[Nick Cox]

set obs isn't relevant to the case where your data are already in memory. It's only needed because I create a fake dataset.

[Lucas Bordure]

If I delete set obs, the code becomes this:

*Fixed effects: Heterogeneity across entities (stripplot)
clear
set seed 2803

gen performance = exp(rnormal(-2, 1))
gen type_industry = ceil(_n/ 100)

label def type_industry 1 `" "Basic" "Materials" "'  ///
2  `" "Communi-" "cations" "' 3 `" "Consumer" "Cyclical" "' 4 `" "Consumer" "Non-Cyclical" "' ///
5  "Energy"  6 "Financial"  7 "Industrial"  8 "Technology" 9  "Utilities" , modify

label val type_industry type_industry

* get a gmean() function (don't remove the *) 
* ssc inst egenmore

* sort on geometric mean (don't remove the *) 
* ssc inst myaxis
myaxis xaxis=type_industry, sort(gmean performance)

* ssc inst stripplot
stripplot performance, over(xaxis) centre vertical cumul cumprob refline scheme(s1color) xla(, noticks labsize(small)) xsize(7) xtitle("") ysc(log) yla(5 2 1 0.5 0.2 0.1 0.05 0.02, ang(h))  reflevel(gmean) mc(blue)

And I get the error no observations (r 2000).

[Nick Cox]

Already answered. That code creates a fake dataset. Therefore you need set obs. Otherwise put, how does Stata know how many observations to use unless you spell it out?

[Lucas Bordure]

But if you create a fake database. I imagine that it does not correspond to the observations of my base file named "File4". By deduction, I imagine that the result is not related by my initial observations but the result is related by the generation of random observations created by this following 2 commands.

set obs 900
set seed 2803

So the code beginning from **Fixed effects: Heterogeneity across entities (stripplot) is wrong.

use"File4.dta", clear


*means and standard deviations of intangible assets by industry
preserve
collapse (mean) mean_intangible_assets_industry=intangible_assets (sd) sd_intangible_assets_industry=intangible_assets, by (industry)
asdoc list, replace
restore
*means and standard deviations of intangible assets by year
preserve
collapse (mean) mean_intangible_assets_year=intangible_assets (sd) sd_intangible_assets_year=intangible_assets, by (year)
asdoc list, replace
restore
*descriptive statistics for control variables (average, mean, sd)
asdoc tabstat enterprise_value_added market_capitalization leverage stock_growth dividend_payout_ratio stock_volatility, stat(mean p50 sd), replace
*correlation for control variables 
corr enterprise_value_added market_capitalization leverage stock_growth dividend_payout_ratio stock_volatility
*setting panel data
xtset i year
*exploring panel data
xtline ln_firm_performance
xtline ln_firm_performance, overlay



*OLS regression
regress firm_performance intangible_assets
scatter firm_performance intangible_assets, ms(none) ylabel(minmax) xlabel(minmax) mlabel(industry_l) mlabpos(0) || lfit firm_performance intangible_assets, clstyle(p2)



*Fixed effects: Heterogeneity across entities (stripplot)
clear
set obs 900
set seed 2803

gen performance = exp(rnormal(-2, 1))
gen type_industry = ceil(_n/ 100)

label def type_industry 1 `" "Basic" "Materials" "'  ///
2  `" "Communi-" "cations" "' 3 `" "Consumer" "Cyclical" "' 4 `" "Consumer" "Non-Cyclical" "' ///
5  "Energy"  6 "Financial"  7 "Industrial"  8 "Technology" 9  "Utilities" , modify

label val type_industry type_industry

* get a gmean() function (don't remove the *) 
* ssc inst egenmore

* sort on geometric mean (don't remove the *) 
* ssc inst myaxis
myaxis xaxis=type_industry, sort(gmean performance)

* ssc inst stripplot
stripplot performance, over(xaxis) centre vertical cumul cumprob refline scheme(s1color) xla(, noticks labsize(small)) xsize(7) xtitle("") ysc(log) yla(5 2 1 0.5 0.2 0.1 0.05 0.02, ang(h))  reflevel(gmean) mc(blue)

I would be grateful if a code integrating my initial observations could be runned.

[Nick Cox]

As you have downloaded stripplot, the suggestion is to use your own dataset and your own variable names and to work out what options you need.