Table 4.
Regression results with number of authors
| Number of authors | Title length | Citations | Year | Colon % | Question % |
|---|---|---|---|---|---|
| 1 | −0.0582** (10.13) | −45.6** (11.42) | −0.161** (10.71) | 0.122** (6.59) | −0.0293 (1.12) |
| 2–4 | 0.0154** (2.95) | −42.7** (10.81) | −0.135** (9.49) | −0.0095 (0.54) | 0.0245 (0.96) |
| 5–9 | 0.0889** (16.12) | −36.7** (9.22) | −0.101** (6.66) | −0.0753** (3.89) | −0.243** (8.02) |
| 10–24 | 0.111** (18.01) | −19.7** (4.76) | −0.00727 (0.42) | −0.0914** (4.01) | −0.447** (10.54) |
| 25–49 | 0.0258* (2.13) | 20.30** (3.31) | 0.12** (3.39) | 0.0357 (0.74) | −0.65** (4.72) |
| Observations | 139,705 | 62,878 | 139,705 | 139,705 | 139,705 |
The independent variables were dummy variables proxying the number of authors. The regressions are: (i) an OLS regression on the log of the title length, (ii) a Tobit regression on the number of citations with a lower bound of zero, (iii) an ordered probit regression on year of publication, (iv) a probit regression on whether a colon is present in the title and (v) a probit regression on whether a ‘?’ is present in the title. All regressions contain dummy variables for different UoAs and a constant term. Standard errors corrected for heteroscedasticty
*, ** denotes significance at the 1/5 % levels