Statistical Inference: non-significant results

In preparation

2 Responses to “Statistical Inference: non-significant results”

  1. dianakornbrot Mon, 27-Feb-12 at 10:33 am #

    In my view the most important information one needs when a result is NS is the a priori power to detect an effect if present.
    the METHOD sections should have a statement of the form, for example: tit was planned to have N participants in each group, which would give a power of 80% to detect a shift from 65% to 75% in the number of patients who improved (say) or an effect size (contingency coefficient or whatever) 0f .5. Then the results section can say something of the form: teh effect was ns at 95% confidence level with a power of 80% to detect an effect of given size. this suggests that there is no difference bewteen groups in the population
    Note. Power needs to be in method as well as results. Give effect size in real (e.g. % improvement, decrease in blood pressure) rather than statistical effect sizes if possible
    If powe is less than 80% then results are not worth reporting

  2. jose luis Sun, 26-Feb-12 at 10:01 pm #

    I will like to know about the interesting approache for inference in non-significant results!!

    Jose Luis (jl.col.her@gmail.com)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: