Future space telescopes will provide us with the opportunity to characterize the atmospheres of terrestrial planets orbiting within the habitable zones of their stars. In the near future, the James Webb Space Telescope (JWST) will allow us to spectrally characterize terrestrial planets orbiting M-stars. In the more distant future, telescopes such as the Large UV/Optical/IR Surveyor (LUVOIR) and the Habitable Exoplanet Observatory (HabEx) will allow us to direct image Earth-sized habitable zone terrestrial planets orbiting Sun-like stars (exoEarths). In this dissertation, I explore the habitability of both G-star and M-star habitable zone planets. In Part 1 of this dissertation, I focus on habitable zone planets around Sun-like stars, and explore testing key theories of habitability statistically with direct imaging missions. These tests would exploit statistical marginalization of uncertainty inherent to both terrestrial planets and observations by using large samples of detected exoEarths. In Chapter 2, I propose a statistical test for the silicate-weathering feedback, which is a stabilizing negative feedback believed to help maintain habitable conditions on habitable zone planets. I find that if the silicate-weathering feedback is common on exoEarths, we are likely to detect it with LUVOIR-A, and may detect it with LUVOIR-B if the natural spread of planetary parameters is small. We are unlikely to detect it with HabEx due to its small expected exoEarth yield. In Chapter 3, I develop a statistical test for whether exoEarths tend to be Earth-like. Specifically, I consider whether observations with LUVOIR and HabEx could inform us on the fraction of exoEarths that are Earth-like (fE) in that they develop Earth-like, O2-producing life (oxygenic photosynthesis) that oxygenates their atmospheres roughly following Earth's oxygenation history. I explore this specifically in the event of a null detection of O2 or O3 on every exoEarth. I find that missions with larger aperture mirrors are more robust to uncertainties in the exoEarth occurrence rate (eta_earth), but all missions are vulnerable to inconclusive null detections if eta_earth is very low. In Chapter 4, I quantify the detectability of O2 and O3 for future direct imaging missions LUVOIR and HabEx using a 3D Global Climate Model that includes photochemistry and realistic cloud coverage. I find that realistic cloud coverage increases the detectability of both O2 and O3, despite the fact that clouds obstruct some of the absorbing gas column mass, due to the strong effect of clouds on planetary reflectivity for planets with a low-to-intermediate surface albedo. In Part 2 of this dissertation, I focus on habitable zone planets orbiting M-stars, which are expected to be tidally locked. Specifically, I explore the Snowball bifurcation on tidally locked planets. In Chapter 5, I show that tidally locked planets do not exhibit a Snowball bifurcation, regardless of stellar spectrum, as a direct result of the spatial pattern of insolation they receive. In Chapter 6, I explore whether realistic ocean heat transport can reintroduce the Snowball bifurcation for habitable tidally locked planets by destabilizing partially glaciated states. I show that including ocean heat transport does not reintroduce the Snowball bifurcation and that the lack of a Snowball bifurcation on tidally locked planets is robust to realistic levels of ocean heat transport. In Chapter 7, I explore whether limit cycling between Warm and Snowball states at the outer edge of the habitable zone occurs for tidally locked planets. I show that tidally locked planets with an active carbon cycle will not experience limit cycling as a result of their lack of a Snowball bifurcation, and instead will settle into "Eyeball" states with a small unglaciated substellar region if they orbit near the outer edge of the habitable zone.