This chapter introduces a range of explainable AI (XAI) techniques relevant to STEM education research, aiming to improve transparency, usefulness, and reliability. We begin with intrinsically explainable models such as linear and tree-based approaches, which are inherently explainable due to their design and structure. Next, we explore counterfactual explanations for understanding how individual input changes impact predictions and support personalized interventions. To overcome their limitations in capturing feature interactions, we present Shapley values as an advanced method that considers combined feature contributions. Gradient-based attributions are presented as tools for visualizing and quantifying how input features influence model predictions, including but not limited to analyzing student responses in STEM education research. We also introduce prompt design as a key approach to improving the explainability and controllability of LLMs. Strategies like in-context examples, knowledge augmentation, and chain-of-thought prompting help users shape model responses and gain clearer insights into model reasoning. Across the chapter, we present practical techniques to help researchers and STEM educators use explainable AI more effectively for clear and useful explanations.