# Large t values

#### carruthers

##### New Member
Hi,

I'm studying the use of certain teaching techniques in Korean high schools, specifically comparing students' perceptions of the frequency of use of traditional and modern teaching practices. I would expect very strong results based on my experience.

I have data from 5-point Likert scales, and having run paired samples t tests on it (I know there is controversy on whether parametric tests should be used with Likert data - but it's been okayed from 'up top') SPSS is giving me (as an example) -the following:

Korean teachers utilise traditional methodology significantly more frequently (M=4.06, SE=0.03) than they utilise modern methodology (M=2.18, SE=0.04), t(305)=34.42, p<.01. (in SPPS p actually shows up as .000)

I'm getting similar t and p values for the other comparisons too. My question is, that t value seems very large, so is this just telling me I have an extremely significant difference, or have I cocked something up somewhere??

Thanks!

#### statsforpsychyinstructor

##### New Member
Hi,

I'm studying the use of certain teaching techniques in Korean high schools, specifically comparing students' perceptions of the frequency of use of traditional and modern teaching practices. I would expect very strong results based on my experience.

I have data from 5-point Likert scales, and having run paired samples t tests on it (I know there is controversy on whether parametric tests should be used with Likert data - but it's been okayed from 'up top') SPSS is giving me (as an example) -the following:

Korean teachers utilise traditional methodology significantly more frequently (M=4.06, SE=0.03) than they utilise modern methodology (M=2.18, SE=0.04), t(305)=34.42, p<.01. (in SPPS p actually shows up as .000)

I'm getting similar t and p values for the other comparisons too. My question is, that t value seems very large, so is this just telling me I have an extremely significant difference, or have I cocked something up somewhere??

Thanks!
A large t would mean that there is only a small probability that your result was due to chance. The reason you got such a big t is because there was very little variability in your samples or large sample sizes (or both)...which is a good thing..it would signify that your means would be a good estimate of the population parameter.