Sigit Haryadi (July 18, 2019)
The story is related to a calculator https://www.haryadi.org/linear-regression-without-intercept/
About forty years ago, when I became a teaching assistant who was in charge of supervising students doing Basic Physics practicum in a laboratory whose tools were less accurate and precise, it was often overwhelmed by student errors without being able to provide a solution.
One story is when students are required to carry out a measurement of resistance (R) of a resistor by measuring the voltage (V) if the resistor is fed by an electric current (I), where measurements are repeated using the electric current that is changed, so it is expected to get a graph in the form of a linear line that has the function V = IR, then compare the R value obtained from the measurement with the actual value of the resistor that has been known based on measurements using accurate instruments.
Long story short, because the equipment used is less accurate, the students will get an R value from the experiment that is not the same as the actual R value. The problem is, students in their reports always say that the measurements taken are correct, where the results of their calculations do using Pearson correlation and get a straight line V = a + IR which is accompanied by a confidence level close to 100%. At that time I was very upset, because I did not have a way to calculate the level of trust in the results of the students’ measurement of the line V = IR rather than the line V = a + IR.
And, a year after discovering the Haryadi Index, recalling the events around forty years ago, I just knew an accurate and precise way to assess the level of trust in a measurable result that was not precise and accurate.