Table 3.

Multiple linear regression models explaining positive associations towards Questionable Research Practices (QRP, columns 1–2) and Falsification, Fabrication and Plagiarism (FFP, column 3–4)

	(1)	(2)	(3)	(4)
	QRP	QRP	FFP	FFP
Humanities & Social Sciences	Ref.	Ref.	Ref.	Ref.
Life Sciences	−0.0107	−0.0303	−0.00127	−0.0113
	(−0.60)	(−1.33)	(−0.09)	(−0.61)
Natural Sciences	−0.00798	−0.0327	−0.00320	−0.00680
	(−0.44)	(−1.39)	(−0.22)	(−0.35)
Engineering Sciences	−0.00261	−0.00638	−0.0132	0.0284
	(−0.13)	(−0.21)	(−0.82)	(1.18)
Female	0.0463***	0.0568**	−0.00471	0.00572
	(3.34)	(3.05)	(−0.42)	(0.38)
Predoc	Ref.	N.a.	ref.	N.a.
Postdoc	−0.0991***	Ref.	−0.0274*	Ref.
	(−6.65)		(−2.28)
Professor	−0.173***	−0.0763***	−0.0515***	−0.0254
	(−9.69)	(−4.08)	(−3.50)	(−1.66)
Europe		ref.		Ref.
Americas		−0.0405		0.00453
		(−0.84)		(0.12)
Asia		0.235**		0.103
		(2.94)		(1.58)
Constant	−0.190***	−0.286***	−0.717***	−0.752***
	(−12.47)	(−15.95)	(−58.45)	(−52.93)
N	5835	3300	5912	3333
$R^2$	0.0221	0.0129	0.00226	0.00236

t statistics in parentheses

* $p<0.05$, ** $p<0.01$, *** $p<0.001$

QRP (1) and FFP (3) include all scientists.QRP (2) and FFP (4) add PhD country information

The sample size is therefor restricted to postdocs and professors