Personal values alone cannot fully explain moral decisions. They are often treated as the main driver of moral choice, yet situational pressures can still shift actions when values are fixed. To study this effect, we introduce VISDA (Value-Informed Scenario-Driven Actions), a dataset of 80 binary moral scenarios across various themes. Keeping value profiles constant, we evaluate three open-weight LLMs (Gemma 4 31B, Qwen 2.5 32B, LLaMA 3.1 8B), a utility-based mathematical baseline over 95 Schwartz profiles, and humans for their decisions under modifiers. Each scenario is augmented with eight situational modifier axes designed to capture contextual pressures on decision-making. LLMs reverse decisions (flips) less frequently (2.03%-8.92%), with larger models showing greater sensitivity. In LLMs, authority and self-preservation modifiers drive most flips and remain statistically significant. Humans showed a pooled shift (12.36%) in their decisions. Humans prioritized stakes and affective modifiers, while LLMs overweight personal-cost modifiers. This work reveals axis-level misalignment between human and LLM sensitivity to situational moral factors.