Is it the car that's deadly, or does the car attract deadly drivers?

For example, are there other, stronger predictors of deadly accidents such as age, geography, or occupation? And are those people more likely to buy certain cars? I notice Buick has fewer deaths per million, but could that be because Buick attracts more cautious or less frequent drivers?

I assume the IIHS has competent statisticians. But I'd still be interested to see how strongly the differences between drivers factor into this.