Do you know about this statistic method using GC/HPLC

[Ronaldo]

Hi mentors,

Actually , my supervisor asked my to do analysis of my analyte on GC and HPLC THEN compare the two method by but the peak value of GC in X axis and HPLC in y axis .Do you know why i use GC just in x and not in y .

Can you provide me with some materials can help me understand this method .

Thnks in advance because i am sure i will find the best help

[shaun78]

generally, when the results from one method are plotted against the results from another method the overall goal is to show that the two methods are in good agreement with each other. This is shown via a linear relationship between method one and method two.

[yangz00g]

search "student's t" test

[mardexis]

Hi Renaldo,

The y axis can be used to tell you how much substance is in your sample. The x axis tells you how good the separation is. If you could use gc or lc which would you choose? An important consideration is 'which method gives better separations' The answer to that is tricky but one general answer could be found by measuring the combined widths (x) of all the peaks found compared to the width of the total chromatogram. Try looking up 'peak capacity'

[jiang295]

by saying peak value, do you mean numbers of peak or peak retention time?

Hi mentors,

Actually , my supervisor asked my to do analysis of my analyte on GC and HPLC THEN compare the two method by but the peak value of GC in X axis and HPLC in y axis .Do you know why i use GC just in x and not in y .

Can you provide me with some materials can help me understand this method .

Thnks in advance because i am sure i will find the best help

[aceto_81]

You can swap your x and y axis if you want, there isn't a prescribed order to do this.
By plotting one method versus the other, you can see if your values are comparable (slope 1, intercept 0).
By checking your slope, intercept, r^2, residuals,... you can state that your methods are equivalent.

This is just one way, another way is to use a paired t-test on different samples.
Or an ANOVA with replicated analysis of different samples.

And there will be other (good) methods...

So pick your method and go for it

Ace