A well-known effect of the near-spherical shape of Earth is that longitude-latitude map projections result in a spatial stretching of the information in the zonal direction in high latitudes. It is also well known that the variance of mean value estimates of climatic data varies inversely with the size of the area or length of interval over which the means are taken. A perhaps lesser-known implication is that the real degrees of freedom in the zonal mean value vary with latitude. Therefore taking zonal means for different latitudes involves averaging over different number of truly independent observations. It is shown here that a comparison between zonal mean values from different latitudes can give misleading results if the effect of latitudinal varying real degrees of freedom is not properly accounted for.