0
$\begingroup$

I am a beginner on Statistical analysis and I'm honestly not confident with the tools I'm choosing. I conducted an experiment on measuring a person's pulse rate using our project/ device and compared it with their ACTUAL pulse rates

the results are given below:

enter image description here

We want to know if the Measured values can be interpreted as "effective" (regarding the device) compared with the Actual Pulse Rates measured by a commercial oximeter. The average outcomes from the trials will be compared with the Actual pulses.

Improve this question
$\endgroup$
4
  • 2
    $\begingroup$ Welcome to the site. What do you mean by "effective"? $\endgroup$
    – Peter Flom
    Commented Jan 24, 2020 at 14:57
  • 2
    $\begingroup$ It would be helpful to know more about what you mean by "effective". For example, do you want to know if pulse rate has, on average, increased/decreased relative to the initial measurement? Or do you want to know if individual rates are being measured reliably across devices? Or some other question? $\endgroup$
    – TPM
    Commented Jan 24, 2020 at 14:57
  • $\begingroup$ We would like to test if the average outcomes of the trials are "accurate" when compared with the actual pulses. $\endgroup$
    – Sydney Alyson V. Nudo
    Commented Jan 25, 2020 at 0:34
  • $\begingroup$ In designing these experiments, it would also be useful to make multiple measurements with the "gold standard" device to see how variable it is. If this is really about pulse rate, it happens that I monitor my own blood pressure (and pulse rate is reported too), and the results are rather variable in a single sitting. Two different things to consider with the new device are accuracy and precision. $\endgroup$
    – Sal Mangiafico
    Commented Jan 26, 2020 at 9:53

1 Answer 1

Reset to default
0
$\begingroup$

If your N is actually 4, there isn't much you can do. But if the data shown are just a sample, then here are some ideas:

You could just look at the average of the trials vs. the true measure. You can improve your looking ability by using graphs. Certainly a scatter plot. Perhaps a quantile quantile plot and maybe a Tukey mean difference plot (aka Bland Altman plot) and a plot of the differences between the two numbers.

If you want a formal test, you can look into tests of equivalence; more particularly TOST (two one sided t test) may be useful. You would have to specify a difference that you regard as too small to matter.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.