We propose to study macro logging techniques data by taking into account for each measure the volume affected by the physical process. The volume of sediment affected by a measure carried out with a logging tool (sonic logs, density logs, etc.) depends on the nature of the surrounding formations and on the setting of the instrument. The various parameters which influence the measurement are: the source-detector spacing, the spacing between detectors, the sampling interval along a well, and the Volume of material considered (the measured entity). Depending on the tool, the measured entity can be either cylindrical, spherical or ellipsoidal. Three main groups of measured entities have been identified: jointed entities, overlapping entities and disjoined entities. When the length of the measured entities exceeds the sampling interval, there is an overlapping of the entities (macro logging tools). In that case, each measurement on an entity is partially correlated to the measurement on a neighbour entity, addition of data is unmeaningful and vertical definition is low. Simple statistical treatments, geostatistics or multivariate analysis are then biased, and the bias increase with the ratio between the length of the measured entity and the sampling interval increases. Starting from the assumption that, for a given logging tool, the measured entity remains stable throughout a well, it is possible to make the data additive by transforming the measured entities. A geostatistical approach allows to study the integrator effect of the tool on logging data and three methods to homogenize logging techniques data are proposed. The interest of these techniques for the homogenization of data is shown in several cases (synthetics and actuals logs data). The homogenization technique can give more rigorous log data, which can then be treated without bias. These methods also give the opportunity to increase the vertical resolution of logging tools: the improvement really depending of the sampling interval.