A Gift of Fire: Questions 4.8 and 4.10 p. 182-3

4.8 a.) Yes, it is more likely to occur in computerized records, because on paper, text and writing does not need to be coded and decoded in order to be transferred, it simply must be physically transferred, whereas with an electronic record, different coding schemes may have an error in which the transfer of data causes the data to be misinterpreted and say something completely different when it is decoded and stored in the new space.

b.) One practice that can help prevent these errors is to run checks to make sure that the coding systems are compatible with each other and that they don't cause errors in the transfer of data. Another practice would be to update the software so that the different types of coding are directly compatible and recognize each other, shrinking the margin for error. An alternative to this would be for every school in a given area use the same coding schemes for things such as disciplinary coding.

4.10) Had the scores been reported higher instead of lower, there would be no complaining by the schools and, if someone within the school did suspect something, they would probably stay quiet about it because the higher test scores would mean that they would be looked upon as better and their careers would become better. However, if another school suspected something and noticed that many schools had a sudden rise in test scores, they would eventually suspect that a software error may be present, and, if found, it would be corrected and it would demote the people of the school whose lives were boosted from the error. One situation where a computer error might not be reported would be if a company's stock values were to increase suddenly causing more people to invest in the company. The company would most likely not report the error, because their business would be doing better. Another case might be if an electronic survey was taken across a large scale of people. A malfunction in the software could cause the results of the survey to be skewed, and so if an organization had planned to take action depending on the results, they may perform the wrong action based on incorrect results. Since the survey would be set on a wide scale, the people who see the results would most likely not collaborate and discuss their answers; they would just accept that the answer they gave was merely not the popular choice. The company would most likely not see it either because they probably are not familiar with the behavior and personalities of people on a very broad spectrum.