Twoway scatter + rcap graph

[Federica Di Marcantonio]

Dear All

I ma trying to make a graph with mean and CF of 3 vars by country and product.
I have generated the following code

foreach v in mean ub lb {
    gen `v'1 = .
    gen `v'2 = .
    gen `v'3 = .
    gen `v'4 = .
}

forval j = 1/3 {
    ci mean mSPS_`j' if product== 1 & coun==2  
    quietly {
        replace mean1 = r(mean) in `j'
        replace ub1 = r(ub) in `j'
        replace lb1 = r(lb) in `j'
    }
    
    ci mean mSPS_`j' if product== 2 & coun==2  
    quietly {
        replace mean2 = r(mean) in `j'
        replace ub2 = r(ub) in `j'
        replace lb2 = r(lb) in `j'
    }
    
    ci mean mSPS_`j' if product== 1 & coun==3  
    quietly {
        replace mean3 = r(mean) in `j'
        replace ub3 = r(ub) in `j'
        replace lb3 = r(lb) in `j'
    }
    
    ci mean mSPS_`j' if product== 2 & coun==3  
    quietly {
        replace mean4 = r(mean) in `j'
        replace ub4 = r(ub) in `j'
        replace lb4 = r(lb) in `j'
    }
    local call `call' `j' "mSPS_`j'"  
}




twoway scatter mean1 prod, mc(blue)  ms(O) || rcap lb1 ub1 prod, lc(blue) ///
|| scatter mean2 prod, mc(red) ms(Th) || rcap lb2 ub2 prod, lc(red) ///
|| scatter mean3 prod, mc(red) ms(Th) || rcap lb3 ub3 prod, lc(red) ///
|| scatter mean3 prod, mc(red) ms(Th) || rcap lb4 ub4 prod, lc(red) ///
xla(`call', tlc(none)) ytitle(means and 95% confidence intervals) by(country)

I would like to get a graph similar to this one (though i prefer to group by country and then for each country display the two products.

Could you please help me to produce the right graph?
thanks
Federica

[Nick Cox]

You may need to arrange offsets.

We need a data example to give good advice on code.

dataex mean? lb? ub? in 1/3

but also details on what are products 1 and 2 and countries 2 and 3, etc., etc.

Please show the results of

tab product, nola

tab product

tab coun, nola

tab coun

describe MSPS_?

[Federica Di Marcantonio]

Hi Nick,
here an example from my dataset. mSPS_* are scores from 0 to 10.

