Quasar variability: an astrostatistical challenge
Variable behavior on all time scales seems to be ubiquitous phenomenon in almost all classes of active galactic nuclei. This work deals with two topics. One being detection of microvariability (intra-night variability) in the sources which are obscured and were believed to be non variable - type 2 quasars. The detection of a variable behavior in obscured sources is challenging as the detection is hampered by the low contrast between the presumably variable nucleus and the host galaxy. The second part deals with comparing of the variable behavior with larger amplitudes amongst two types of unobscured quasars: core dominated radio loud quasars (CRLQ) and radio quiet quasars (RQQ). Microvariability studies of obscured quasars are the focus of chapters 2 and 3. We have explored the possibility to search for short, small amplitude variations, with statistical tests, namely the F-test and one-way analysis of variance, ANOVA. The data for our sample were obtained during two observing runs, one using the 1.5 m telescope at San Pedro M\'artir observatory in Mexico and another one with Nordic Optical Telescope (NOT) at Observatorio Roque de los Muchachos, La Palma. Both samples were observed in optical wavelengths. The fourth chapter is dedicated to a study independent on the previous two chapters, which studies data set obtained during a year long campaign carried out at San Pedro M\'artir observatory. The data set contains observations of a paired set of unobscured CRLQ and RQQ. The observations were obtained in a weekly cadence, in optical wavelengths, thus focusing on a low frequency observations. We used Bartels test to search for variability in this case. As a result, we report detection of microvariability phenomenon in a subset of targets observed during the microvariability campaigns described in chapters 2 and 3. It is a result previously unaccounted for in obscured type 2 quasars. We explore possible explanations such as structure in the obscuring medium. In the second part of the thesis, we focused no comparing whether it is more likely that CRLQ are variable than RQQ. We have conducted this study on a set of targets which was studied previously. The previous study focused on a similar comparison on high frequencies. We have confirmed the results of the previous study using low frequency observations and report no statistically significant difference between RQQ and CRLQ. We have demonstrated feasibility of different statistical tests in variability studies.