Accuracy, reliability, validity, and precision (uncertainty)
For a set of points that can be used to calculate an average, precision and accuracy can be thought of like this, the centre of the target being the actual or theoretical value:
Accuracy
The accuracy of the data from an experiment can be found by comparing the experimental value to an expected value. Some values have been measured by scientists many times and are accepted as true. Not all experiments are investigating an expected value.
An error percentage can be calculated to give a value for accuracy:
Validity (related to accuracy, informal/ non-statistical)
The validity of an experiment is how effectively it answers a question using the scientific method. The validity of a method is how effectively it measures what it intended to measure.
This includes factors such as whether the measuring instrument measures what it is supposed to measure, the conditions of the experiment and the manipulation of variables (fair testing).
Questions that should be asked are:
- How effectively were the variables controlled?
- How effectively was the independent variable manipulated and measured?
- How effectively was the dependent variable manipulated and measured?
- What assumptions were made?
Precision
The precision of a measurement or calculation depends upon the amount of significant figures and the uncertainty of any measurements. An electronic balance that measures to 0.001 grams is more precise than an electronic balance that measures to 0.1 grams.
Uncertainty (related to precision)
Every measurement has uncertainty. The amount of precision in the measurement can be used to find the uncertainty of the measurement. For example, a rock could have a mass of 2.4 on electronic balance A, a mass of 2.37 on electronic balance B and 2.366 on electronic balance C. For digital instruments (such as electronic balances) the uncertainty is the smallest possible measurement. The actual mass could be 2.3658g.
Reliability (also related to precision, informal/ non-statistical)
Reliability is whether the same method would get the same results if repeated. If a collection of data points is precise, the experiment can be considered reliable. A reliable experiment can still give results low in validity and accuracy.
- Are the trials result similar to each others?
- If the method is repeated by someone else, it it clear (justified) how it should be done to ensure result is similar?
(IA related) How to evaluate a method?
Reliability, validity, and precision are different and important concepts to stress in your evaluation section. Discuss your strengths and weaknesses in terms of validity, precision, and reliability
Few other notes:
- Justification that compare their data and literature value (found in background) → scientific context to say that the findings have high/low validity.
- Weakness and limitations should be based on the method: procedure and/or equipments used.
- Reliability discussion: sources of error must have directionality.
- Random and systematic error: random error will be within the uncertainty of the measurement (+/-). Systematic error will have directionality Eg. all points are lower than the accepted value. Uncertainty is NOT error, use language correctly.
- Improvement should NOT just more precise equipment. How to collect data differently and why?
- Extensions or improvements:
- Don’t: do the same exp with more trials or more number of independent
- Do: Extensions- Developments for further inquiry as related to the outcome of the investigation.This time did primary alcohols, next time secondary
- Improvement: how to make data move closer to expected outcomes.