* Example generated by -dataex-. To install: ssc install dataex
clear
input int id byte country float(product mSPS_1 mSPS_2 mSPS_3)
 1 2 1  6  7 10
 1 2 1  4  2 10
 1 2 2  7 10  4
 1 2 2  4  7  8
 2 2 1  9  1  4
 2 2 1  5  5  1
 2 2 2  6  8  2
 2 2 2  3  2  6
 3 2 1  8  7  3
 3 2 1  6  5  4
 3 2 2  9  6  3
 3 2 2  2  7  3
 4 2 1  3  6  5
 4 2 1  3  2  4
 4 2 2  6  5  7
 4 2 2  1  1  2
 5 2 1  6  9  5
 5 2 1  5  7  3
 5 2 2  8  2  4
 5 2 2  6  7  3
 6 2 1  2  3  2
 6 2 1  2  2  3
 6 2 2  2  5  2
 6 2 2  2  2  3
 7 2 1  8  3  5
 7 2 1  3  1  8
 7 2 2  9  7  4
 7 2 2  5 10  1
 8 2 1  6  8  7
 8 2 1  6  4  8
 8 2 2  6  9  7
 8 2 2  4  6  5
 9 2 1  7  7  7
 9 2 1  9  3  8
 9 2 2  9  9  9
 9 2 2  9  9  9
10 2 1  6  5  7
10 2 1  5  6  4
10 2 2  8  5  8
10 2 2  4  3  5
11 2 1  8  2  4
11 2 1  5  1  2
11 2 2  8  6  2
11 2 2  7  2  4
12 2 1  5  5  8
12 2 1  5  6  8
12 2 2  7  4  6
12 2 2  7  3  8
13 2 1  6  8  5
13 2 1  3  6  7
13 2 2  6  4  7
13 2 2  7  5  6
14 3 1  8  6  8
14 3 1  7  5  7
14 3 2  5  4  6
14 3 2  5  5  6
15 3 1  8  4  6
15 3 1  3  3  6
15 3 2 10  7  5
15 3 2  6  9  7
16 3 1  9 10  9
16 3 1  8  6  9
16 3 2  6  6  8
16 3 2  7  8  7
17 3 1  2  6  3
17 3 1  4  2  1
17 3 2  7  7  4
17 3 2  5  6  3
18 3 1  8  9 10
18 3 1 10  9 10
18 3 2  9  9  9
18 3 2  9  9 10
19 3 1  5  5  3
19 3 1  3  4  4
19 3 2  6  7  6
19 3 2  8  8  9
20 3 1  5  7  6
20 3 1  4  5  6
20 3 2  4  5  6
20 3 2  7  4  1
21 3 1  5  6  9
21 3 1  8  5  6
21 3 2  4  7  9
21 3 2  9  7  5
22 3 1  7  9  7
22 3 1  8  5 10
22 2 2 10  8  9
22 3 2  8  8  7
23 3 1  5  5  5
23 3 1  5  3  6
23 3 2  5  6  3
23 3 2  4  5  5
24 3 1  6  9  7
24 3 1  5  6  7
24 3 2  5  6  8
24 3 2  7  5  6
25 3 1  3  5  2
25 3 1  3  2  6
25 3 2 10  9  9
25 3 2  5  3  7
end
label values country country
label def country 2 "DE" 3 "HU", modify
label values product product
label def product 1 "Prod 1", modify
label def product 2 "Prod 2", modify

product	Freq.	Percent	Cum.
1	1,600	50.00	50.00
2	1,600	50.00	100.00
Total	3,200	100.00

product	Freq.	Percent	Cum.
Prod 1	1,600	50.00	50.00
Prod 2	1,600	50.00	100.00
Total	3,200	100.00

Co	Freq.	Percent	Cum.
Germany	1,600	50.00	50.00
Hungury	1,600	50.00	100.00
Total	3,200	100.00

	storage	display	value
variable	name	type	format	label	variable	label
mSPS_1	float	%9.0g	mSPS_DE
mSPS_2	float	%9.0g	mSPS_HU
mSPS_3	float	%9.0g	mSPS_LI

Thanks for helping
F

[Nick Cox]

Could you confirm that mSPS_1 refers to DE (Germany) even though that is country 2 and similarly that mSPS_2 refers to Hungary (not Hungury, by the way) even though that is country 3?

[Nick Cox]

This may help. Using #1 as well as #3 I have made some guesses, including about what you ask for but don't really want and vice versa.

