Well, simply.. you cant ensure your data is valid unless you have quantitative evidence. Therefore, by extrapolating.. you are only extending the trend of the graph, which would be invalid, as there would be no quantitative evidence to support your conclusion based on that extrapolation.