*DO-FILE FOR "NATO SOURCES DATASET" *This do-file corresponds with tables and figures featured in the article "Who Matters for Memory: Sources of Institutional Memory in International Organization Crisis Management" in the Review of International Organizations *Tables and Figures indicated below are presented in the order in which they appear in the Main Manuscript. *The codebook for the dataset and a list of all variables in the dataset are located at the bottom of this file. **********MAIN MANUSCRIPT********** *Table 2: Average Treatment Effects (ATEs) *(a) IS Source Cue as Treatment reg sccsr iscontrol reg colleague iscontrol reg sprvisr iscontrol *(b) US Source Cue as Treatment reg sccsr uscontrol reg colleague uscontrol reg sprvisr uscontrol *Figure 2: Likelihood of Recording for Self or Successors by Institution twoway scatter sccsr institution if treated==0, jitter(7) twoway scatter sccsr institution if treated==3, jitter(7) *Table 4-6 *Quotes from NATO elites *APPENDIX *Table 1: Subject Pool *Indicates how many subjects sampled at each of six NATO institutions *Figure 1: Recording for Self or Successors versus Discussing with Colleague *Recording for Self or Successors reg sccsr uscontrol adjust, by(uscontrol) ci *Copied these coefficients and confidence intervals into Excel and made the bar graphs reg sccsr iscontrol adjust, by(iscontrol) ci *Copied these coefficients and confidence intervals into Excel and made the bar graphs reg sccsr mediacontrol adjust, by(mediacontrol) ci reg colleague uscontrol adjust, by(uscontrol) ci *Copied coefficients and confidence intervals into Excel and made bar graphs reg colleague iscontrol adjust, by(iscontrol) ci *Copied coefficients and confidence intervals into Excel and made bar graphs reg colleague mediacontrol adjust, by(mediacontrol) ci *Table 3: Effect of US Source Cue on Recording the Error for Self or Successors **Model 1 (just treatment) ologit sccsr uscontrol **Model 2 (treatment with all controls) ologit sccsr uscontrol institution age gender nationality hypsc workfor civmil fe currentyrs totalyrs **********ONLINE RESOURCE********** *TABLE OF CONTENTS *I. THEORETICAL ARGUMENT *Additional Discussion of Terminology *II. HYPOTHESES *Discussion of Hypothesis 4: The International Media *III. METHODOLOGY *Study Design Assumptions *Subject Recruitment *Dates and Locations of Experiment on Subjects *Demographic and Institutional Characteristics sum age gender nationality hypsc institution workfor civmil fe currentyrs totalyrs *Block Randomization *Pre-test Questions *Survey Instrument as Treatment *Outcome Questions *IV. RESULTS *1. OVERVIEW OF TREATMENT EFFECTS *Likelihood of Discussing with Supervisor graph bar (mean) sprvisr, over(treated) blabel(bar) *Successor: Differences in Means across Treatments versus Control mean sccsr, over(treated) oneway sccsr treated, tabulate *Supervisor: Differences in Means across Treatments versus Control mean sprvisr, over(treated) oneway sprvisr treated, tabulate *Colleague: Differences in Means across Treatments versus Control mean colleague, over(treated) oneway colleague treated, tabulate *Joint Statistical Test – Seemingly Unrelated Regression with Supervisor sureg (sccsr uscontrol)(colleague uscontrol)(sprvisr uscontrol) *Cell Sizes across Source Cue Conditions on Recording for Self or Successors tab sccsr treated *Cell Sizes across Source Cue Conditions on Discussing with Supervisor tab sprvisr treated *Cell Sizes across Source Cue Conditions on Discussing with Colleague tab colleague treated *Balancing Statistics mean institution if iscontrol==1 mean institution if uscontrol==1 mean institution if treated==0 mean age if iscontrol==1 mean age if uscontrol==1 mean age if treated==0 mean gender if iscontrol==1 mean gender if uscontrol==1 mean gender if treated==0 mean nationality if iscontrol==1 mean nationality if uscontrol==1 mean nationality if treated==0 mean hypsc if iscontrol==1 mean hypsc if uscontrol==1 mean hypsc if treated==0 mean workfor if iscontrol==1 mean workfor if uscontrol==1 mean workfor if treated==0 mean civmil if iscontrol==1 mean civmil if uscontrol==1 mean civmil if treated==0 mean fe if iscontrol==1 mean fe if uscontrol==1 mean fe if treated==0 mean currentyrs if iscontrol==1 mean currentyrs if uscontrol==1 mean currentyrs if treated==0 mean totalyrs if iscontrol==1 mean totalyrs if uscontrol==1 mean totalyrs if treated==0 *2. INTERNATIONAL STAFF SOURCE CUE *Effect of International Staff Cue on Recording for Self or Successors **Logistic Model 3 (just treatment) ologit sccsr iscontrol **Logistic