While it's certainly the norm that you'd fail in that circumstance, it's not a given. There are up to 10 marks for the model, which include those for getting the right approach, and doing the steps correctly.
However, if you're able to produce results from your wrong model, explain them adequately (that is, draw decent conclusions from them), and present them properly, there's no reason why you shouldn't pass.
It's just a lot more difficult, as the questions are usually set up in such a way that doing them correctly provides you with a lot to say regarding your results. Doing it wrong may lead to results that don't lend themselves to pithy conclusions.