Compute and re-code into new variables in JASP

PerPalmgren

Hi.

Could somebody help me🙏.

How do I go about to:

a) create a new nominal (binary) variable from a scale variable. So, I have an age variable (denoted as year of age) and I want to dichotomize this to a nominal variable with two levels, thus, under or equal to the mean age =1 and over the mean age = 2.

b) collapse an ordinal scale (Likert) with five steps to an ordinal scale with three steps. Thus, I want 1 & 2 =1, 3=2 and 4 & 5 =3.

Grateful for some help :-)

Per

EJ

For others who may land here: in the meantime this was addressed at https://github.com/jasp-stats/jasp-issues/issues/1844

E.J.

PerPalmgren

Hi@vandenman

You helped me earlier on the GitHub post a above (it is closed now) as I wanted to create a new variable where I cluster 5-level data to 3-level data. Thus, 1 & 2 =1, 3=2 and 4 & 5 =3. 

You gave me a great tip with the following R code (that I have used so many times since, despite I hate syntax and coding):

ifElse(facFive <= 2, 1, 

ifElse(facFive <= 3, 2, 3)

)

Could you just explain the mechanics behind the codes. I can’t get my head around it. In my logic I ask the program to:

• If the variable is less or equal to 2, make it a 1
• If the variable is equal to 3, make it a 2
• If the variable is more or equal to 4, make it a 3

Could you explain the like I was a kid?🤪🙏🏻

All the best

Per

vandenman

Hi Per,

No worries! The key is to realize that the two statements are nested, so it is actually one statement like so:

ifElse(facFive <= 2, 1, ifElse(facFive <= 3, 2, 3))

We can also see this as a decision tree, like below:

You start at the top and ask the question, is this value <= 2? If yes, you go left, and if no you go right. If you go left, you end up with 1 and are done. If no, you ask another question, is this value <= 3? If yes, you go left and end up at 2. If no then you go right and end up at 3. In general, you repeat this until there are no more questions and you are at a value. We repeat this process for each row in the dataset.

I hope this makes things more understandable, but let me know if anything else is unclear!

Don

PerPalmgren

@vandenman

Thanks Don🙏🏻. That was extremely helpful😍.

All the best

Per😎