If future improvements to computer simulations of automobile crashes enable computers to provide as much reliable information about the effectiveness of automobile safety features as is provided by actual test crashes, then manufacturers will use far fewer actual test crashes.
Conclusion
If computer simulated crash tests become as reliable as actual test crashes, then manufacturers will use far fewer actual test crashes.
Evidence
The cost of doing computer simulations is lower than paying for actual test crashes.
Evaluate
At the heart of this question is a comparison between computer simulated crash tests vs. actual physical ones with cars and dummies and brick walls.
This argument makes a Prediction, that if computer tests were as reliable as actual tests, then we'd see more computer tests and fewer actual tests based on a Comparison between the two types of tests:
Computer tests are cheaper than actual tests.
If you were a car manufacturer deciding whether or not to go with computer tests vs. actual tests, so far you know that they're equally reliable, but computer tests are cheaper. Is there anything else you'd like to find out?
This is an Incomplete Comparison. We know the two options are tied when it comes to reliable information. Computer tests win when it comes to how costly. But do actual tests have any advantages? Faster turnaround time? Tax incentives? Do they provide some other kind of data that computer tests don't?
Goal
We want to hear some other way in which computer tests are better than actual tests (or at least, not worse than actual tests). Or we'd like an answer that stressed that 'how costly is it?' is an important part of the decision making process for manufacturers deciding whether to go with computer tests or actual tests.