Creating country specific linear and quadratic time trends

[Himani Srihan]

Hi everyone,

I was trying to include linear and quadratic id specific time trends in my model (separately in two different models). I used the following code. Can anyone confirm if I did the correct code? Any help would be highly appreciated:

Linear time trend: i.Country##c.Year
Quadratic time trend: ***Generating variable for quadratic time trend.
egen Year1=group(Year)
gen Year2=Year1^2
i.Country1##c.Year2

One additional question I was facing was, how will the results differ if I use one "#" instead of two "##"? Thank you!

[Maarten Buis]

No, country specific quadratic trends: i.Country##c.Year##c.Year

[Himani Srihan]

No, country specific quadratic trends: i.Country##c.Year##c.Year

Hi, Thanks for your response! Why do we need to include "##"? this would give me an interaction of Country1#c.Year. However, I just need need the quadratic time trends. So should both the linear and quadratic time trends not contain only "#" insteadof "##" ?

[Joseph Coveney]

Why do we need to include "##"?. . . I just need need the quadratic time trends. So should both the linear and quadratic time trends not contain only "#" insteadof "##" ?

You know, you can try things out for yourself easily enough in Stata, creating compact toy examples. See an illustration below. (I tried to read your mind as to what you're suggesting as the alternative to what Maarten recommended. If it's not quite what you had in mind, then you can easily modify the code or create new code that does do what you're trying to suggest.)

.ÿ
.ÿversionÿ16.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿsetÿseedÿ`=strreverse("1581797")'

.ÿ
.ÿquietlyÿsetÿobsÿ20

.ÿgenerateÿbyteÿcidÿ=ÿmod(_n,ÿ2)

.ÿ
.ÿbysortÿcid:ÿgenerateÿbyteÿtimÿ=ÿ_n

.ÿ
.ÿgenerateÿdoubleÿoutÿ=ÿ///
>ÿÿÿÿÿÿÿÿÿ(cidÿ-ÿ0.5)ÿ*ÿ2ÿ+ÿ///
>ÿÿÿÿÿÿÿÿÿ(cidÿ-ÿ0.5)ÿ*ÿ(timÿ-ÿ5.5)ÿ+ÿ///
>ÿÿÿÿÿÿÿÿÿ(cidÿ-ÿ0.5)ÿ*ÿ(timÿ-ÿ5.5)ÿ*ÿ(timÿ-ÿ5.5)ÿ+ÿ///
>ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿrnormal()

.ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ
.ÿ*
.ÿ*ÿBeginÿhere
.ÿ*
.ÿ//ÿMaarten'sÿrecommendation
.ÿregressÿoutÿi.cid##c.tim##c.tim

ÿÿÿÿÿÿSourceÿ|ÿÿÿÿÿÿÿSSÿÿÿÿÿÿÿÿÿÿÿdfÿÿÿÿÿÿÿMSÿÿÿÿÿÿNumberÿofÿobsÿÿÿ=ÿÿÿÿÿÿÿÿ20
-------------+----------------------------------ÿÿÿF(5,ÿ14)ÿÿÿÿÿÿÿÿ=ÿÿÿÿ110.31
ÿÿÿÿÿÿÿModelÿ|ÿÿ825.864045ÿÿÿÿÿÿÿÿÿ5ÿÿ165.172809ÿÿÿProbÿ>ÿFÿÿÿÿÿÿÿÿ=ÿÿÿÿ0.0000
ÿÿÿÿResidualÿ|ÿÿ20.9622647ÿÿÿÿÿÿÿÿ14ÿÿ1.49730462ÿÿÿR-squaredÿÿÿÿÿÿÿ=ÿÿÿÿ0.9752
-------------+----------------------------------ÿÿÿAdjÿR-squaredÿÿÿ=ÿÿÿÿ0.9664
ÿÿÿÿÿÿÿTotalÿ|ÿÿ846.826309ÿÿÿÿÿÿÿÿ19ÿÿ44.5698058ÿÿÿRootÿMSEÿÿÿÿÿÿÿÿ=ÿÿÿÿ1.2236

---------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿoutÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿtÿÿÿÿP>|t|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
----------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿ1.cidÿ|ÿÿÿ24.70432ÿÿÿ2.035324ÿÿÿÿ12.14ÿÿÿ0.000ÿÿÿÿÿ20.33898ÿÿÿÿ29.06965
ÿÿÿÿÿÿÿÿÿÿÿÿtimÿ|ÿÿÿ4.269738ÿÿÿ.6010669ÿÿÿÿÿ7.10ÿÿÿ0.000ÿÿÿÿÿ2.980578ÿÿÿÿ5.558898
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿcid#c.timÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿ1ÿÿ|ÿÿ-9.217564ÿÿÿÿ.850037ÿÿÿ-10.84ÿÿÿ0.000ÿÿÿÿ-11.04071ÿÿÿ-7.394416
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿc.tim#c.timÿ|ÿÿ-.4414601ÿÿÿ.0532523ÿÿÿÿ-8.29ÿÿÿ0.000ÿÿÿÿ-.5556749ÿÿÿ-.3272454
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
cid#c.tim#c.timÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿ1ÿÿ|ÿÿÿ.9435872ÿÿÿ.0753101ÿÿÿÿ12.53ÿÿÿ0.000ÿÿÿÿÿ.7820631ÿÿÿÿ1.105111
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿ_consÿ|ÿÿÿ-11.0878ÿÿÿ1.439191ÿÿÿÿ-7.70ÿÿÿ0.000ÿÿÿÿ-14.17456ÿÿÿÿ-8.00104
---------------------------------------------------------------------------------

