Why do you think sample size differs between untransformed and log-transformed variables?

answer:

• the occurrences where both the needy variable and free variable(s) are log-changed factors, the elucidation is a mix of the direct log and log-straight cases above.
• the understanding is given as a normal rate change in Y when X increments by some rate.
• the others have noted, individuals frequently change with expectations of accomplishing typicality before utilizing some type of the general straight model (e.g., t-test, ANOVA, relapse, and so forth). Be that as it may, I expect that much of the time, individuals commit two errors while doing as such:
• the they take a gander at typicality of the result variable as opposed to ordinariness of the mistakes. For OLS models, the mistakes are thought to be autonomously and indistinguishably conveyed as ordinary with mean = 0.
• the They overestimate the significance of the typicality presumption. Or then again putting it another way, they disparage the strength of OLS models to non-typicality of the blunders.
• the regards to OLS models, changes are all the more frequently about balancing out the fluctuation, it appears to me- - e.g., the log change when the SD is corresponding to the mean.
• these are the that as it may, in a few settings, one may change to get a test measurement that has an around typical examining circulation - e.g., the inspecting dispersion of the chances proportion (OR) isn't ordinary, however the inspecting conveyance of ln(OR) is asymptotically typical with SE = SQRT(1/a + 1/b + 1/c +1/d) where a-d are the 4 cell tallies in the 2x2 table.
• There are two principle motivations to utilize logarithmic scales in diagrams and charts. The first is to react to skewness towards substantial qualities; i.e., cases in which one or a couple of focuses are a lot bigger than the greater part of the information.
• The second is to indicate percent change or multiplicative variables.