Normal distribution + calculation of Z values

Click For Summary

Discussion Overview

The discussion centers around the calculation of Z-values in the context of a normal distribution, specifically how to determine Z-values for given percentages and the implications of the normal distribution's properties. Participants also explore the relationship between cumulative distribution functions (CDFs) and normality, as well as methods for estimating standard deviation from CDF data.

Discussion Character

  • Exploratory
  • Technical explanation
  • Mathematical reasoning

Main Points Raised

  • One participant inquires about the method to determine Z-values for specific percentages in a normal distribution, questioning whether it involves calculations or can be derived from observing the distribution.
  • Another participant mentions that tables of the error function, which are constructed through numerical integration, can be used to look up Z-values for a normal distribution with mean 0 and variance 1.
  • A participant introduces the "1-2-3 rule" (68-95-99.7 rule) as a helpful guideline for understanding the distribution of data within standard deviations from the mean, suggesting that this can aid in approximating Z-values.
  • A further contribution discusses a specific CDF with corresponding x-values and questions how to determine if the data follows a normal distribution without the original random numbers, as well as how to derive a probability density function (PDF) from the CDF.
  • The same participant proposes a method for estimating standard deviation based on the range of values at the 5th and 95th percentiles, seeking validation or alternative suggestions for this approach.

Areas of Agreement / Disagreement

Participants present various methods and rules for calculating Z-values and understanding normal distributions, but no consensus is reached on the best approach for determining normality from a CDF or the correctness of the proposed standard deviation estimation method.

Contextual Notes

Some assumptions about the normality of the data and the methods for deriving Z-values and standard deviations are not explicitly stated, leading to potential limitations in the discussion.

Who May Find This Useful

Readers interested in statistical methods, normal distributions, and the application of cumulative distribution functions in data analysis may find this discussion relevant.

JamesGoh
Messages
140
Reaction score
0
For a normal distribution with E[x]=0 and Var(X)=1, how do we determine the Z-value of a particular percentage ?

i.e. if the percentage is 5%, how do we know that Z(5%)= 1.645 ?

is there a calculation involved or do we get it from observing the x-axis of the normal distribution ?
 
Physics news on Phys.org
There are tables of the error function (integral of normal) for mean 0 and variance 1. These have been constructed by numerical integration. For a particular value, just look it up.

Google "normal distribution table".
 
A nice rule to remember too, is the "1-2-3 rule" aka 68-95-99.7 rule:

In a normal distribution, 68% of the data is within 1σ of the mean ,

(so that, by symmetry, 34% is right of the mean and 34% is left- of the mean)

95% of the data is within 2σ, and 99.7% of all data is within 3σ of μ.

Also, using the fact that the normal distribution is symmetric also simplifies

a lot of other calculations.

Notice an approximation for your 5% question: you know that the percentile for

the mean ; z(μ)=0 , is 50-percentile. Then, by symmetry, the value σ=1 gives

you the 84th percentile. Now, z=2 would give you the 97.5th percentile--

too far. So 95th percentile is somewhere between z=1 and z=2 . More

advanced tricks will allow you to zone-in more carefully, but this is a nice

rule- of- thumb.
 
To add to this question itself. I have a CDF so a column of 19 values. [0.05, 0.1, 0.15...0.95] and i have the corresponding x values [779, 784, 793...877 ]...again 19 values

When i plot graphically each other, it gives a smooth CDF following a normal curve however i am not sure if its normal, how do u derive if its normal since i do not have the random numbers.

Also I made a PDF formula for these values with d(CDF)/dx which means...(cdf2-cdf1)/(x2-x1) as coming from various textbooks. Is it right?

How do i generate Standard deviation from such a CDF?? Currently thinking that as 95% data is under 4σ area...[x(95) - x(5)]/4 will approximately give me the Standard deviation...Can someone suggest me the right way here
 

Similar threads

  • · Replies 4 ·
Replies
4
Views
2K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 5 ·
Replies
5
Views
3K
  • · Replies 24 ·
Replies
24
Views
4K
  • · Replies 1 ·
Replies
1
Views
2K
  • · Replies 14 ·
Replies
14
Views
2K
  • · Replies 30 ·
2
Replies
30
Views
5K
Replies
20
Views
3K