Ok, based on your very limited documentation, I would do the following:
Fact Tables - Your fact table is your measurement table. It is the thing that happened. It is the meeting together of dimension tables, typically at a point in time. In your diagram,
- RESPONSES is your fact table.
- OPTIONS could be a fact table. If you denormalize it you could have dimensions associated with it.
Dimension Tables - Dimension tables contain the textual context associated with a measurement event. It describes the “who, what, where, when, how, and why.” Based on your diagram:
- SAMPLE would be a dimension table, describing the people who make a response. It would be the closest you have to a "customer" dimension table.
- QUESTIONS would be a dimension table, describing the question in detail. You could extend this dimension to include more data on the question type, such as length, IsMultipleChoice, IsFreeText, IsPersonal, isPrivate, etc. I would think of this dimension as a "product" dimension table, based on the limited schema you have.
- OPTIONS could be a dimension table. It looks like it is something you base the response field on. The possibility of responses for a given question.
Personally, I think you need to read up on Data Warehousing. The Data Warehouse Toolkit by Kimball is an invaluable resource.