. 2018 Sep 26;13(9):e0195773. doi: 10.1371/journal.pone.0195773

Table 5. Models of self-citation behavior of first and last authors based on 41.6 million references from 1.6 million articles with 2 or more authors published during 2002-2005.

Predictor^†	First author effects, simple models^‡	First author effects, complete model	Last author effects, complete model
Intercept	—	`-1.957 (0.031)`	`-2.790 (0.027)`
gender (vs. Unknown)
Female	`0.021 (0.003)`	`-0.029 (0.003)`	`0.026 (0.003)`
Male	`0.451 (0.002)`	`-0.039 (0.003)`	`-0.002 (0.003) X`
age
log₁₀(pub count + 1)	`1.440 (0.007)`	`1.610 (0.007)`	`2.172 (0.008)`
log₁₀(pub count + 1)²	`-0.119 (0.002)`	`-0.091 (0.003)`	`-0.340 (0.002)`
pub count = 0	`-1.375 (0.009)`	`-1.421 (0.009)`	`-1.232 (0.016)`
pub count = 1	`-0.210 (0.005)`	`-0.236 (0.005)`	`-0.187 (0.011)`
ethnicity (vs. ENGLISH)
ARAB	`-0.470 (0.006)`	`-0.055 (0.006)`	`-0.085 (0.006)`
CHINESE	`-0.444 (0.003)`	`0.046 (0.004)`	`0.034 (0.004)`
DUTCH	`-0.068 (0.004)`	`0.030 (0.007)`	`-0.004 (0.005) X`
FRENCH	`-0.205 (0.003)`	`-0.074 (0.004)`	`-0.038 (0.003)`
GERMAN	`-0.056 (0.003)`	`-0.014 (0.004)`	`0.062 (0.003)`
HISPANIC	`-0.261 (0.003)`	`0.053 (0.004)`	`0.082 (0.003)`
INDIAN	`-0.418 (0.004)`	`-0.064 (0.005)`	`0.020 (0.005)`
ITALIAN	`0.084 (0.003)`	`-0.060 (0.006)`	`0.026 (0.004)`
JAPANESE	`0.018 (0.003)`	`-0.102 (0.006)`	`-0.048 (0.006)`
KOREAN	`-0.656 (0.007)`	`-0.187 (0.007)`	`-0.076 (0.006)`
NORDIC	`0.037 (0.004)`	`0.136 (0.005)`	`0.070 (0.004)`
OTHER	`-0.362 (0.004)`	`-0.181 (0.004)`	`-0.057 (0.003)`
SLAV	`-0.103 (0.004)`	`0.044 (0.005)`	`0.032 (0.004)`
UNKNOWN	`-0.318 (0.017)`	`0.013 (0.018) X`	`-0.046 (0.014)`
country (vs. USA)
Australia	`0.004 (0.005)`	`0.088 (0.005)`	`-0.099 (0.004)`
Canada	`-0.121 (0.004)`	`-0.004 (0.004) X`	`-0.069 (0.003)`
China	`-0.751 (0.009)`	`-0.412 (0.010)`	`-0.432 (0.007)`
France	`-0.133 (0.004)`	`-0.044 (0.005)`	`-0.116 (0.004)`
Germany	`0.029 (0.003)`	`0.024 (0.004)`	`-0.169 (0.003)`
India	`-0.359 (0.009)`	`-0.044 (0.011)`	`-0.206 (0.008)`
Italy	`0.268 (0.003)`	`-0.101 (0.007)`	`-0.298 (0.005)`
Japan	`0.154 (0.003)`	`0.267 (0.006)`	`-0.093 (0.006)`
Netherlands	`-0.022 (0.005)`	`0.068 (0.007)`	`-0.179 (0.005)`
Other	`-0.124 (0.002)`	`-0.064 (0.003)`	`-0.213 (0.002)`
Spain	`-0.002 (0.005)`	`-0.122 (0.006)`	`-0.184 (0.005)`
Sweden	`0.113 (0.005)`	`0.107 (0.007)`	`-0.083 (0.005)`
UK	`-0.050 (0.003)`	`-0.005 (0.003) X`	`-0.090 (0.002)`
Unknown	`0.091 (0.006)`	`-0.045 (0.006)`	`-0.159 (0.005)`
author count
author count > 20	`0.018 (0.014)`	`0.000 (0.015) X`	`-0.050 (0.013)`
