Missing values

[lucysbridge]

Hi,

I already imported a data set into Jamovi and started data cleaning etc. I have now realised before I start analysis I need to add in missing values as I currently just have empty cells. I have tried setting the missing values in the option bar before just importing the .omv file I have been working on, when I do this none of the empty cells take the new value. How can I just assign a missing value across the entire data set

Hope this makes sense,
Thanks
Lucy

[jonathon]

hi,

so normally, with missing values, you might load in a .csv file, and it might have all these -99 values everywhere. in this situation, you want jamovi to treat these -99 values as though there's no value there at all. so you can using the missing value system to set this up.

however, it doesn't sounds like this is your situation. if you've got empty cells, then those are missing values, and jamovi will treat them as such. so i don't think you need to do anything here -- if i understand correctly. those empty cells will be treated as missing values, and the analyses will handle them correctly without you needing to do anything.

cheers

jonathon

[lucysbridge]

Thanks Jonathon,

That is exactly the situation, I thought I would need to put a value to them as I tried using a SUM formula to add a couple of scales together and it was not doing anything, as in I would press enter and nothing happens. Do you know why else this could be occurring?

Lucy

[jonathon]

so in statistics, you can think of a 'missing value' as a "don't know". so if you're summing two values, and one of them is "don't know", can you see that the result will be "don't know" as well.

however, we can pass an argument to the SUM() function which is ignore_missing. take a look at the following screenshot

Screen Shot 2021-03-02 at 21.49.27.png (594.52 KiB) Viewed 33772 times

does that address the issue?

kind regards

jonathon

[leehw]

In SPSS it's possible to stipulate how many valid numbers there must be in the formula for it to work. Does Jamovi have anything similar?

E.g. if we want to still take a mean of 10 variables even if 1 or 2 of them are missing, but to return MISSING if any more than 2 are missing, we can do this kind of thing...

Code: Select all

MEAN.8(Vbl1, Vbl2, Vbl3, Vbl4, Vbl5, Vbl6, Vbl7, Vbl8, Vbl9, Vbl10)

It allows pro-rating for a certain percentage of missing data on psychometric scales where that's the approach suggested by the scale developers.

Is there any similar approach in Jamovi?

[MAgojam]

Hey @leehw,
you were looking for something like this:

tempsnip.png (51.33 KiB) Viewed 32265 times

Cheers,
Maurizio

[Bobafett]

That's a very nifty solution, Maurizio. I'll be using that feature in future!
I'd like to achieve something similar with SUM scores - is there a way to calculate a pro rata version?

I haven't looked in while, but does jamovi have any solution to missing data - like mean imputation?

Cheers!

[MAgojam]

jamovi have any solution to missing data - like mean imputation?

You could combine it like this:

Code: Select all

IFMISS(myvarname, VMEAN(myvarname), myvarname)

Cheers,
Maurizio

[jonathon]

this blog post describe mean imputation:

https://blog.jamovi.org/2018/10/23/transforming-variables.html

cheers

[Bobafett]

God I LOVE this forum!!!!!

Thank you so much for the tips, much appreciated