3

I am checking the accuracy of one device that finds the heart rate against the other device that gives correct measurements. Specifically, I have performed an experiment on a group of people. I used both device 1 and device 2 simultaneously on the each person and collected 7 values. Having 10 participants, I have in total 70 values for each device. I placed the data from all participants together and I have two columns, one for each device, of 70 values.

Now, I need to compare the two set of values to find whether they are significantly different or not. I am not sure what test to use here.

If I use the t-test, which one should I use? Paired, Two Samples assuming equal Variances or Two samples assuming unequal variances?

I am a bit confused.

phedon rousou
  • 133
  • 3

1 Answers1

1

I don't think you want to know if they're significantly different - you want to know if they agree, and that's not quite (but nearly) the same thing.

If you're interested in agreement you should look at the limites of agreement method of Bland and Altman (often called Bland-Altman plots, but that's not really true). Martin Bland has a section on his web page here: http://www-users.york.ac.uk/~mb55/meas/meas.htm with the papers that describe the approach. There's also a Wikipedia page: http://en.wikipedia.org/wiki/Bland%E2%80%93Altman_plot

Jeremy Miles
  • 17,812
  • 1
    Since apparently you (@phedonrousou) have repeated measures on the same subjects, this paper might prove useful: Myles & Cui, Using the Bland–Altman method to measure agreement with repeated measures - British Journal of Anaesthesia 309–11 (2007), doi:10.1093/bja/aem214 – boscovich Sep 29 '13 at 11:33