log₁₀(author count)	`-1.347 (0.016)`	`0.082 (0.018)`	`-0.444 (0.014)`
log₁₀(author count)²	`0.941 (0.012)`	`-0.256 (0.013)`	`0.067 (0.010)`
language
ref. English	`-0.456 (0.009)`	`-0.185 (0.010)`	`-0.031 (0.009)`
English	`0.488 (0.007)`	`0.649 (0.008)`	`0.701 (0.007)`
references
log₁₀(count)	`-1.977 (0.028)`	`-1.647 (0.034)`	`-2.221 (0.031)`
log₁₀(count)²	`0.742 (0.020)`	`0.820 (0.023)`	`1.463 (0.022)`
log₁₀(count)³	`-0.125 (0.004)`	`-0.229 (0.005)`	`-0.387 (0.005)`
count = 1	`-0.390 (0.019)`	`-0.300 (0.022)`	`-0.405 (0.021)`
MeSH
log10₁₀(ref. count + 1)	`0.832 (0.004)`	`-0.168 (0.006)`	`-0.092 (0.004)`
ref. count = 0	`1.978 (0.009)`	`-0.093 (0.015)`	`0.005 (0.011) X`
log10₁₀(count + 1)	`-0.693 (0.004)`	`-0.170 (0.006)`	`-0.251 (0.004)`
count = 0	`-0.812 (0.009)`	`-0.667 (0.013)`	`-0.892 (0.010)`
novelty
log10₁₀(ref. score + 1)	`-0.329 (0.003)`	`0.095 (0.008)`	`0.048 (0.005)`
log10₁₀(ref. score + 1)²	`0.079 (0.001)`	`-0.028 (0.001)`	`-0.022 (0.001)`
log10₁₀(score + 1)	`-0.062 (0.001)`	`-0.106 (0.001)`	`-0.139 (0.001)`
pub type
ref = Case Report	`-0.423 (0.006)`	`-0.637 (0.006)`	`-0.731 (0.005)`
ref = Journal	`0.103 (0.011)`	`0.328 (0.011)`	`0.324 (0.009)`
ref = Letter	`-0.044 (0.011)`	`-0.284 (0.011)`	`-0.292 (0.009)`
ref = Review	`-0.723 (0.003)`	`-0.783 (0.003)`	`-0.641 (0.002)`
Case Report	`-0.892 (0.006)`	`-0.940 (0.007)`	`-0.948 (0.005)`
Journal	`-0.219 (0.011)`	`0.266 (0.011)`	`0.229 (0.010)`
Letter	`0.468 (0.011)`	`-0.430 (0.011)`	`-0.343 (0.010)`
Review	`0.080 (0.002)`	`-0.079 (0.003)`	`-0.249 (0.002)`
venue
same journal	`0.492 (0.003)`	`0.457 (0.003)`	`0.534 (0.002)`
log₁₀(journal similarity + 1)	`0.214 (0.009)`	`0.063 (0.010)`	`0.126 (0.008)`
log₁₀(journal similarity + 1)²	`0.050 (0.003)`	`0.006 (0.003) *`	`-0.050 (0.003)`
journal similarity = 0	`-0.178 (0.008)`	`-0.281 (0.008)`	`-0.296 (0.006)`
time lag
time lag < 0	`-1.884 (0.015)`	`-0.556 (0.016)`	`-0.383 (0.015)`
log₁₀(time lag + 1)	`-2.099 (0.013)`	`0.539 (0.015)`	`1.232 (0.011)`
log₁₀(time lag + 1)²	`-0.275 (0.009)`	`-1.281 (0.010)`	`-1.402 (0.007)`
time lag = 0	`0.016 (0.005)`	`0.538 (0.006)`	`0.735 (0.005)`
ref. citation
log₁₀(count + 1)	`-0.769 (0.006)`	`-0.443 (0.006)`	`-0.079 (0.004)`
log₁₀(count + 1)²	`-0.199 (0.002)`	`-0.170 (0.003)`	`-0.169 (0.002)`
count = 0	`0.161 (0.004)`	`0.127 (0.004)`	`0.240 (0.003)`

^†Format: logit (SE) signif., where X = p ≥ 0.05, * = p < 0.05, . = p < 0.01, and p < 0.001 otherwise.

^‡Each category represents a simple model (with only one predictor); intercepts not shown.