What is a Bad Statistical Graph

When analyzing a graph there is only one factor that makes it of less quality or “bad” and that is correlation. Correlation is how well your data follows a certain pattern. It doesn’t matter if it follows a liner relationship, exponential relationship, an inverse relationship, or any other possible kind of mathematical relationship; as long as your data follows some form of pattern is it correlated.

In statistics a few factors can influence your data and skew it so it may not follow the pattern you would expect. These factors would be called ‘outliers’ which are individual data points that do not line up with the rest because you either made an error when recording them or you recorded information from something that was not within your set. Examples of these two types of error would be:

Error in Recording: You were measuring how fast a car was traveling on the highway to get the average but your scanner jammed and displayed the wrong number.

Error in Set Management: You wanted to know how many state facts people in a particular state knew about their own state but by accident you polled people that lived in other states but just happened to be temporarily visiting in your state.

Some may think how would we find this error and remove it from our data set so we can have a correct representative model. Firstly we would have know that our data is quantitative (has actual numerical values) and is not qualitative (Categorical). This means that whomever we are testing we have to make sure that they are within our target group, and we can remove data that we think are unrepresentative and outliers. People may find this sinister and have used this fact of gathering data to build an infamous title for a statistician.

To actually correct this data we need to take as many tests as we can. Of course we cannot take a survey of everyone in the world; or let alone an entire state. So we just make sure we have a hefty sample group size to make sure we encompass the entirety of the group you are trying to represent. Than once you find can find an overwhelming pattern appearing within the set you can strike through the data cells that seem to not be in place ‘outliers’ and state that this is your representative relationship for whatever you are testing.