Microtubules (MTs) are biological polymers that perform crucial roles in eukaryotic cells such as formation of the mitotic spindle for chromosome separation during cell division, providing infrastructure for intracellular transport, and contributing largely to overall cellular organization. A fundamental behavior of MTs, dynamic instability (DI), is generally described as the stochastic switching between periods of growth and shortening. While this definition is widely accepted, it has limited the quantification of MT dynamics to simply include measurements of growth, shortening, and frequencies of transitions in-between. This definition ignores variability in growth and shortening rates and additional characterization of transitions themselves, which further limits mechanistic understanding. To address the issue of quantification, we first present a more complete analysis of MT dynamics using the Statistical Tool for Automated Dynamic Instability Analysis (STADIA); this analysis is performed using both in silico and in vitro data (Chapter 2). Using this data-driven approach, not constrained by a priori assumptions regarding MT behavior, we find that MT behavior is not well-approximated when constrained to measurements of simply growth and shortening; specifically, we report that a previously identified but not fully characterized behavior, `stutter', should be included in future studies of MT dynamics and in quantification of DI behavior. Additionally, we find that stutters precede catastrophes both in silico and in vitro, and that the anti-catastrophe factor CLASP2 prevents catastrophe by promoting stuttering MTs back to growth. To further validate these conclusions, we test their stability across a wide range of STADIA user-input parameters and data temporal resolutions (Chapter 3). Importantly, we report that the identification of `stutters' is generally robust with respect to both user-input parameters and data temporal resolutions. Of note, however, is that analysis must be conducted at a scale capable of detecting stutter behaviors. Lastly, we continue our dissection of MT dynamics by approaching MT velocity as a continuous random variable, as opposed to a binary variable (i.e., growth vs. shortening) (Chapter 4). Specifically, analysis of data generated by our dimer-scale in silico model demonstrates that MT velocity is best described as following multiple probability distribution functions, adding to our understanding of the spectrum of MT behavior. Further, our results suggest that MT velocity is determined primarily by the off-rate which is itself controlled by variations in the MT tip structure. We also establish and validate analytical relationships between rate constants, MT tip structures, and overall MT behavior; a key result of these analytical solutions is the identification of three random variables that sufficiently describe the tip structure and strongly correlate with the o -rate. For future work related to this dissertation (Chapter 5), we suggest two remaining areas of interest: 1) further optimization of methodologies used in STADIA (i.e., segmentation and classification procedures), and 2) the use of MTs as a model system for studying deterministic processes evident in steady-state systems.