between two communities and the actual transmission or strong ties and their influence on the actual transmission.
St. Gallen, Switzerland
Tel +41 71 224 27 47 _____________________________________________________________________ SOCNET is a service of INSNA, the professional association for social network researchers (http://www.insna.org). To unsubscribe, send an email message to [log in to unmask] containing the line UNSUBSCRIBE SOCNET in the body of the message.