See also https://journals.sagepub.com/doi/pdf...867X1001000112

* Example generated by -dataex-. To install: ssc install dataex
clear
input int id byte country float(product mSPS_1 mSPS_2 mSPS_3)
 1 2 1  6  7 10
 1 2 1  4  2 10
 1 2 2  7 10  4
 1 2 2  4  7  8
 2 2 1  9  1  4
 2 2 1  5  5  1
 2 2 2  6  8  2
 2 2 2  3  2  6
 3 2 1  8  7  3
 3 2 1  6  5  4
 3 2 2  9  6  3
 3 2 2  2  7  3
 4 2 1  3  6  5
 4 2 1  3  2  4
 4 2 2  6  5  7
 4 2 2  1  1  2
 5 2 1  6  9  5
 5 2 1  5  7  3
 5 2 2  8  2  4
 5 2 2  6  7  3
 6 2 1  2  3  2
 6 2 1  2  2  3
 6 2 2  2  5  2
 6 2 2  2  2  3
 7 2 1  8  3  5
 7 2 1  3  1  8
 7 2 2  9  7  4
 7 2 2  5 10  1
 8 2 1  6  8  7
 8 2 1  6  4  8
 8 2 2  6  9  7
 8 2 2  4  6  5
 9 2 1  7  7  7
 9 2 1  9  3  8
 9 2 2  9  9  9
 9 2 2  9  9  9
10 2 1  6  5  7
10 2 1  5  6  4
10 2 2  8  5  8
10 2 2  4  3  5
11 2 1  8  2  4
11 2 1  5  1  2
11 2 2  8  6  2
11 2 2  7  2  4
12 2 1  5  5  8
12 2 1  5  6  8
12 2 2  7  4  6
12 2 2  7  3  8
13 2 1  6  8  5
13 2 1  3  6  7
13 2 2  6  4  7
13 2 2  7  5  6
14 3 1  8  6  8
14 3 1  7  5  7
14 3 2  5  4  6
14 3 2  5  5  6
15 3 1  8  4  6
15 3 1  3  3  6
15 3 2 10  7  5
15 3 2  6  9  7
16 3 1  9 10  9
16 3 1  8  6  9
16 3 2  6  6  8
16 3 2  7  8  7
17 3 1  2  6  3
17 3 1  4  2  1
17 3 2  7  7  4
17 3 2  5  6  3
18 3 1  8  9 10
18 3 1 10  9 10
18 3 2  9  9  9
18 3 2  9  9 10
19 3 1  5  5  3
19 3 1  3  4  4
19 3 2  6  7  6
19 3 2  8  8  9
20 3 1  5  7  6
20 3 1  4  5  6
20 3 2  4  5  6
20 3 2  7  4  1
21 3 1  5  6  9
21 3 1  8  5  6
21 3 2  4  7  9
21 3 2  9  7  5
22 3 1  7  9  7
22 3 1  8  5 10
22 2 2 10  8  9
22 3 2  8  8  7
23 3 1  5  5  5
23 3 1  5  3  6
23 3 2  5  6  3
23 3 2  4  5  5
24 3 1  6  9  7
24 3 1  5  6  7
24 3 2  5  6  8
24 3 2  7  5  6
25 3 1  3  5  2
25 3 1  3  2  6
25 3 2 10  9  9
25 3 2  5  3  7
end
label values country country
label def country 2 "DE" 3 "HU", modify
label values product product
label def product 1 "Prod 1", modify
label def product 2 "Prod 2", modify

* my code starts here 

save product_country, replace 

forval j = 1/3 { 
    use product_country, clear 
    statsby , by(country product) : ci mean mSPS_`j'
    gen version = `j'
    save results`j', replace 
}

append using results2 results1 

list, sepby(version country)

label def version 1 "DE version" 2 "HU version" 3 "LI version"
label val version version

foreach v in mean ub lb { 
    separate `v', by(version) veryshortlabel 
}

gen productL = product - 0.2 
gen productR = product + 0.2 


twoway scatter mean1 productL, by(country, note("")) mc(blue)  ms(O) || rcap lb1 ub1 productL, lc(blue) ///
|| scatter mean2 product, mc(red) ms(Th) || rcap lb2 ub2 product, lc(red) ///
|| scatter mean3 productR, mc(black) ms(Dh) || rcap lb3 ub3 productR, lc(black) ///
xla(1 2, tlc(none) valuelabel) xtitle(product) ytitle(means and 95% confidence intervals) legend(order(1 3 5))

[Federica Di Marcantonio]

Hi Nick,

thanks a lot. This is exactly what I was trying to do.
For the readers I confirm that your assumptions are correct. There are two countries DE and HU and two products (let's say product 1 and 2). For each product we have 3 scores representing a rating for the DE version (mSPS_1), HU version (mSPS_2) and LI version (mSPS_3).

Thanks
Federica

[Nick Cox]

Good!

On offsets see also https://journals.sagepub.com/doi/pdf...867X0700700112

[Federica Di Marcantonio]

Ok thanks
F