***** To join INSNA, visit http://www.insna.org ***** Hello,

I am trying to sort out the implications of sampling from a known set of ties in link-tracing designs with an unknown total node-set and would appreciate help or related references.

For example, a web-crawler might be designed to sample 10% of the links on a given webpage. The seed nodes are known and the nodes that the crawler could sample from at each step are known (and if/when those nodes also appear in subsequent sampled pages' links), but the total population of nodes is not known.

Is it possible to draw legitimate conclusions on network properties from such a data set, perhaps by using the sampling probabilities at each stage of the design in a weighting system? Or does the nature of the non-random selection on the dependent variable preclude such analysis?

Work I have found on link-tracing designs in ERGM (Pattison et al 2012, Handcock and Gile 2010, Koskinen et al 2013) assumes that sampling takes place on nodes but not ties, e.g. that link-tracing designs exhaustively sample the ties from given node-sets. Respondent Driven Sampling uses link tracing without the assumption of exhaustive sampling, but with non-sampled ties unknown (and other restrictions that limit the ability to draw conclusions about the network itself).

The end goals of this question are twofold:
(1) To collect multi-wave online link-tracing data for network analysis with less strain on (the collector's) computer resources and (the pages of interest's) server resources
(2) I'm also hoping that if it were simpler to collect valid network data with greater depth (number of waves) without producing as intimidatingly large data sets, it might encourage more social scientists without extensive computer science or mathematical backgrounds to pursue internet-originated structural data that is relevant to their research questions

Thanks for any help you have to offer.

Nathaniel Porter
Department of Sociology, Penn State University
Research Associate, The Association of Religion Data Archives
_____________________________________________________________________ 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.