|
Good design is critical
Once you are familiar with the terminology used in carrying out a trial, you need to understand how to actually design a 'fair trial'.
What design features do you think experiments need to have?
A well-designed experiment needs to:
- be as simple as possible
- generate answers to the hypothesis that is formulated
- take into account the variables under the same condition
- include some replication to reduce errors.
A good experiment can always be replicated. Replication ensures the:
- statistical analysis of the observations made has a high degree of accuracy
- collection of data is easily maintained
- elimination of any bias that may influence the results.
A fair trial is always said to have three important criteria. What do you think each criterion means in relation to carrying out an experiment? Click on each to reveal an explanation.
If you can't access the interaction below, you can use this alternative instead.
accuracy
|
An accurate measurement method is one that gives the measurer the correct value or very close to it. An unreliable method cannot possibly be accurate. Consistency also does not guarantee accuracy. Accuracy allows groups to be compared and results to be statistically tested.
Example: An incorrectly calibrated thermometer might yield highly consistent readings on an interval scale, but they will all be inacurrate because because the initial calibration method was incorrect.
|
| reliability |
The same measurement leads to the same results and thereby demonstrates the consistency of the results.
Example: If you were carrying out diet trials, the results would clearly show if one diet was always outperforming the other diet or if there were times when the results were inconsistent.
|
| validity |
Determines whether the observation recorded measures what is required.
Example: If you were carrying out diet trials to determine the performance of each, the measurements that would reveal this information would be the length, weight and FCR.
|
It is easy to see from the feedback above that replication is the key to ensuring the experimental design meets the accuracy, reliability and validity criteria. It is always a good idea to spend a little time thinking about the processes involved with designing and carrying out an experiment. Remember, good experimental design is simple, generates an answer to the hypothesis, takes into account the variables under the same conditions and can be replicated to eliminate errors.
|
