Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
parametric
Message posted by sanja on June 16, 2000 at 12:00 AM (ET)
what are the similarities and differences, advantages and disadvantages between parametric and non-parametric statistics ?
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: parametric
Message posted by Doug Mahoney on June 23, 2000 at 12:00 AM (ET)
Parametric and non-parametric analyses are similar in that they both try to answer the same question (or hypothesis test), but with different assumptions regarding the distribution of the data. Most parametric analyses require the gaussian assumption (ie normal distribution) for the response variable understudy. This assumption should always be verified when preforming a test of signficance. If the assumption isn't valid for your particular data, then non-parametric tests should be used. One advantage for parametric analyses is that estimation of effect size, confidence intervals, and data displays are simple and straight forward. One advantage of performing non-parametric tests is that even if the normal assumption is valid, the test is usually 90 to 95% efficient. That is, the results would match that of the parametric approach 90 to 95% of the time. One could always argue the "Law Of Large Numbers" when performing an analysis and blindly go with parametric approaches. However, that approach severly reduces the sensitivity of the analysis. This is because the estimation of standard errors are too large (due to skewness or long tailed distributions). A non-parametric approach is better able to handle this, thus increasing your test sensitivity.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.