Maxipol: Data analysis and results

We present results from and the analysis of data from MAXIPOL, a balloon- borne experiment designed to measure the polarization in the cosmic microwave background (CMB). MAXIPOL is the first CMB experiment to obtain results using a rotating half-wave plate as a rapid polarization modulator. We report results from observations of a sky area of 8 deg(2) with 100 resolution, providing information up to l similar to 700. We use a maximum likelihood method to estimate maps of the Q and U Stokes parameters from the demodulated time streams, and then both Bayesian and frequentist approaches to compute the EE, EB, and BB power spectra. Detailed formalisms of the analyses are given. We give results for the amplitude of the power spectra assuming different shape functions within the l bins, with and without a prior C-l(EB) = C-l(BB) = 0, and with and without inclusion of calibration uncertainty. We show results from systematic tests including differencing of maps, analyzing sky areas of different sizes, assessing the influence of leakage from temperature to polarization, and quantifying the Gaussianity of the maps. We find no evidence for systematic errors. The Bayesian analysis gives weak evidence for an EE signal. The EE power is 55(-45)(+51) mu K-2 at the 68% confidence level for l = 151-693. Its likelihood function is asymmetric and skewed positive such that with a uniform prior the probability of a positive EE power is 96%. The powers of EB and BB signals at the 68% confidence level are 18(-34)(+27) and 31(-19)(+31) mu K-2, respectively, and thus consistent with zero. The upper limit of the BB-mode at the 95% confidence level is 9.5 mu K. Results from the frequentist approach are in agreement within statistical errors. These results are consistent with the current concordance Lambda CDM model.