Whereas the majority of works and research about context-awareness in ubiquitous computing provide context models that make use of context features in a particular application, one of the main challenges these last years has been to come out with prospective standardization of context models. As for Information Retrieval, the lack of consensual Context Models represents the biggest issue. In this paper, we investigate the importance of good context modelling to overcome some of the issues surrounding a search task. Thus, after identifying those issues and listing and categorizing the modelling requirements, the objective of our research is to find correlations between the appreciations of context quality criteria taking into account the user dimension. Likewise, the results of a previous survey about search habits have been used such that many socio-demographic categories were considered and the Kendall's W evaluation performed together with the Friedman test provided very interesting results that encourage the feasibility of building large scale context models.