geom_jitter: is it allowed?

[Nana]

Hi guys,

I'm very new in R and currently learning geom_jitter. I've read many sources about this geom but still can't understand how it is allowed to add noise in your data.

For example, in these picture, OP explains that the bottom plot is using geom_jitter thus avoiding overplotting. But from what I understand, geom_jitter makes the plot changes the data point. In the geom_point, there is no x 6,1 and y 20,1 but in geom_jitter there is.

My question is: is it allowed?

Sorry if it sounds stupid, I learned basic statistic in college but never about jitter.

Thanks so much ^^

[Ron Crump]

Hi Nana,

> I'm very new in R and currently learning geom_jitter. I've read many
> sources about this geom but still can't understand how it is allowed to
> add noise in your data.

> pls_open.png
>
> For example, in these picture, OP explains that the bottom plot is using
> geom_jitter thus avoiding overplotting. But from what I understand,
> geom_jitter makes the plot changes the data point. In the geom_point,
> there is no x 6,1 and y 20,1 but in geom_jitter there is.
>
> My question is: is it allowed?

Yes.

> Sorry if it sounds stupid, I learned basic statistic in college but
> never about jitter.

It isn't a statistical technique, it's a presentational one.

Providing it is used with care - eg as in the example you presented
with the jitter applied to number of cylinders - so that it gives an
indication of the number of observations at each value of hwy, and does
not infuence the variability in hwy, then it is fine. There are
alternatives - you could colour each value based on the number
of observations at that unique point, or vary the size. But often those
would still overlap a lot.

Regards,
Ron.

PS There isn't much traffic on this list these days, with
community.rstudio.com being the preferred support forum for all
things tidyverse-related, including ggplot2, so I'd recommend
asking future questions over there.

[Nana]

Hi Ron,

Thanks for taking your time to reply - it's such a helpful explanation.

I followed your suggestion and made an account in the RStudio community. 

Thanks for being kind,

Nana