If I have some actual data, and a model produced by someone else how do I measure the accuracy of their model?
You should make a similar model that you know the accuracy of then compare it. If you have paid someone to create the model for you then asking for a measurement of confidence in the model would not be unfair.
What kind of data do you have? What kind of model is it?
I have research data with 1000s of questions and answers, I then want to measure the accuracy of a model produced by CACI they have 100's of variables. Some are actual data and others a modelled based on their actual data. Mainly demographic data like, age, household income and some binary answers like gender.
I don't want them to tell me what the confidence in data is. I would like to test it myself. And produce some kind of measurement of accuracy for each variable.
There are many types of analysis you can do with a big data set. When looking at a very simple data set i.e. one binary variable you can set up a simple pie chart to analyse the proportion of respondents. Or you can compare different variables that may have different metrics to find correlations.
You can make some models to help you make predictions. The most common type would be a linear regression model. Have a look at Linear regression - Wikipedia, the free encyclopedia ms Excel will make the model for you.