Graph of predicted probabilities using mlogit

[Lara ingram]

Hello,

I am trying to replicate figure 3 in the paper by Ho & Park (2019) – I have attached an image of the graph. I am running a multinomial logistic regression and I would like the predicted probabilities of employment transitions by living arrangement transitions.

My outcome has four categories (consistently employed (base category), unemployed-employed, employed-unemployed and consistently employed). My living arrangement is also categorical (no change, both to one and both to neither). This is what I have managed to produce, however, I am unable to format the graph so it looks like Ho & Parks.

svyset Fpsu [pw = panel_wt_12_18 ], strata(gov1)
svy: mlogit change2 i.releducLG i.matempLG  i.fatempLG i.coresidenceCH, rrr
margins coresidenceCH
mplotoffset, recast(scatter) offset(0.1)

Attached are the two graphs (the one I produced in Stata and the one by Ho & Park). In the graph by Ho & park, predicted probabilities for the base category are plotted on the left vertical axis and the predicted probabilities for the remaining three categories are plotted on the right vertical axis.


Article:
Ho, P., & Park, H. (2019). Young Adults' Patterns of Leaving the Parental Home: A Focus on Differences Among Asian Americans. Journal of Marriage and Family, 81(3), 696-712.


Thank you.

[Lara ingram]

Forgot the data example:

* Example generated by -dataex-. To install: ssc install dataex
clear
input double Findid float(change2 coresidenceCH matempLG fatempLG)
601004004 . . . .
601004004 0 0 0 1
601004603 . . . .
601004603 0 0 0 1
601005902 . . . .
601005902 3 0 0 1
601012406 . . . .
601012406 0 0 0 1
601020204 . . . .
601020204 0 2 1 1
601029206 . . . .
601029206 0 0 0 1
601032102 . . . .
601032102 2 0 0 1
601032309 . . . .
601032309 0 0 0 1
601037505 . . . .
601037505 0 0 0 1
601038302 . . . .
601038302 0 0 0 1
601045003 . . . .
601045003 0 0 0 1
601050804 . . . .
601050804 0 0 0 1
601051404 . . . .
601051404 0 0 0 1
601054402 . . . .
601054402 0 0 0 1
601055607 . . . .
601055607 1 0 0 1
601056302 . . . .
601056302 0 0 0 1
601058302 . . . .
601058302 3 0 1 1
601058602 . . . .
601058602 0 0 0 1
601060002 . . . .
601060002 1 0 0 1
601060202 . . . .
601060202 0 0 0 1
601060302 . . . .
601060302 0 0 1 1
601061602 . . . .
601061602 0 0 1 1
601062002 . . . .
601062002 0 0 0 1
601062602 . . . .
601062602 0 0 0 1
601063102 . . . .
601063102 1 0 0 1
601067002 . . . .
601067002 2 0 1 1
601072002 . . . .
601072002 1 0 0 1
601072702 . . . .
601072702 0 0 1 1
601072902 . . . .
601072902 1 0 0 1
601073102 . . . .
601073102 0 0 0 1
601074502 . . . .
601074502 1 0 0 1
601075902 . . . .
601075902 0 0 0 1
601077402 . . . .
601077402 0 0 0 1
601078802 . . . .
601078802 3 0 0 1
601080302 . . . .
601080302 1 0 0 1
601082202 . . . .
601082202 0 0 0 1
601083402 . . . .
601083402 0 0 0 1
601085202 . . . .
601085202 0 0 0 1
601085602 . . . .
601085602 0 0 0 1
601085703 . . . .
601085703 0 0 0 1
601086402 . . . .
601086402 0 0 0 1
601088002 . . . .
601088002 0 0 0 1
601088202 . . . .
601088202 0 0 0 1
601088305 . . . .
601088305 0 2 0 1
601088306 . . . .
601088306 0 0 0 1
601089002 . . . .
601089002 0 0 1 1
601089303 . . . .
601089303 3 0 1 1
601090502 . . . .
601090502 3 0 0 1
601090602 . . . .
601090602 0 0 0 1
601091602 . . . .
601091602 0 0 0 1
end
label values change2 change2
label def change2 0 "Consistently unemployed", modify
label def change2 1 "Unemployed - employed", modify
label def change2 2 "Employed - unemployed", modify
label def change2 3 "Consistently employed", modify
label values coresidenceCH coresCH
label def coresCH 0 "No change", modify
label def coresCH 2 "Both - Neither", modify
label values matempLG memp
label def memp 0 "Unemployed", modify
label def memp 1 "Employed", modify
label values fatempLG femp
label def femp 1 "Employed", modify