I'm a PhD student who recently had an abstract accepted for a conference. It was based on a survey that was scheduled to take place before the conference, but external issues caused the survey to be delayed. I don't have the data I thought I would. I've tried to rework a paper that is more theory-based and introduces some descriptive data from another data set, but it's not a strong paper. Should I cancel late notice or just suck it up? It would be my first conference.
You don't mention your exact field, but it is very common in the humanities for conferences papers to be, in effect, works (of larger scale) in progress.
One of the really nice things about that is that when you present it, you will have a number of people with expertise in your topic listening and often times will provide very constructive comments or ask questions that ultimately help you deepen or extend or improve the paper.
So as you mention, moving it to be more theory based isn't a bad thing. You can focus on the design of your survey, and mention the intended use, and you will undoubtedly get some solid feedback (it's also totally fine to be upfront and explain why your paper doesn't match the submitted abstract/title).
The end result will be an even better survey that, once you've done it and analyzed it, will lead to a more easily published article (perhaps with another conference paper along the way). You may also get people to come up to you afterwards who are doing similar work, and that could result in collaborations down the road.
As it is an oral presentation go and do your best. This assumes that there is some novelty and your scheduled approach is of interest. You must decorate (verbally) it and gives to the audience the envy to proceed in the same way.
You can also represent related previous data and results.
I assume that you have a minimum of a solid base, as it seems from the question. I also assume that you have limited time, say 15 minutes max.
Otherwise don't go.