Radio and mm observations play an important role in determining the star formation properties of high redshift galaxies. However, most galaxies at high redshift are too faint to be detected individually at these wavelengths. A way to study this population of galaxies is to use stacking. By averaging the emission of a large number of galaxies detected in optical or near infrared surveys, we can achieve statistical detection.

We investigate methods for stacking data from interferometric surveys. Interferometry poses unique challenges in stacking due to the nature of imaging of this data. We have compared directly stacking the uv data with stacking of the imaged data, the latter being the typically used approach. Using simulated data, we find that uv-stacking may provide around 50% less noise and that image based stacking systematically loses around 10% of the flux.