Annual

Individual

It does not contain an over-representation during the periods of strong influx to limit the effects of cluster. The compositions of the flights on periods July-August (holidays of summer and southern winter) on the one hand and on December-January (holidays of southern summer) on the other hand are similar in term of characteristics of the passengers. Airport of Rolland Garros: (95% of the influx)

a- The sampling plan rests at first on a selection of 126 distributed days of survey on one year (every 3 days systematically).

b- Then, planes are distributed in three strata, which correspond to three big types of destinations from Reunion Island: metropolitan France, Mauritius and the other destinations of the Indian Ocean. For every day of survey, a flight to metropolitan France and a flight to Mauritius are selected on a random basis.

Then every two days of survey, we drawn by lot the third flight either to Mauritius or to another destination of the Indian Ocean.

Airport of Pierrefonds: The lighter sampling plan rests at first on a selection of 24 days of survey spread on one year. Then, planes are distributed in three strata, which correspond to three types of destinations from Reunion: metropolitan France, Mauritius and Madagascar. For every day of survey, a drawing lots allows to select the flight to be investigated: a flight to metropolitan France, a flight to Mauritius and a flight to Madagascar (one day of survey on two). The Port of Pointe des Galets: No survey is realised considering the low passenger number (1,5% of the total flow). The destinations are quite due to Mauritius or Madagascar. The passengers statistics are communicated to INSEE by the departmental equipment directorate - department of Ports for the passengers due to leave Reunion Island which also supplies an estimation, considered reliable, on behalf of the tourists in this clientele. This tourists' flow is thus only aggregated into the total number of tourists from Reunion Island.

c- Finally, we distribute the questionnaire to each of the households of the surveyed flight. Questionnaires will be exploited. We do not know, a priori, the size of the households sample. It depends on the number of flights, on the rate of filling of planes, on the size of the households and on the rate of answering. The survey at the airport allows the exploitation of more than 20 000 index cards a year. The data can be compared without recourse to seasonal adjustment.