Home >> Science & Nature >> Science

Three Reasons Poor Precision in a Scientific Measurement May Arise

In science, an accurate measurement reflects the degree of closeness to the measured object's actual, real-world value, whereas a precise measurement is one that is the same each time you measure it. While scientists can never obtain a truly accurate measurement, they can obtain as precise a measure as possible. The three reasons for poor precision in a scientific measurement are human, systematic and random error. While scientists can control measurement and operator errors, they accept random error as inherent in all measurement processes and beyond their control. Instead of futilely attempting to control error, scientists instead report statistical results within a range of values, known as a confidence interval.

Human Errors
- Even given a high level of vigilance, it̵7;s easy for a tired, overworked or misunderstanding lab assistant to read a gauge wrong, incorrectly use an instrument or make changes to the experiment halfway through the procedure. All of these variations from protocol are known as human error and detract from the precision of a scientific measurement. It̵7;s important to carefully and meticulously outline your experimental protocol, follow instructions to the letter, and check and recheck all your calculations. Train everyone involved in the experiment using the same protocols, emphasizing the importance of following instructions exactly to reduce human error to a minimum.
Systematic Errors
- Precision will be sacrificed due to systematic error if your measurement device has been calibrated incorrectly, you have given incorrect instructions resulting in the uniform misuse of a measuring device, or an external factor, such as wind resistance, is influencing your results. Systematic error will throw off your results, causing them to be systematically too high or too low, on every measurement. Balance and calibrate measurement devices according to manufacturer̵7;s instructions and ensure you have accurate instructions for using measurement devices to avoid systematic errors.
Random Errors
- Random error differs from both human and systematic errors in that the experimenter cannot control it. Random errors occur because not all factors can be controlled, no matter how careful an experimenter you are. A gauge, for example, no matter how precise, will probably not yield exactly the same results on every single measurement if used an infinite number of times. If a person makes several mistakes in an experiment in weighing items on a scale due to his hand shaking, for example, these errors will result in individual measurements being either too high or too low. This bias will be introduced randomly and throughout the results. In this example, therefore, the error type is known as random, even though a person made the mistakes, because normal human hand shakiness is outside human control.
Standard Error of Measurement
- Statisticians coined the term, standard error of measurement, to account for random error in experimental results. The standard error of measurement is calculated by taking the standard deviation of your test results and multiplying this by the square root of 1, then subtracting the reliability coefficient. If you are using a standardized instrument, such as the Wechsler test for intelligence, the test manual will have already calculated the reliability coefficient. If you are developing your own instrument, you will have to calculate this coefficient by having several different experimenters use the measurement device and comparing their results. The standard deviation will likewise be given in the testing manual if the instrument is well-known and standardized; otherwise, you will have to use your data from the reliability measurements to calculate standard deviation.
Confidence Interval
- Once you have calculated the standard error of the measurement, your results are reported within a range of values using this figure. If the standard error of measurement is .005 units, for example, and you obtained a measurement of 3, you would report your results as between 2.9 and 3.0. By using a confidence interval, you increase your chances of reporting the true measurement.

Science