The Student Room Group
Uni students - Complete our survey >>
Uni students - Complete our survey >>
Uni students - Complete our survey >>

STATISTICS question????

can something look similar but be significantly different?

I'm analysing some data and basically 2 variables were not significantly different with the stats test i used which did make sense looking at their mean values but the difference between means was around 6 but then one variable was REALLY significant (p=<0.0001) despite the means only being 0.1 in difference?

is that possible or am i messing up?
Original post by Moura
can something look similar but be significantly different?

I'm analysing some data and basically 2 variables were not significantly different with the stats test i used which did make sense looking at their mean values but the difference between means was around 6 but then one variable was REALLY significant (p=<0.0001) despite the means only being 0.1 in difference?

is that possible or am i messing up?


Yes, this is quite possible; it's not just the mean that matters! Typically, for simple statistical tests, two other things contribute: the standard deviation and the sample size. If you have a larger standard deviation, then a difference of means has to be larger to get a "significant" p-value; if the sample size is larger, p-values tend to be smaller.

Remember that the standard deviation represents one aspect of how spread out your data is. If it is more spread out, there is more uncertainty in estimating it's mean; there is more uncertainty in estimating quantities such as differences of means.

Quick Reply

Related discussions

Latest

Trending

Trending

Articles for you