Transforms to acheive linearity

[Moose352]

I was reviewing some statitics and got a little confused with transformations to achieve linearity in bivariate data. The book is really vague and rather than trying to figure it out, I figure someone here will be able to help. I'm not so sure as to what transformations are best applied to which type of relationships and how to apply the transformations. Any help would be appreciated.

 

[Moose352]

No one wants to help? The book lists the square root transformation and the reciprocal transformation and the square transformation but does not say when they are used. For example, it says that a square root transformation is used when the spread of observations increases with the mean...what is that supposed to mean?

 

[quantumdude]

OK, help me understand a little better.

The book lists the square root transformation and the reciprocal transformation and the square transformation but does not say when they are used.

Looking through the indices of all 3 of my stats books, I find nothing by those names. Can you please type out those transformations, so that I can correlate them to the ones in my books?

For example, it says that a square root transformation is used when the spread of observations increases with the mean...what is that supposed to mean?

It means that you use that particular transformation when the standard deviation is an increasing function of the mean.

 

[Moose352]

It means that you use that particular transformation when the standard deviation is an increasing function of the mean.

I don't really understand what you mean by that. I don't understand how the standard deviation can be a function of the mean in a single data set.

 

[quantumdude]

I don't really understand what you mean by that. I don't understand how the standard deviation can be a function of the mean in a single data set.

I'm just trying to go by what you said. The spread in the data is parametrized by the standard deviation.