I'm using SPSS to run some statistical tests on some data for a project. The data is comparing the antimicrobial activity of 2 different kinds of mouthwash and 2 different kinds of toothpaste, on 2 different species of bacteria.
Not counting controls, there are 8 groups of results (i.e. 4 treatments in total, each carried out on 2 bacteria species), and each group has 6 repeats.
I had assumed that these groups would be too small to be normally distributed, and from quickly making some histograms in excel they certainly don't look like it, so I assumed I would be using non-parametric tests. However, I decided to carry out the Kolmogorov-Smirnov test on it anyway, and some of the groups came out as normally distributed.
Is it possible to have normally distributed data when you only have 6 data points? Or should I follow my common sense and only do non-parametric tests?
Help with SPSS -- normal distributions? watch
- Thread Starter
Last edited by RUBecSO; 14-05-2013 at 18:20.
- 14-05-2013 18:15
Offline20ReputationRep:TSR Support TeamVery Important PosterPS Reviewer
- TSR Support Team
- Very Important Poster
- PS Reviewer
- 15-05-2013 12:21
This kind of data would be normally distributed but I would try both a parametric and non-parametric test, though try the parametric one first because it really should be the one you should be using.
- 03-06-2013 16:09
I know that this is an old(ish) post, but I thought I'd reply anyway
I’m just wondering whether you did any tests to check for normal distribution, first? This would then allow you to make the right choice when it comes to choosing a statistical test. If you run descriptive stats, and ask it to produce a histogram and a Q-Q plot, you will be able to physically see whether you data is normally distributed or not, rather than having to assume that it would or wouldn’t be. In terms of writing this up in a report, I think it would be much better if you had confirmed normal distribution – if the histogram has a bell curve, then go ahead with parametric. If there is a skew, or kurtosis, then you can choose the best test for that issue.