version 12.1 * Stata program for citation impact/margins analysis, updated Feb 2013 * esttab from SSC needs to be installed. * Also, fre from SSC * Some tables will need additional formatting by hand use "https://www3.nd.edu/~rwilliam/margins/bornmann.dta", clear * Define vars, select sample encode oecdma, gen(noecdma) encode doctype, gen(ndoctype) fre n* drop if noecdma == 1 | noecdma == 3 | noecdma == 6 * For py2, 1980 = year 31, 2010 = 1 gen py2 = 2011 - py * Create dummies for labeling purposes in the tables tab1 univ, gen(University) tab1 noecdma, gen(subject) ren subject1 Engineering ren subject2 MedHealth ren subject3 NatSciences tab1 ndoctype, gen(doctype) rename doctype1 Article rename doctype2 Note ren doctype3 Proceedings ren doctype4 Review clonevar PPTop10 = pr2 label variable jif "Journal Impact Factor" label variable py2 "# Years since publication" label variable PPTop10 "Highly Cited Publication" label define univ 1 "Univ 1", modify label define univ 2 "Univ 2", modify label define univ 3 "Univ 3", modify label define univ 4 "Univ 4", modify * Table 1 - Descriptives sum PPTop10 University1-University4 Engineering-NatSciences Article-Review jif py2 auth prange *** Table 2 * Model 1 - univ only logit PPTop10 i.univ est store m1 * Model 2 - All vars, no transformations logit PPTop10 i.univ i.noecdma i.ndoctype jif py2 auth prange est store m2 * Model 3 - Final Model with Squared Terms logit PPTop10 i.univ i.noecdma i.ndoctype jif py2 auth prange c.jif#c.jif c.prange#c.prange est store m3 * Create Table 2. It will still need hand-editing esttab m1 m2 m3 using table2, z pr2 aic bic scalars(chi2 df_m) mti("Baseline" "All vars" "Squared trms") rtf * Table 3 - AAPs and AMES, Categorical Variables * Table will still need hand-editing est restore m3 margins univ noecdma ndoctype, post est store m11 est restore m3 margins, dydx(univ noecdma ndoctype) post est store m12 esttab m11 m12 using table3, z mti(AAPS AMES) rtf title("Table 3. Average Adjusted Predictions and Average Marginal Effects") replace *** Figures follow * Graph 1 AAPs for JIF est restore m3 margins , at(jif = (0 (1) 35)) marginsplot, scheme(s1mono) yti(`"Pr(Highly Cited Publication)"') title("") recastci(rarea) ci1opts(fintensity(30)) ylabel(#10) xlabel(#10) * Graph 2 MERs for JIF est restore m3 margins , dydx(jif) at(jif = (0 (1) 35)) marginsplot, scheme(s1mono) yti(`"Effects on Pr(Highly Cited Publication)"') title("") recastci(rarea) ci1opts(fintensity(30)) ylabel(#10) xlabel(#10) * Graph 3 APRs for # Pages est restore m3 margins , at(prange = (1 (1) 120)) marginsplot, scheme(s1mono) yti(`"Pr(Highly Cited Publication)"') title("") recastci(rarea) ci1opts(fintensity(30)) ylabel(#10) xlabel(#20) * Graph 4 MERs for # Pages est restore m3 margins , dydx(prange) at(prange = (1 (1) 120)) marginsplot, scheme(s1mono) yti(`"Effects on Pr(Highly Cited Publication)"') title("") recastci(rarea) ci1opts(fintensity(30)) ylabel(#10) xlabel(#20) ****************************************************************************** * Graph 5 APRs for univ, JIF est restore m3 margins univ, at(jif = (0 (1) 13)) marginsplot, scheme(s1mono) yti(`"Pr(Highly Cited Publication)"') title("") ylabel(#10) xlabel(#20) * Graph 6 MERs for Univ, JIF est restore m3 margins r.univ, at(jif = (0 (1) 13)) marginsplot, scheme(s1mono) yti(`"Effects on Pr(Highly Cited Publication)"') title("") ylabel(#10) xlabel(#20) * Graph 7 APRs for univ, # Pages est restore m3 margins univ, at(prange = (1 (1) 25)) marginsplot, scheme(s1mono) yti(`"Pr(Highly Cited Publication)"') title("") ylabel(#10) xlabel(#10) * Graph 8 MERs for univ, # Pages est restore m3 margins r.univ, at(prange = (1 (1) 25)) marginsplot, scheme(s1mono) yti(`"Effects on Pr(Highly Cited Publication)"') title("") ylabel(#10) xlabel(#20)