Model 4 (treatment with all controls) ologit sccsr iscontrol institution isinstit age gender nationality hypsc workfor civmil fe currentyrs totalyrs **OLS Model (just treatment) regress sccsr iscontrol *Effect of International Staff Cue on Discussing with Colleague *Logistic Model (just treatment) ologit colleague iscontrol *Logistic Model (treatment with all controls) ologit colleague iscontrol institution isinstit age gender nationality hypsc workfor civmil fe currentyrs totalyrs **OLS Model (just treatment) regress colleague iscontrol *Effect of International Staff Cue on Discussing with Supervisor *OLS Model (just treatment) regress sprvisr iscontrol *3. US SOURCE CUE *US Source Cue on Recording for Self or Successors ** OLS Model (just treatment) regress sccsr uscontrol ** OLS Model 10 (treatment with all controls) regress sccsr uscontrol institution age gender nationality hypsc workfor civmil fe currentyrs totalyrs *US Source Cue on Discussing with a Colleague **Logistic Model 5 (just treatment) ologit colleague uscontrol **Logistic Model 6 (treatment with controls) ologit colleague uscontrol institution age gender nationality hypsc workfor civmil fe currentyrs totalyrs **OLS Model (just treatment) regress colleague uscontrol *US Source Cue on Discussing with a Supervisor regress sprvisr uscontrol *Effect of US Source Cue on American Elites **1. I generated a new variable "american" indicating 1 if American, 0 if non-American *gen american = 0 *replace american = 1 if nationality==28 **2. I created an interaction term for American receiving the US treatment *gen american_uscontrol = american*uscontrol **3. I then regressed successor variable on US treatment and americans who received the US treatment reg sccsr uscontrol american american_uscontrol **4. I also regressed colleague variable on US treatment and americans who received the US treatment reg colleague uscontrol american american_uscontrol *4. INTERNATIONAL MEDIA SOURCE CUE *Effect of International Media Source Cue on Recording for Self or Successors *Logistic Model (just treatment) ologit sccsr mediacontrol *Logistic Model (treatment with controls) ologit sccsr mediacontrol institution age gender nationality hypsc workfor civmil fe currentyrs totalyrs **OLS Model (just treatment) regress sccsr mediacontrol *Effect of International Media Source Cue on Discussing with a Colleague *Logistic Model (just treatment) ologit colleague mediacontrol *Logistic Model (treatment with controls) ologit colleague mediacontrol institution age gender nationality hypsc workfor civmil fe currentyrs totalyrs **OLS Model (just treatment) regress colleague mediacontrol *Effect of International Media Source Cue on Discussing with a Supervisor **OLS Model (just treatment) regress sprvisr mediacontrol *5. CORRELATIONS **Correlations between US Source Cue and Independent Variables / Control Variables corr uscontrol institution corr uscontrol age corr uscontrol gender corr uscontrol nationality corr uscontrol hypsc corr uscontrol workfor corr uscontrol civmil corr uscontrol fe corr uscontrol currentyrs corr uscontrol totalyrs **Correlations between IS Source Cue and Independent Variables / Control Variables corr iscontrol sccsr corr iscontrol colleague corr iscontrol sprvisr corr iscontrol institution corr iscontrol age corr iscontrol gender corr iscontrol nationality corr iscontrol hypsc corr iscontrol workfor corr iscontrol civmil corr iscontrol fe corr iscontrol currentyrs corr iscontrol totalyrs **Correlations between Media Source Cue and Independent Variables / Control Variables corr mediacontrol sccsr corr mediacontrol colleague corr mediacontrol sprvisr corr mediacontrol institution corr mediacontrol age corr mediacontrol gender corr mediacontrol nationality corr mediacontrol hypsc corr mediacontrol workfor corr mediacontrol civmil corr mediacontrol fe corr mediacontrol currentyrs corr mediacontrol totalyrs *Correlations between Sources and Control Variables *Data from above all entered into Excel as a bar graph *6. IRB Protocol as Pre-Registration **********DATA************* *The following variables are included in the dataset: *iscontrol *mediacontrol *uscontrol *colleague *institution *sccsr *sprvisr *treated *isinstit *The following demographic data has been removed from the dataset in order to protect the anonymity of human subjects, in accordance with IRB protocol, as anonymity was assured in exchange for subjects' participation. *age *currentyrs *fe *gender *hypsc *civmil *nationality *totalyrs *workfor *american *american_uscontrol **********CODEBOOK********** *VARIABLES