Cleaning and Preparing Data in UCINET

[@MrBaroter]

Thanks for the video! Really helps out

[@marcoslvsfilho]

Thanks for the videos, they've been helping a lot! Although, could you explain why, when I'm making a dataset, the colums are read as a node? For example, the file campattr.##d ran in NetDraw keeps making colums "gender" and "role" a node with relations and etc. Thank you!!!

[@JessKropczynski]

It sounds like you're loading attribute files (descriptors of nodes) as sociomatrix (relational datasets). This task might help you: open the spreadsheet editor and open the "campnet" dataset. That's the sociomatrix describing who knows whom. Then load "campattr". That's the attribute file describing the characteristics of each of the nodes in the sociomatrix. They're two different files and you will load them separately into NetDraw. When you open NetDraw, you'll see there's a few different folder icons. The folder with the "U" next to it and the folder by the "A" next to it toggle the default settings for you so that you can load your sociomatrix. Start by loading the sociomatrix, "campnet" by clicking the folder with the "U" next to it. You should see a sociogram with links and nodes in the shape of a wonky bow tie. Next, click the folder with the "A" next to it and load the "campattr" file. Nothing should seem to have happened. You can then click on the button with the color pallet and set colors by gender. Again, I suspect that you loaded the campattr file first with the settings toggled such that it was loaded as your relationships -- doing the above might help your understanding of formatting and loading the data.

[@marcoslvsfilho]

Yeah, it worked perfectly!!! I can see now the difference between one file and the other. I really appreciate it!! I just started using UCINET and it's been difficult. Anyway, your videos are helping A LOT. Please keep up the great job! Thanks

[@JessKropczynski]

Sounds like you're well on your way. Once you can get your data collected, formatted, and in the software, it gets easier... until you decide to take things to the next level ;)

[@marcoslvsfilho]

Do you mind one (or a few) more questions? In this video you used an affiliation matrix as an example (Davis.##d case). Do they recquire a more sophisticated technique of data treatment, such as normalizations and etc? Is it possible to work and manage them directly, without such? Other question is more general about such data. Davis, Gardner and Gardner (1941) used it to know whether these women shared proximity or were from the same community. Although, I collected data about nodes I know a priori that they actuate on the same organization. My intention was to measure whether they are close from one another and more than that, if they are truelly envolved with the organization - by the share of co-presences. Because of Davis et ali dataset, I've been having second thoughts about the validity of my data and the necessity of "cleaning" or "nomarlyzing" it.

[@JessKropczynski]

Sorry for the delay, it really depends on what you're doing. If you open the Davis dataset is is 2 modes: one mode is women and the other mode is events. Since most measures cannot be performed on 2-mode date as is, we typically deal with 2-mode data by converting it to a 1-mode affiliation matrix (how nodes along one mode are affiliated by their connections to the other mode). This creates a weighted matrix. Depending on the way you'll analyze the data, you may need to dichotomize or normalize the data... but you might not need to do that. It all depends on the analysis.

[@wolton20]

Hello. Thank you for the video. I wonder if you can help me with calculating centrality measures. I am aware of the different concepts/ measures and what they mean. I have some confusion regarding directed data. I have a directed network and was looking to maybe measure the betweeness centrality. Will this work with directed data or would I have to recode it to undirected? But in doing this, is it not potentially misleading as I can infer there are possible routes to particular nodes which in practice do not exist? Thanks.

[@JessKropczynski]

Hi Joanna, Are you familiar with in-degree and out-degree centrality? If you have a directed dataset and symmetrize it, you'll be loosing valuable information about in and out degree centrality. With UCINET, you will not need to recode your data before calculating multiple measures (including betweenness) of centrality. Just use Ctrl+M to bring up this function. For other questions about centrality, this may be helpful to you: www.faculty.ucr.edu/~hanneman/nettext/C10_Centrality.html

[@wolton20]

Thanks for your response. I have already used in-degree and out-degree centrality but its not really clear if you can use more than one centrality measure. I have used the UCINET function to calculate multiple centrality measures but I am unsure how to read them because some of the numbers in my mind don't tally up i.e they are too high. I also can't seem to find a guide to help interpet the numbers that I am seeing. Thanks for the link, I will re-visit the guide and see if I can decipher the numbers UCINET is generating

[@JessKropczynski]

The Hanneman and Riddle online text that I sent you should help you to interpret. It also sounds like you could use a good introductory text. If you're using UCINET, I recommend "Analyzing Social Networks" by the authors of the software. I teach a graduate methods course at Penn State (both online and in-residence) that utilizes that textbook. While you're troubleshooting, consider if you have a valued/weighted network (although UCINET will dichotomize some measures for you), and how you've input your data and whether it was formatted correctly. You can certainly utilize more than one measure of centrality in an analysis--but it all depends what you're analyzing to know which measures are right for answering your research questions. Best of luck.

[@coronarysulcus]

Hi Jess, thanks for this. Do you recommend any tools or have access to any power shell scripts that will create matrix data sets out of Microsoft Exchange logs?

[@JessKropczynski]

NodeXL used to have a plugin for this but I'm not sure if they've been maintaining it through the years. My strategy is typically to take a look at the way the data is formatted when aggregated from a service and reformatting in a way that works well with the analysis tool you're interested in using. There's a lot of simple ways to reformat data in UCINET, Excel, or R that can be find in KZbin tutorials and StackOverflow

[@fatemehtehranifar1265]

dear Jess,I want to draw a netdraw for my reasech but ucinet hasn;t netdraw and Ihave a problem do you help me? Ican't insert excel to ucinet

[@fatemehtehranifar1265]

dear jess I;m Iranian and my english isn;t fluence but a litle useful to me