log transformations of variables

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

log transformations of variables
Message posted by Shelly (via 142.165.70.20) on September 24, 2001 at 10:59 PM (ET)

I have variables for which the skewness and kurtosis is not acceptable. I want to do log transformations for these variables and am using the SPSS for Windows function : LG10(variable) to perform the task. However, my variables include vales of 0 which the SPSS does not accept. Any suggestions for the transformations?

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: log transformations of variables
Message posted by Darius (via 200.23.217.10) on September 25, 2001 at 11:14 AM (ET)

In case of log(0), you can do two tings:

1- You can ignore (take out) the "0" values.

2- You can add a constant to every value (ie. 0+0.1), the behabiur of the data will still be present.

Re: log transformations of variables
Message posted by Jack Tomsky (via 208.249.113.130) on September 26, 2001 at 2:05 PM (ET)

You can also consider other transformations, such as the square-root, cube-root, etc., which are defined at zero.

Re: log transformations of variables
Message posted by Dave (via 192.231.113.84) on October 2, 2001 at 5:28 PM (ET)

Some trnasformations require you add a constant to values of zero. But don't worry, It won't affect the test.

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.