Measure type allocation for CSV files

Discuss the jamovi platform, possible improvements, etc.
Post Reply
prog_frog
Posts: 6
Joined: Mon Jan 11, 2021 6:26 pm

Measure type allocation for CSV files

Post by prog_frog »

Hi!

One thing I've noticed in Jamovi is that when opening a CSV file, variables that have whole numbers (no decimals) tend to be set as Nominal, even if those whole numbers span a large range or extend into quite large numbers, whereas numerical variables with decimals tend to be set as Continuous. In my experience, setting variables with whole numbers as Nominal is almost always wrong, and it's a bit tedious to have to manually change most or all of my continuous variables from Nominal to Continuous.

Is there a reason why Jamovi does this? I would love to see a revision to this so that unless the values are limited to a very small number of whole numbers (e.g., 0 and 1), Jamovi defaults to Continuous instead. But I'm curious what others' experiences have been too.

Thanks!
~Doug
User avatar
jonathon
Posts: 2609
Joined: Fri Jan 27, 2017 10:04 am

Re: Measure type allocation for CSV files

Post by jonathon »

hi doug,

jamovi's behaviour is the same as spss ... if there's a large number of unique integer values (>50 i think), it will treat it as continuous, otherwise nominal.

the spss philosophy, continued with jamovi, is that you don't really need to worry about measure type if you don't want to. you can use nominal integer variables anywhere you can use a continuous variable. the measure types are more of a guide, and you can ignore them if you like.

cheers

jonathon
prog_frog
Posts: 6
Joined: Mon Jan 11, 2021 6:26 pm

Re: Measure type allocation for CSV files

Post by prog_frog »

Thanks, Jonathon!

The biggest reason I ask is because in teaching statistics I think it's helpful for students to apply the concepts they're learning in class, and so I'm having to tell them to check the measure type for their variables to ensure they're accurate. I worry that telling them it doesn't matter and they can ignore it will cause more confusion.

A related question - is there a reason why it's not possible to get a frequency table for continuous variables? Perhaps to prevent overly long tables if there are many, many values? I've occasionally had to change measure type to an inaccurate type (nominal or ordinal) to 'trick' Jamovi into letting me get a frequency table.

Overall, I've been very happy with the switch from SPSS to Jamovi in my undergraduate statistics course, especially during remote teaching, so thanks to you and the entire team for your continued work on it!

Thanks again!
~Doug
User avatar
jonathon
Posts: 2609
Joined: Fri Jan 27, 2017 10:04 am

Re: Measure type allocation for CSV files

Post by jonathon »

> Perhaps to prevent overly long tables if there are many, many values?

yes, mostly. most of the time you don't want frequency tables for continuous variables.

> Overall, I've been very happy with the switch from SPSS to Jamovi in my undergraduate statistics course, especially during remote teaching, so thanks to you and the entire team for your continued work on it!

great to hear. you can always write us a testimonial for our testimonials page.

cheers

jonathon
User avatar
MAgojam
Posts: 421
Joined: Thu Jun 08, 2017 2:33 pm
Location: Parma (Italy)

Re: Measure type allocation for CSV files

Post by MAgojam »

Hi, @prog_frog.
jonathon wrote:> Perhaps to prevent overly long tables if there are many, many values?
yes, mostly. most of the time you don't want frequency tables for continuous variables.
Jonathon has already answered you, but allow me a possible suggestion, if you need to show the frequencies of the values of a continuous variable without intervening on "Measurement type" to 'trick' Jamovi, try to take a look among the modules of the library.
The "Rosetta" module has a "Frequencies" independent of the type of measurement.

Cheers,
Maurizio
sbalci
Posts: 124
Joined: Sat Jan 06, 2018 10:25 pm
Contact:

Re: Measure type allocation for CSV files

Post by sbalci »

The "Rosetta" module has a "Frequencies" independent of the type of measurement.
Similarly "tableone" from ClinicoPath Descriptives shows frequencies of continuous variables if table type is janitor. It is not esthetic but I use that output to see the data.
Post Reply