Wednesday, August 14, 2013

The Webometrics Methodology

Isidro F. Aguillo of Webometrics has kindly sent me a summary of the methodology:

- The ranking intends to measure global performance of the 
universities using the web only as a proxy. Web design is mostly 
irrelevant, web contents are key if the web policy intends to mirror 
all the university missions on the web.

- We have a MODEL for the weighting of the variables in the composite 
indicator. It is the traditional "impact factor" developed several 
decades ago in bibliometrics adapted to the web: A ratio 1:1 (50%:50%) 
between ACTIVITY and IMPACT.

- For measuring IMPACT (visibility?, impact?, quality?) there are 
three alternatives: prestige surveys (THE, QS), peers citations 
(Leiden, NTU, URAP) or link visibility (number of external inlinks or 
backlinks). We use this last option because in this way we acknowledge 
a larger diversity of activities and missions and (very important) by 
a huge amount of users, a truly global audience.

- Personally I have two important objectives with the ranking: First, 
I am a scholar (scholar.google.com/citations?user=SaCSbeoAAAAJ) who is 
paid by the Spanish government to make scientific research, so the 
ranking provides me with a lot of valuable data useful for analysis 
and papers. Second, I have a "political" agenda, that is supporting 
Open Access initiatives.

- So, for measuring ACTIVITY the key issue is considering the 
full-text documents, so Openness consists of the number of files in 
pdf, doc, ppt and similar formats.

- An important innovation in the 3 latest editions is the Excellence 
indicator that is not really web related but intends to acknowledge 
the research intensive institutions. The data is provided by Scimago 
and reflects the top 10% more cited papers in 21 disciplines.

2 comments:

  1. Really good blog for ranking. But i have one question if you can clarify. On Webometrics Site it says
    "Webometrics log-normalize the variables before combining according to a ratio 1:1 between activity/presence and visibility/impact groups of indicators."

    I am not clear about log normalization. Can you please write some thing about that.

    Yours Sincerely,

    Haroon Sharif

    ReplyDelete
  2. I have doubt on webometrics will increase the rankings. After reading post, I clarified my doubt.

    Web design company in hyderabad

    ReplyDelete