Now that ELISAAnalysis.com provides 95% confidence intervals, many clients have asked how they can tighten the confidence intervals for their standard curves. Here are some comments and suggestions:
- You will usually see the confidence intervals widening for higher concentrations. This is normal when the percentage variability is roughly constant for all concentrations. For example, if the confidence interval is plus or minus 10% then for higher concentrations this will translate into a larger confidence interval.
- For a given level of variability, to tighten the confidence interval you should increase the number of replicates for your standard curve points. While duplicates for standard curve points are common, from a statistical point of view this is a very low level of replication and as a result this leads to low confidence.
- For a given number of replicates, to tighten the confidence interval you should explore strategies for reducing the variability between replicates. ELISA involves a large number of steps and the challenge is to implement experimental procedures to reduce variability at each step.
- The 4PL curve has asymptotes at both the positive and negative extremes. As the curve moves closer to the asymptote, the confidence intervals will widen. For a given concentration, the tightest confidence intervals for a 4PL curve will be near the inflection point of the curve, where it looks roughly like a linear curve at 45%. If your unknown values are near the asymptotes then you could consider diluting the samples to move them towards the centre of the curve or consider a more sensitive kit.