The aggregates on the Linkscape tool are computed only on the top 3k links. From the API you can get a deeper, more complete look.

To answer your question, take a look at the comparison reports (advanced) in Linkscape. There are tabs for key terms and keyphrases by anchor. These reports give the top 8 most popular terms and include a value passed metric.

Hope that’s of some use!

This is a really interesting post and I love getting down into research and trying to make sense of all the nitty gritty details that can effect the SERPs. This post really does give away some great data from your research.

One question (and maybe I have just forgotten something here as I have only just woken up and it’s pretty early right now!) – How do you find the ‘mR Passed by Relevant Anchor Text’ in Linkscape? Did you just average out the mR values assigned to each exact match anchor per ranking domain?

We met very briefly at Distilled/SEOmoz seminar in London, I have been enjoying seogadget since. Quality reading!

That said, Open Site Explorer (http://www.opensiteexplorer.org) has most of the data available in the Linkscape tool and more and supports download to spreadsheet. So you could pull those down and into excel and then do a neat pivot table.

To answer your question, I’m currently using Linkscape’s web interface, so we’re only looking at the most valuable inbound links. In both cases that is actually most of them – but a complete snapshot would be even more accurate.

I’ve tried to extract data via the api to MS excel via Data > From Other Sources > XML Data Import. I’d like to assign more time to this after SES as I’m sure it’s possible to set up the equivalent of a SQL style cube, perhaps with a VBA interface.

Always great to see what you have to say I like your points about the interplay between page and domain-level factors. And I like what you’re doing with anchor text. Are you getting that from the Linkscape tool or from the anchor-text API? The tool only aggregates data for the first few thousand links (which might be most of them in this case). But it would be really neat to see that stuff from the anchor-text API which aggregates across all links.

It would be great to see this on more SERPs with averages and error bars. But I don’t think anyone will argue with some of your conclusions:
* page level and domain level factors are important and interact for ranking
* relevant anchor text is valuable
* a weakness in one factor (like relevant anchor text) can be made up for in another (like overall links)

Neat!