IN THE "NATO SOURCES DATASET" *Variables correspond with outcome questions, treatments and pre-test questions as indicated in the Methodology section of the Online Resource *DEPENDENT VARIABLES *sccsr = likelihood of recording for self or successors ***Question: How likely would you be to record this for yourself or successors? ** 0 if Not at all likely ** .25 if Unlikely ** .5 if Somewhat likely ** .75 if Very Likely ** 1 if Definitely likely *sprvisr = likelihood of discussing with supervisor ***Question: How likely would you be to discuss this with your supervisor? ** 0 if Not at all likely ** .25 if Unlikely ** .5 if Somewhat likely ** .75 if Very Likely ** 1 if Definitely likely *colleague = likelihood of discussing with close colleague ***Question: How likely would you be to discuss this with a close colleague? ** 0 if Not at all likely ** .25 if Unlikely ** .5 if Somewhat likely ** .75 if Very Likely ** 1 if Definitely likely *INDEPENDENT VARIABLES *treated = indicator of whether elite received one of three treatments or a placebo condition ** 0 if Placebo ** 1 if International Staff ** 2 if International Media ** 3 if US *iscontrol = International Staff treatment as source cue ***Elite received a hypothetical scenario in which the International Staff was used as treatment ** 0 if elite did not receive IS treatment ** 1 if elite did receive IS treatment *mediacontrol = international media treatment as a source cue ***Elite received a hypothetical scenario in which the International Staff was used as treatment ** 0 if elite did not receive media treatment ** 1 if elite did receive media treatment *uscontrol = US treatment as a source cue ***Elite received a hypothetical scenario in which the International Staff was used as treatment ** 0 if elite did not receive US treatment ** 1 if elite did receive US treatment *CONTROL VARIABLES *institution = Elite's current institution at NATO where he/she is physically working - irrespective of his/her employer **NOTE: For example, an elite may be working as a military staff member in IMS at NATO Headquarters but be employed by his/her member state. * 1 if Allied Command Operations (ACO) * 2 if Allied Command Transformation (ACT) * 3 if International Military Staff (IMS) * 4 if International Staff (IS) * 5 if Military Committee (MC) * 6 if North Atlantic Council (NAC) *isinstit = interaction term of NATO elites working for the IS and who also received the IS source cue **Original code for creating this variable was the following: generate isinstit = iscontrol*institution *civmil = Elite's job status as civilian or military ** 0 if civilian ** 1 if military *Question: Are you considered civilian staff or military staff? *workfor = Elite's employer ** 0 if a state (e.g. France) ** 1 if NATO *Question: Are you employed by NATO, a nation or do you work as a VNC? **NOTE: Ultimately, I did not recruit any voluntary national contribution (VNCs) for this study because they did not fit the requirements for subject recruitment as indicated in the paper. Therefore, there is no code listed here for VNC. *currentyrs = Numerical value indicating the elite's current years working in the current position in NATO *Question: How many years have you worked in this position in NATO? *totalyrs = Numerical value for the elite's total years in his/her lifetime working in NATO *Question: How many years have you worked in your lifetime in NATO? *fe = Elite has had or not had field experience of at least three months in a NATO crisis management operation ** 0 if no ** 1 if yes *Question: Have you had field experience of at least three months in a NATO crisis management operation? *hypsc = Elite has had or not responded to a hypothetical scenario as part of a study before ** 0 if no ** 1 if yes *Question: Prior to this, have you responded to a hypothetical scenario as part of a study before? *nationality = Elite's nationality * 1 Albania * 2 Belgium * 3 Bulgaria * 4 Canada * 5 Croatia * 6 Czech Republic * 7 Denmark * 8 Estonia * 9 France * 10 Germany * 11 Greece * 12 Hungary * 13 Iceland * 14 Italy * 15 Latvia * 16 Lithuania * 17 Luxembourg * 18 Netherlands * 19 Norway * 20 Poland * 21 Portugal * 22 Romania * 23 Slovakia * 24 Slovenia * 25 Spain * 26 Turkey * 27 United Kingdom * 28 United States *Question: What is your nationality? *gender = Elite's gender identity ** 0 if Male ** 1 if Female *Question: What is your gender? *age = Numerical variable to indicate elite's age *Question: What is your age? *american = Elite is American or not ** 0 if not American ** 1 if American