We perform observational tests of statistical isotropy using data from large-scale structure surveys spanning a wide range of wavelengths. Using data from 2MASS, 2MRS, and NVSS galaxies, and BATSE gamma-ray bursts, we constrain the amplitude and direction of dipolar modulations in the number count of sources projected along the line of sight. We pay particular attention to the treatment of systematic errors and selection effects, and carefully distinguish between different sources of dipole signal previously considered in the literature. Dipole signals detected in these surveys are consistent with the standard, statistically isotropic expectation, except for the NVSS result, which is likely biased by remaining systematics in the data. We place constraints on the amplitude of any intrinsic dipole driven by novel physics in the early universe.

36 pages, 20 figures. v3: minor additions to theory section; matches the published MNRAS version