# how to check normal distribution of data - real-time PCR data (Jul/11/2007 )

Hi all
I am analysing six different treatments with n of 6 or more for real time PCR
When using Sigmastat, it says normality failed.
what can be done to make sure that normal distribution is achieved.
i know that increasing sample size is one way to do it

listing down my questions?
1-can outliers be removed without justification
2-would that help achieving normal distribution
3- can someone explain normal distibution in an easier way

thanks
any suggestions n advice is grealt appreciated. thanks for ur precious time on this topic

-Watson-

Hi Watson,

1. it depends on the outliers, but i wouldn't remove them just yet.
2. what you can sometimes do to achieve normality with a function that by itself is not normal, is to transform it using a monotonic function. A good example is to log your function. Sometimes if the starting function is very asymmetric, the log of this function is roughly normal.
3. the normal or Gaussian distribution is a curve that MODELS data that are spread around a certain value (the mean), with the same probability of being above or below this value. The shape of the curve depends on two values: the mean and the standard deviation. For example: suppose a factory makes steel rods of roughly equal length (1 m). The length of the rods will be distributed normally with the mean 1 m. But not all rods will be exactly 1 m long. There will be some that are 99 cm and some that are 101 cm long. The standard deviation depends on the exactness of the cutting of the rods, that is the number of rods that are different from 1 m. Because there is no real reason that there would be more rods of 99 cm than 101 cm, the curve is symmetric around the mean 1 m.

Hope it helps.
Regards,
Miha

-BioMiha-

QUOTE (BioMiha @ Jul 13 2007, 02:52 AM)
Hi Watson,

1. it depends on the outliers, but i wouldn't remove them just yet.
2. what you can sometimes do to achieve normality with a function that by itself is not normal, is to transform it using a monotonic function. A good example is to log your function. Sometimes if the starting function is very asymmetric, the log of this function is roughly normal.
3. the normal or Gaussian distribution is a curve that MODELS data that are spread around a certain value (the mean), with the same probability of being above or below this value. The shape of the curve depends on two values: the mean and the standard deviation. For example: suppose a factory makes steel rods of roughly equal length (1 m). The length of the rods will be distributed normally with the mean 1 m. But not all rods will be exactly 1 m long. There will be some that are 99 cm and some that are 101 cm long. The standard deviation depends on the exactness of the cutting of the rods, that is the number of rods that are different from 1 m. Because there is no real reason that there would be more rods of 99 cm than 101 cm, the curve is symmetric around the mean 1 m.

Hope it helps.
Regards,
Miha

thanks for the feedback..was really helpful n timely

-Watson-