The power of assuming normality

Daphne R. Raban, Eyal Rabin

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This study focuses on the power law distributions found on the web and proposes a method to perform statistical inference on data from such distributions. Beyond describing the state of a community, the power law nature of social interactions can be used to explain some of the variance associated with social behavior. Inference based on data on interval or ratio scales rests on the assumtion that the data is normally distributed. To obtain normal distributions the power law data is logarithmically transformed and subsequently used in a regression model. Data retrieved from the Google Answers service is used as an example. The regression model suggests that participation in the Google Answers information market is catalyzed both by social and by economic incentives with the most influential incentive being tip, a form of socially-driven economic incentive. This type of analytical approach is seldom found in the internet research literature and is hereby recommended as a very useful analysis tool.

Original languageEnglish
Title of host publicationProceedings of the European and Mediterranean Conference on Information Systems, EMCIS 2007
Pages301-308
Number of pages8
StatePublished - 2007
Event4th European and Mediterranean Conference on Information Systems, EMCIS 2007 - Valencia, Spain
Duration: 24 Jun 200726 Jun 2007

Publication series

NameProceedings of the European and Mediterranean Conference on Information Systems, EMCIS 2007

Conference

Conference4th European and Mediterranean Conference on Information Systems, EMCIS 2007
Country/TerritorySpain
CityValencia
Period24/06/0726/06/07

Keywords

  • Incentives for participation
  • Power law distributions
  • Social networks
  • Statistical analysis

ASJC Scopus subject areas

  • Information Systems

Fingerprint

Dive into the research topics of 'The power of assuming normality'. Together they form a unique fingerprint.

Cite this