Difference Between Constant & Proportional Error

Understanding the difference between constant and proportional error in statistical analysis will allow a function to be properly graphed. Once a graph is completed any value on the y axis can be found if the x value is known and vice versa.

A constant error is an average of the errors over the range of all data. The x value will be independent of the y value. For example, an affixed scale will always have deviation from the zero setting whether the item being weighed is 100 lbs., 600 lbs. or anywhere in between and this error has nothing to do with the actual weight of the object. The average deviation of a single instance will decrease as the number of instances increases.

Proportional error is an error that is dependent on the amount of change in a specific variable. So the change in x is directly related to the change in y. This change is always an equally measurable amount so that x divided by y always equals the same constant. The amount of error will always be a consistent percentage.

An indeterminate error is one that is neither constant or proportional. These errors are often the result of observer bias or inconsistent methodology during an experiment. Indeterminate errors can also be a sign that there is absolutely no correlation between the two items being compared. In cases like this it is important to revisit all facets of data collection including experimental bias and inconsistent measurements.

A constant error will be reflected in a change in the y intercept on the graph. A proportional error will change the slope of the line on the graph. Indeterminate errors will cause a scatter plot effect on the graph, making the determination of the line of best fit impossible.