Following a methodological approach similar to Le Guel et al. , the analysis of the decision to access the Internet and its uses are carried out here using discrete probit type choice models (Le Guel et al., 2005). These models of current use in economics based on survey data make it possible to identify the determinants of individual choices (McFadden, 2001).
The variables explained
12A first modeling focuses on the decision of the individual to access the Internet. On this point, it is important to emphasize that the majority of the work on the developed countries opts for the study of the decision to adopt Internet which they assimilate that of having a connection at home.
But although the study of adoption would have been probably very interesting, knowing that less than 2% of our respondents have a home computer and that over 97% of Internet users among them regularly access the Internet outside their home. We will focus here on the decision whether or not to access this technology, this decision being simply indicated by having already had to use the Internet, regardless of the place of connection. As a result,
13 Second modeling focuses on the choice of services to consume the Internet, in other words on the choice of uses that will be made of them. In the questionnaire sent to them, one question gave respondents the opportunity to indicate whether or not they used eighteen Internet services, of which, in addition to the ten listed in the previous graph: games, video consumption(YouTube, Dailymotion …), consultation of catalogs of products or services, participation in forums / blogs, training, use of administrative services, participation in a social network site (Myspace, Facebook …) and the realization of banking or stock exchange transactions.
Considering each of these services as good that the individual asks and consumes according to the utility that he hopes to draw from it, the reasoning here is similar to that of access: given each of the uses of the Internet that offers itself to him, the individual decides whether to exploit it. Consequently, for each of these uses, the model explained variable will also be a binary variable with modalities 1 or 0 depending on whether the individual concerned will have chosen to exploit the use in question.
14 This being said, it must be recognized that it may be inappropriate to study all the eighteen listed uses, especially since some of these uses appeared to be particularly seldom chosen by the respondents. It is then necessary to find a criterion for selecting the uses to be analyzed. In this quest, we are led to retain the criterion of uses constituting the upper quartile of the most solicited by the populations.
We admit that this criterion may be open to discussion. But in this, it would be no exception, because any other criterion would also be open to discussion and we retained it simply by supposing that the fact of being limited to the 25% of the most common uses is acceptable.
15 Since this quartile is at the level of a usage rate of 40.42%, the uses that we will retain are: e-mail (or e-mail), search for news information international search for information on local or national news, live discussions with friends / families outside the country and search for information related to the profession or studies.
It should be noted that this selection makes it possible to analyze the determinants of uses for asynchronous (e-mail) and synchronous (live chat) communications, as well as the determinants of uses for event information (national and international news) and professional (professional and personal). studies).
16 However, note that for every person, the two decisions analyzed are linked in a two-step sequence: in a first step, the individual decides whether to access the Internet and, in a second step, he decides to whether to exploit a given use or not. It is therefore obvious that the access model will always concern a larger number of individuals than the usual one since the tree of the parts of the two modeled decisions takes the following form: