Using data from the ECB's Survey of Professional Forecasters, we investigate the reporting practices of survey participants by comparing their point predictions and the mean/median/mode of their probability forecasts. We find that the individual point predictions, on average, tend to be biased towards favourable outcomes: they suggest too high growth and too low inflation rates. Most importantly, for each survey round, the aggregate survey results based on the average of the individual point predictions are also biased. These findings cast doubt on combined survey measures that average individual point predictions. Survey results based on probability forecasts are more reliable.