concentration from standards 0 to 5 looks good on the trendline as r value is 0.9 and above but on inserting the OD of the last two standards, the r value becomes awful
Maybe you reach a saturation plateau with later standards (higher conc.) which disrupts your trend line obtained from lower concentrated standards. You should avoid to use standards in too high concentrations to obtain relyable standard curves!
There could be multiple explanations, Andreas is correct that it likely is a problem of saturation. You might be saturating your reaction and looking in the nonlinear phase. Or you might merely be operating outside the linear range of your spectrophotometer, at high densities they are normally no longer accurate.
As Michael J. Benedik says, there are multiple explanations, and without looking at the data it is difficult to say with certainty. Most of the time in lab practice the culprit is non-linearity of the data due to limitations of the spectrophotometer, saturation of the reaction, etc. as mentioned above by both Andreas and Michael (in those cases the data looks as in the 'Systematic' picture I attached, although I exaggerated the effect for clarity). In that case you can either dilute the samples so that they fall within the linear range or you can use nonlinear regression, but if you stick with linear regression you should not include the higher-concentration points, as they deviate systematically from the model (linearity) and if included, would result in less accurate estimates of slope, intercept, and the concentration of the samples.
But including the high OD values might lower r simply because their variance is higher, as exemplified in the "random" figure below. The distinction is important, because in this particular case, although including them will lower the regression coefficient, the accuracy of the resulting estimates will actually be better, as sample size (the number of points used to derive the standard curve) is larger. In other words, the confidence intervals of the parameters of the curve and of the inferred concentration of the samples will be narrower, so in this case you should not discard those points.
Hope it helps!
Prof. Michael J. Benedik
thank you sincerely for the enlightenment.
i will look into those items highlighted Prof.
Alejandro Martin
thank you so much for you explanation.
now the issue is well understood
- Badges
- Science method
- Similar topics
- Medicine
- Immunology
- Immunology Techniques
- Immunoassay
- ELISA