Number of injection for Standrad for Related substance test

[kalidassa]

Hi to all we are following for all Related substance duplicate injection of diluted standard and the % RSD limit we following is Not mote than 2.0%.

Any guideline is there to address our standard injection?, We found that for six replicate injection % RSD is given Not More than 10.0%. Is our %RSD for two injection is stringent? Can any one help me.

[HPLCaddict]

Statistically speaking, %RSD of just two values doesn't make sense. You need at least three values to calculate a "meaningful" RSD. Nevertheless it's often done with just two values...
Generally, the more values you have the lower the %RSD should be (IF there's just the random experimental error involved). Taking this into account, an RSD of 2.0% with two values is a MUCH harder specification than a %RSD of 10.0% with 6 values.

Disclaimer I'm neither mathematician nor statistican, maybe someone else may clarify this issue better...

[unmgvar]

in general for related substances 6 repetitions are injected because it is statistically what you should inject for a 0.1% limit.
and the limit of %RSD depends on the level that you need to report.
the lower the level the higher the RSD limit that you get. all of this based on statistical rules.
it is also very acceptable to set as the final limit of your method something that is based both on the guidelines and the actual set of results that you have received from your validation.

the % relative standard deviation is something that cannot be done on 2 results only. you can make a calculation and get a result, but it is a wrong one. it is a meaningless one.

[Sallybeetle]

I have seen laboratories perform two injections, then calculate the %RSD on peak area #1, peak area #2, and average of PA #1 and #2. This gave them 3 numbers to use for %RSD. Can anyone tell me if this is statistically or mathematically correct? It never seemed to be a correct way of calculating %RSD to me.

[tom jupille]

You certainly can't calculate an RSD from one measurement. You *can* calculate it from two measurements, but the result is essentially meaningless. Using the mean as a third input merely compounds the meaninglessness.

Always remember Murphy's Paradox: "If you add a teaspoon of wine to a barrel of sewage, you end up with a barrel of sewage; if you add a teaspoon of sewage to a barel of wine, you also end up with a barrel of sewage."