Anomalous microwave emission (AME) is believed to be due to electric dipole radiation from small spinning dust grains. The aim of this paper is a statistical study of the basic properties of AME regions and the environment in which they emit. We used WMAP and Planck maps, combined with ancillary radio and IR data, to construct a sample of 98 candidate AME sources, assembling SEDs for each source using aperture photometry on 1°-smoothed maps from 0.408 GHz up to 3000 GHz. Each spectrum is fitted with a simple model of free-free, synchrotron (where necessary), cosmic microwave background (CMB), thermal dust, and spinning dust components. We find that 42 of the 98 sources have significant (>5σ) excess emission at frequencies between 20 and 60 GHz. An analysis of the potential contribution of optically thick free-free emission from ultra-compact H ii regions, using IR colour criteria, reduces the significant AME sample to 27 regions. The spectrum of the AME is consistent with model spectra of spinning dust. Peak frequencies are in the range 20-35 GHz except for the California nebula (NGC 1499), which appears to have a high spinning dust peak frequency of (50 ± 17) GHz. The AME regions tend to be more spatially extended than regions with little or no AME. The AME intensity is strongly correlated with the sub-millimetre/IR flux densities and comparable to previous AME detections in the literature. AME emissivity, defined as the ratio of AME to dust optical depth, varies by an order of magnitude for the AME regions. The AME regions tend to be associated with cooler dust in the range 14-20 K and an average emissivity index, βd, of +1.8, while the non-AME regions are typically warmer, at 20-27 K. In agreement with previous studies, the AME emissivity appears to decrease with increasing column density. This supports the idea of AME originating from small grains that are known to be depleted in dense regions, probably due to coagulation onto larger grains. We also find a correlation between the AME emissivity (and to a lesser degree the spinning dust peak frequency) and the intensity of the interstellar radiation field, G0. Modelling of this trend suggests that both radiative and collisional excitation are important for the spinning dust emission. The most significant AME regions tend to have relatively less ionized gas (free-free emission), although this could be a selection effect. The infrared excess, a measure of the heating of dust associated with H ii regions, is typically >4 for AME sources, indicating that the dust is not primarily heated by hot OB stars. The AME regions are associated with known dark nebulae and have higher 12 μm/25 μm ratios. The emerging picture is that the bulk of the AME is coming from the polycyclic aromatic hydrocarbons and small dust grains from the colder neutral interstellar medium phase.