Internal validity is whether a study's design lets you confidently conclude that the variable being tested actually caused the observed effect, rather than something else.
Imagine a study claims that a new mindfulness app reduces anxiety, based on the fact that people who used it for a month reported feeling calmer. That sounds like good news for the app, but internal validity asks a sharper question: what else changed during that month? If participants also knew they were being watched, or the study happened to run during a season when anxiety tends to dip anyway, either of those could be the real cause, not the app. High internal validity means the study's design has ruled out those alternative explanations, usually through things like random assignment, control groups, and controlling for confounding variables.
Internal validity is really about the strength of a causal claim, specifically. A well-designed randomized controlled trial tends to have high internal validity, because random assignment spreads outside influences evenly across groups. A study where people chose whether to use the app themselves has much weaker internal validity, since the kind of person who opts into a mindfulness app may already be different from those who don't.
There's often a tradeoff with external validity: the tightly controlled conditions that maximize internal validity can also make a study's setting less like everyday life, which is part of why no single study settles a question on its own.