A: a large event, such as an earthquake, that happens in the middle of the study

    Incorrect.
An earthquake in the middle of an experiment COULD threaten internal validity, however this situation is not a selection bias.  We can imagine how the presence of an earthquake may allow us to question whether the independent variables (causes) had an effect on the dependent variables (effects), because a confounding variable is present, in this case, the earthquake.  Make sure you read the question carefully!  In this case the question asks specifically about selection biases.  Choice A is incorrect, because a selection bias may occur when random assignment procedures have been disrupted.  Typically, this would involve certain subjects being chosen over others, or a situation in which subjects select their own experimental condition.  An earthquake in the middle of the study would not do this.