Earthquake ground motion intensity measures (IMs) are commonly estimated for engineering applications using ground motion models (GMMs) that estimate the median and lognormal standard deviation of IMs conditional on a series of source, path, and site parameters. Different GMMs are used in different tectonic regimes, and in the case of central and eastern North America (CENA), GMMs for stable continental regions are required. A major multi-agency, multi-investigator project (NGA-East) developed GMMs for CENA. That project, which was completed prior to the onset of the research presented in this thesis, provides the “initial conditions” from which the present work built upon. Model development in the NGA-East project had two components: (1) GMMs that predict lognormal IM distributions conditional on magnitude and distance, derived for a reference condition with a shear wave velocity (VS) = 3000 m/s; and (2) site amplification models that modify IMs for softer site conditions that are conditioned on time-averaged shear wave velocity in the upper 30 m of the site (VS30). I investigate whether these models, when used in tandem, are compatible with an expanded inventory of ground motion recordings in CENA (i.e., the database was expanded relative to what had been developed during the NGA-East project). This effort was divided into six major tasks: (1) data processing; (2) consistent metadata compilation and organization of the data into a relational database; (3) residuals analyses; (4) analysis of mean misfits of the combined model (hard-rock GMM and site amplification model) relative to the data; (5) identification of potential path-model misfits for data in different geologic domains within CENA, with model adjustments as needed to remove those misfits; and (6) investigation of dependencies of residuals on site parameters, including VS30 and sediment depth in coastal plain regions using depth models published by the U.S. Geological Survey (USGS) in 2024. The site response work in Task 6 is intended to validate the original model and further investigate depth-dependence of site response, which had not been previously investigated. To facilitate efficient data processing while maintaining protocols for human inspection of waveforms to ensure reliability, I worked with collaborators to expand the capabilities of the USGS’s automated processing code, gmprocess, to include displacement drift checks in the selection of high-pass corner frequencies. These procedures check for, and as needed remove, low-frequency artifacts in the displacement record using polynomial fits, which can be used in combination with existing signal-to-noise ratio (SNR)-based corner frequency selection procedures. The automated selections are then efficiently verified and refined using a graphical user interface (GUI) that plots relevant ground-motion time series and spectra and tracks modifications to signal processing parameters. Ground motion recordings from a total of 186 events from November 2011 (end of NGA-East data curation) to April 2022 were processed using this method and then archived in a relational database along with metadata collected following the recommendations of the NGA project for subduction zones (NGA-Sub; Contreras et al., 2022). The expanded database includes 2096 sites and 16,272 three-component recordings from events having a magnitude range of 4 to 5.8. I computed residuals using 17 NGA-East GMMs and three data selection criteria that reflect within-CENA regional variations in certain ground motion attributes. Partitioning of residuals using mixed-effects analyses reveals a persistent pattern of misfits in which ground motions are overpredicted at short periods (0.01 to 0.6 sec, including PGA) and underpredicted at longer periods. These misfits are regionally variable, with the Texas-Oklahoma-Kansas (TOK) region having larger absolute misfits than other parts of CENA. Two factors potentially influencing these misfits are: (1) differences in the site amplification models used to adjust the data to the reference condition during GMM development relative to those developed in the later stages of NGA-East, and (2) potential bias in simulation-based factors used to adjust ground motions from the hard-rock reference condition to a VS30 = 760 m/s condition. Bias adjustment factors and their epistemic uncertainties are provided, which were considered in the development of the 2023 National Seismic Hazard Model (NSHM; Petersen et al., 2023). I developed an equation-based version of the NGA-East tabular GMMs to provide physically meaningful parameters that could be adjusted, as needed, to remove scaling-trends in regions in CENA. After identifying five CENA regions (two coastal planes, two suture zones, and other areas), I found no appreciable errors in the NGA-East path model for any of the regions other than the Gulf Coastal Plain (GCP). In the GCP, faster attenuation was encountered, which is similar to prior work, and period-dependent adjustments were provided to remove the trends. Using the adjusted path models, residuals were recomputed and used to investigate site response. I find the trends in the data with VS30 to be essentially unchanged relative to current models, whereas significant effects of depth in coastal plain sediments were encountered for GCP and the Atlantic Coastal Plain (ACP). The depth model is formulated using differential depths derived using a VS30-conditioned depth model. Models for amplification conditioned on differential depth are different for GCP and ACP. ACP shows that shallower than average sites have higher short-period amplification and essentially average long-period amplification, whereas deeper-than-average ACP sites have reduced short-period amplification and increased long-period amplifications. For GCP, we see no bias of the VS30-scaling model at short periods for shallow or average depth sites, but reduced amplifications are found for deep sites at short to intermediate periods.