.ÿpredictÿdoubleÿxb,ÿxb

.ÿ
.ÿgraphÿtwowayÿ///
>ÿÿÿÿÿÿÿÿÿscatterÿoutÿtimÿifÿ!cid,ÿmcolor(red)ÿmsize(small)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿlineÿxbÿtimÿifÿ!cid,ÿlcolor(red)ÿlpattern(dash)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿscatterÿoutÿtimÿifÿcid,ÿmcolor(blue)ÿmsize(small)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿlineÿxbÿtimÿifÿcid,ÿlcolor(red)ÿlpattern(dash)ÿ,ÿ///
>ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿylabel(ÿ,ÿangle(horizontal)ÿnogrid)ÿlegend(off)

.ÿquietlyÿgraphÿexportÿCorrect.png,ÿreplace

.ÿ
.ÿ//ÿCompareÿyours
.ÿdropÿxb

.ÿregressÿoutÿi.cid#c.tim#c.tim

ÿÿÿÿÿÿSourceÿ|ÿÿÿÿÿÿÿSSÿÿÿÿÿÿÿÿÿÿÿdfÿÿÿÿÿÿÿMSÿÿÿÿÿÿNumberÿofÿobsÿÿÿ=ÿÿÿÿÿÿÿÿ20
-------------+----------------------------------ÿÿÿF(2,ÿ17)ÿÿÿÿÿÿÿÿ=ÿÿÿÿÿ21.15
ÿÿÿÿÿÿÿModelÿ|ÿÿ604.022178ÿÿÿÿÿÿÿÿÿ2ÿÿ302.011089ÿÿÿProbÿ>ÿFÿÿÿÿÿÿÿÿ=ÿÿÿÿ0.0000
ÿÿÿÿResidualÿ|ÿÿ242.804131ÿÿÿÿÿÿÿÿ17ÿÿ14.2825959ÿÿÿR-squaredÿÿÿÿÿÿÿ=ÿÿÿÿ0.7133
-------------+----------------------------------ÿÿÿAdjÿR-squaredÿÿÿ=ÿÿÿÿ0.6795
ÿÿÿÿÿÿÿTotalÿ|ÿÿ846.826309ÿÿÿÿÿÿÿÿ19ÿÿ44.5698058ÿÿÿRootÿMSEÿÿÿÿÿÿÿÿ=ÿÿÿÿ3.7792

---------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿoutÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿtÿÿÿÿP>|t|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
----------------+----------------------------------------------------------------
cid#c.tim#c.timÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿ0ÿÿ|ÿÿ-.1081245ÿÿÿ.0310055ÿÿÿÿ-3.49ÿÿÿ0.003ÿÿÿÿ-.1735404ÿÿÿ-.0427086
ÿÿÿÿÿÿÿÿÿÿÿÿÿ1ÿÿ|ÿÿÿ.1102438ÿÿÿ.0310055ÿÿÿÿÿ3.56ÿÿÿ0.002ÿÿÿÿÿ.0448279ÿÿÿÿ.1756597
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿ_consÿ|ÿÿÿ.5266589ÿÿÿ1.311959ÿÿÿÿÿ0.40ÿÿÿ0.693ÿÿÿÿ-2.241332ÿÿÿÿÿ3.29465
---------------------------------------------------------------------------------

.ÿpredictÿdoubleÿxb,ÿxb

.ÿ
.ÿgraphÿtwowayÿ///
>ÿÿÿÿÿÿÿÿÿscatterÿoutÿtimÿifÿ!cid,ÿmcolor(red)ÿmsize(small)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿlineÿxbÿtimÿifÿ!cid,ÿlcolor(red)ÿlpattern(dash)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿscatterÿoutÿtimÿifÿcid,ÿmcolor(blue)ÿmsize(small)ÿ||ÿ///
>ÿÿÿÿÿÿÿÿÿlineÿxbÿtimÿifÿcid,ÿlcolor(red)ÿlpattern(dash)ÿ,ÿ///
>ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿylabel(ÿ,ÿangle(horizontal)ÿnogrid)ÿlegend(off)

.ÿquietlyÿgraphÿexportÿIncorrect.png,ÿreplace

.ÿ
.ÿexit

endÿofÿdo-file


.

Exploring things on your own can give you insights that are more difficult to apprehend when they are spoon-fed by others.