# How to Refute Any Analysis By:: [[Brian Heath]] 2023-03-22 Analytics is not without weak points. Most of our posts on this subject focus on the human side and its practice. However, it is very easy to refute any analytics argument presented. All one has to do is point to one detail or assumption that is wrong or missing. Executives in organizations are excellent at finding these issues, especially when the analysis runs counter to their beliefs, vision, and job security. It typically goes something like this. The Analyst presents a model showing the nuanced failure of an activity the Executive is passionate about. In return, the Executive asks if this model accounts for the change in gravitational fields as one moves about the earth. The Analyst says the model does not account for that but doesn't understand how that would matter. "Well," says the Executive, "varying gravitational forces impact our fatigue levels so that might account for why customers on the valley buy differently than customers on the mountain. Without accounting for this, I don't feel confident that the reason we are failing is the marketing strategy in the valley is not working." Three weeks later the Analyst returns to show that the change in gravity is not significant and had no impact. "Interesting," says the Executive, "well, while you were doing this analysis, I found a slight data error. Your model says the amount of money spent on marketing in July was $24,345. In actuality, it was $24,344.51. How can I trust that the rest of the figures are correct?" The Analyst is now stuck in an endless loop defending a model that is wrong but useful. As all models are wrong, a person can always refute them. If you think this example is ridiculous, you are either extremely fortunate or still early in your career. It will happen and drive you insane unless you prepare for it. Recognize when it is happening and decide how much you can put up with it. The problem will likely go away when the company goes out of business, the executive leaves or is fired, or you quit for something better. Advice is only received when the other party is willing to accept it. Don't bother otherwise. #### Related Items [[Analytics]] [[Organizational Analytics]] [[Management]] [[Models]] [[Science]] [[Work]]