Table 6:
Breakdown by Star Scientist Characteristics
| Publications | Citations | Funding | Importance to the Field |
|||||
|---|---|---|---|---|---|---|---|---|
| Below Median |
Above Median |
Below Median |
Above Median |
Below Median |
Above Median |
Below Median |
Above Median |
|
| After Death | 0.059 (0.037) |
0.116* (0.050) |
0.036 (0.042) |
0.125** (0.040) |
0.014 (0.040) |
0.162** (0.052) |
0.063* (0.031) |
0.123** (0.045) |
| Nb. of Investigators | 2,901 | 4,836 | 2,792 | 4,619 | 3,048 | 4,287 | 5,019 | 4,493 |
| Nb. of Fields | 17,210 | 17,008 | 17,328 | 16,890 | 15,731 | 15,487 | 16,985 | 17,233 |
| Nb. of Field-Year Obs. | 632,089 | 627,087 | 636,750 | 622,426 | 578,277 | 570,665 | 625,140 | 634,036 |
| Log Likelihood | −1,377,741 | −1,387,648 | −1,367,337 | −1,396,654 | −1,268,567 | −1,252,952 | −1,462,541 | −1,257,972 |
Note: Estimates stem from conditional (subfield) fixed effects Poisson specifications. The dependent variable is the total number of publications by noncollaborators in a subfield in a particular year. Each pair of columns splits the sample across the median of a particular covariate for the sample of fields (treated and control) in the baseline year. The table examines differences in the extent to which the eminence of the star at death (respectively counterfactual year of death for controls) influences the rate at which non-collaborators enter the field after the star passes away. Eminence is measured through the star’s cumulative number of publications, the star’s cumulative number of citations garnered up to the year of death, and the star’s cumulative amount of NIH funding. We also have a “local” measure of eminence: the star’s importance to the field, which is defined as the proportion of articles in the subfield up to the year of death for which the star is an author. All models incorporate a full suite of year effects and subfield age effects, as well as a term common to both treated and control subfields that switches from zero to one after the death of the star. Exponentiating the coefficients and differencing from one yield numbers interpretable as elasticities. For example, the estimate in the second column implies that treated subfields see an increase in the number of contributions by non-collaborators after the superstar passes away—a statistically significant 100×(exp[0.116]-1)=12.30%.
Robust standard errors in parentheses, clustered at the level of the star scientist.
p < 0.10,
p < 0.05,
p < 0.01.