Line graph with interquartile ranges as shaded area rather than error bars

[Kate Harris]

Hello, I wondered if anyone can help. I followed some guidance on here to generate new variables for the median and IQR.
I now want to generate a line graph for three variables (possibly more), each are showing the median by year. I want to add the interquartile range as a shaded area for two (mcorn and mderby) rather than as error bars (tried in excel but if the series' intersect it won't show both).
I have created variables for the 25th and 75th percentile for both 'mcorn' and 'mderby' but I don't know how to plot these. Please let me know if this is possible?!

the line graph is as follows: twoway (line medianvar1 mcorn mderby year, sort)

Thank you in advance!

[Scott Merryman]

Like this:

webuse grunfeld,clear
collapse (p50) invest_50=  invest (p75) invest_75 = invest (p25) invest_25=invest (p50) /// 
    kstock_50=  kstock (p75) kstock_75 = kstock (p25) kstock_25=kstock, by(year)
twoway rarea invest_75 invest_25 year, fcolor(gs8) ///
    || rarea kstock_75 kstock_25 year, color(blue*.25) /// 
    || line invest_50 year, lc(black) lw(medium)  /// 
    || line kstock_50 year, lc(blue) lw(medium) /// 
    || line invest_25 year, lc(gs8) lw(medium) /// 
    || line invest_75 year, lc(gs8) lw(medium) /// 
    ||, legend(off)

[Kate Harris]

Yes exactly like that thank you Scott! I will try it out on my data now,

[Nazzarena]

This is excellent, thanks, but it's a bit cumbersome to collapse if you have multiple "by" covariates.
As a rough alternative, when using a continuous x variable, is there a way to add IQR to the median band plot?
Thanks!

[Jannis Keller]

Very helpful thanks!!!