As as researcher you must have felt perplexed whether to use a parametric test or a non parametric test, at some point of time at least.
If there is still any doubt in that context, then this surely is a good read that you must take up. It will give you some good insight into what exactly are parametric tests from a critical perspective.
Parametric tests are nothing different from conventional statistical procedures. This is so because in each of the parametric tests, you got to apply statistics so as to get some estimation about the parameter of the population. The fundamental requirements for a parametric test to be applied are:
- Need a sample and a sampling distribution
- A universe from which the sample has been drawn
- Some parametric assumptions about the population
The biggest and the most important benefit of using parametric tests that make it the most popular test to be used by researchers is that it does not require a great amount of data that has to be converted into some order or ranks for the purpose of applying tests. This benefit further makes them very easy to not just administer but also calculate. Also there are a lot of softwares available for calculating parametric tests, which makes the task all the more simpler.
The output of the parametric tests can be quite reliable because they the results that generalise the information about the population state them in terms of confidence intervals, thus adding to their reliability.
However, it is not all good about parametric tests. They have their own disadvantages and need to be used very carefully. They aren’t applicable and lose their validity when we try to apply them on smaller data sets. In addition to that, the prerequisites from the population are quite stringent. Other than the data set being big, the variances in the population should be of similar kinds and the measurement of the variables that are being studied should have been measured at the same scale of intervals.
The most typical feature of the parametric tests is that it only works on continuous data and the results do get distorted because of the outliers present in the data. The strength of the non parametric tests is much greater here because they can also handle ordinal data, rank data and it stays unaffected by the impact of the outliers. However, the non parametric tests have their own set of requirements and assumptions that must be checked well in advance before applying them.
Follow Us
