1
$\begingroup$

I do a ANOVA, when I do the normality test, the data has too many zeros, how to make the data to normality using the transformation? enter image description here

$\endgroup$
2
  • 1
    $\begingroup$ Welcome to CV. You might want to consider expanding the question from asking about a transformation to asking about more appropriate models for zero-heavy data. $\endgroup$
    – user78229
    Nov 15, 2015 at 11:58
  • $\begingroup$ Thank you!@DJohnson, so I can't do ANOVA with the data? $\endgroup$
    – LuYan
    Nov 15, 2015 at 12:13

1 Answer 1

6
$\begingroup$

First, the data in an ANOVA does not have to be normal. The errors (as estimated by the residuals) do.

However, when the dependent variable has a lot of 0's, the residuals will not be normal.

Second, no transformation will get rid of excess 0s. You can't use ANOVA on these data. You should use a zero inflated regression such as zero inflated Poisson or zero inflated negative binomial.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.