0

I'm currently trying to run (what I believe is) an independent samples t-test to determine whether there are significant differences between scores, however I'm a little new to R and am in the process of trying to figure out the landscape.

Essentially, I have a dataset (a sample for which I've provided in the code below), of regions by year and their respective test scores for which I'm trying to gauge levels of significant differences 

data <- read.table(text = 
     "Province       Score  Year
370   Alberta       549    2003   
333   Alberta       517    2012
371   BC            538    2003
444   BC            522    2012
372   Ontario       530    2003
445   Ontario       514    2012
555   PEI           500    2003
373   PEI           479    2012"    
, stringsAsFactors=F, header = T)

So far, I've tried:

pairwise.t.test(data$Score,data$Province, p.adjust="bonferroni")

However I'm not seeing any significant p-values, and I'm getting the feeling I might be overlooking something. I've also tried consulting the literature but have come up short. Just wanted to explore and consult some forums to make sure I'm covering my bases for my specific purposes. For instance, I'd want to know if the mean scores of Alberta were significant from 2003 to 2012, but right now I believe it's giving me a crosstabs of each province compared against other provinces.

Something like:

Saskatchewan 2003
Saskatchewan 2012
p-value = x

Ontario 2003
Ontario 2012
p-value = x

Would be what I'm looking for.

Thanks in advance for any guidance!

Jace

Jace_Riley
  • 1
  • 1
  • Why do you think that with those tiny sample sizes (2 values for each province) that you could possibly see any statistical significance? – IRTFM Jan 26 '20 at 23:45
  • I see what you mean. I thought there might be a way to simply see whether there was a significance between difference in scores, treating it sort of like a "score at time 1" and ", score at time 2" analysis – Jace_Riley Jan 27 '20 at 01:02
  • So you were hoping to find significance by analyzing four differences??? – IRTFM Jan 27 '20 at 01:43
  • Short answer, yes. There's a government-level doc I'm going through that's reporting significance levels between two scores (year one and year two) for a given province. I'm just hoping to replicate that analysis in R, the only difference being between two different years. – Jace_Riley Jan 27 '20 at 02:36
  • In the absence of a better description of how this data was derived, it's really not possible to comment with accuracy. I'm guessing these are summary data and htat you would need the raw data from which it was calculated to replicate the study. – IRTFM Jan 27 '20 at 17:46

0 Answers0