[Rsiena-help] Interpretation of the outdegree activity effect

[Tom Snijders]

Dear Daniel Stefan,

These type of questions should rather be addressed to the User Group for Siena and StOCNET,
http://groups.yahoo.com/groups/stocnet/
This list is meant especially for programming (coding) issues.

In your interpretation of this effect, you should take in the understanding that the outdegree effect itself is also included, and the parameters will be estimated such that the balance between creation and termination of ties agrees with the data. Taking a given function and then adding a positive coefficient multiplied by a quadratic function of the outdegree, (and note that the added quadratic function will because of the estimation be centered at the value where the balance occurs) imply that for current low oudegrees, the push to lower values will be relatively amplified, while for high outdegrees, the push to higher values will be relatively amplified.

Best regards,
Tom


================================================================
Tom A.B. Snijders
Professor of Statistics and Methodology
Dept. of Sociology, University of Groningen
Emeritus Fellow, Nuffield College, University of Oxford
Associate Member, Dept. of Statistics, University of Oxford
http://www.stats.ox.ac.uk/~snijders/

From: rsiena-help-bounces at lists.r-forge.r-project.org [mailto:rsiena-help-bounces at lists.r-forge.r-project.org] On Behalf Of Daniel Stefan Hain
Sent: 16 September 2015 15:05
To: rsiena-help at lists.r-forge.r-project.org
Subject: [Rsiena-help] Interpretation of the outdegree activity effect

Hello dear group,

I have a question regarding the outdegree activity effect in directed networks. The manual describes it as:

The out-degree activity effect (with or without 'sqrt') reflects tendencies to actors with high out-degrees to send out extra outgoing ties 'because' of their high current out-degrees. This also leads to dispersion in out-degrees of the actors.

But, as I see in the formula, it is just the squared amount of an ego's outdegrees. Wouldn't that rather test for an "inverse U-shape" effect of outdegrees, in a way that their marginal effect follows a curvelinear function, where they first matter less, then more, then less again?

In the same manner, how could the squareroot version of this effect, which is basically the squareroot of an egos outdegrees?

Thank you very much in advance helping me solving my confusion.

Best wishes

Daniel S. Hain

Research Assistant, M.Sc. Economics, M.Sc. International Business, Dipl.-Wirt.-Ing.
Department of Business and Management | IKE | DRUID | EIS
T: (+45) 9940 2724  | Email: dsh at business.aau.dk<mailto:dsh at business.aau.dk>
Web: www.ike.aau.dk<http://www.ike.aau.dk/> | www.eis-all.dk<http://www.eis-all.dk/> | www.innoresource.org<http://www.innoresource.org/>
Aalborg University | Fibigerstræde 11, Office 97b | 9220 Aalborg, Denmark

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rsiena-help/attachments/20150916/8c6902b8/attachment.html>