Naked Science Forum

Life Sciences => Plant Sciences, Zoology & Evolution => Topic started by: itadiki on 08/02/2018 22:42:05

Title: How many research subjects do you need for a study to be statistically valid?
Post by: itadiki on 08/02/2018 22:42:05

Hello,
When publishing research paper how do scientist know how many subject they will need for their study to be valid ?
I am intersting in dog behaviour and training subject for example.
Claudia Fugazza publish a paper.
http://www.academia.edu/18023840/Do_as_I_..._Did_Long-term_memory_of_imitative_actions_in_dogs_Canis_familiaris_ (http://www.academia.edu/18023840/Do_as_I_..._Did_Long-term_memory_of_imitative_actions_in_dogs_Canis_familiaris_)
She used 12 subject for her study.(and 12 other for comparaison How did she know that around this number it would be enough subject to be valid ?

Thanks

Title: Re: How many research subjects do you need for a study to be statistically valid?
Post by: petelamana on 10/02/2018 18:30:12

To have confidence that your study results are representative of the larger population, it is important that you have a large number of randomly-selected subjects in each study group, relative to the estimated population size. The larger the sample size, the less the margin of error.  The following link, www.surveysystem.com/sscalc.htm will provide you with a Sample Size Calculator, but a reasonably good approximation is the simple expression 1/n².

For example:  Say you want a confidence level of 95%, and you want the margin of error to be 6.5%, or 0.065, then...

                              sample size = 1/n², where n = margin of error.
                                                  = 1/(0.065)²
                                                  =  236 participants

What was the confidence level of the study Ms. Fugazza performed?  Most researchers use a confidence level of no less than 95%.  At 95%, and only citing 12 subjects, her margin of error would be an enormous 29 (28.29 to be exact).