Skip to main content
Cognitive Research: Principles and Implications

Table 1 Summary of Empirical Studies on Collective Intelligence

From: g versus c: comparing individual and collective intelligence across two meta-analyses

Study name

N

K

Effect 1: % Var

Effect 2: (%) Pos. Manifold

Effect 3: c → criterion (r)

Effect 4: Av.IQ → c (r)

Effect 5: Av.IQ → criterion (r)

RA (Y:N)

aCredé and Howardson (2017a, b)

487

6

     

5:1

aBates and Gupta (2017): Studies 2 and 3 (combined analysis)

80

2

     

Yes

Woolley et al. (2010, ): Study 1

40

 

43.4

100

.52

.19, p = ns

.18, p = ns

Yes

Woolley et al. (2010, : Study 2

152

 

44.1

93

.28

.15, p = .04

.18, p = ns

Yes

Engel, Woolley, Jing, Chabris, and Malone (, 2014a, b): Face-to-face (speaking) condition

32

 

49.3

100

   

Yes

Engel, Woolley, Jing, Chabris, and Malone (, 2014a, b): Online (text-chat) condition

36

 

41.4

100

   

Yes

Engel et al. (2015a, b): Study 2

116

 

40

100

.25

  

No

Engel, Woolley, et al. (2015a, b): Study 3

25

 

40

100

   

Yes

Woolley & Aggarwal (under review); (Also reported in Woolley and Aggarwa2017)

59

   

C1: .29 and C2: .29

-.05, p = 53

C1: -.02, p > .05; C2: -.21, p > .05

No

Meslec et al. (2016)

30

      

Yes

Glikson, Harush, et al. (under review)

115

   

.11

  

No

Chikersal, et al. (2017)

58

      

No

Kim et al. (2017a, 2017b)

248

 

38.38

 

-.15

  

Yes

Aggarwal et al. (2019)d

98

 

44

 

.58e

  

Yes

Barlow and Dennis (2016, )

86

 

42

50

  

.07 (p > .05)

Yes

Barlow (2015, unpublished doctoral thesis): Control Group (CG)

64

  

33

   

Yes

Barlow (2015, unpublished doctoral thesis): Experimental Group (EG)

65

 

46

100

 

.339, p = 026 c

 

Yes

Bates and Gupta (2017): Study 1

26

 

39.8

100

   

Yes

Bates and Gupta (2017): Study 2

40

 

50

100

   

Yes

Bates and Gupta (2017): Study 3

40

  

100

   

Yes

Rowe (2019, unpublished doctoral thesis)

29

 

41

100

.104, p = .59

.294, p = .12

.202, p = .29

No

Mean or Ratio:

71.5

 

43.03

90

.253

.185

.067

13:5

Study name

Subtests (n)

Independent: Woolley

Lab: Field

Online or Face-to-face b

Group size

Country (USA:Other)

aCredé and Howardson (2017a, b)

 ≥ 4 < 17

Yes

Both

Both

 

Multiple

aBates and Gupta (2017): Studies 2 and 3 (combined analysis)

5

Yes

Lab

Both

3

Multiple

Woolley et al. (2010, ): Study 1

5

No

Lab

F2F

3

USA

Woolley et al. (2010, ): Study 2

10

No

Lab

F2F

2 to 5

USA

Engel et al. (2014a, b): Face-to-face (speaking) condition

8

No

Lab

F2F

4

USA

Engel et al. (2014a, b): Online (text-chat) condition

8

No

Lab

online

4

USA

Engel et al. (2015a, b): Study 2

7

No

Field

online

2 to 5

Germany

Engel et al. (2015a, b): Study 3

6

No

Lab

online

4

Japan

Woolley and Aggarwal (under review); (Also reported in Woolley & Aggarwal, 2017)

8

No

Field

online

4 to 5

USA

Meslec et al. (2016)

8

No

Field

online

3 to 6

Netherlands

Glikson, Harush, et al. (under review)

8

No

Field

online

 

USA

Chikersal, et al. (2017)

6

No

Lab

online

2

USA

Kim et al. (2017a, b)

11

No

Field

online

5

Multiple

Aggarwal et al. (2019)d

5 or 10

No

Lab

Both

2 to 5

USA

Barlow and Dennis (2016, )

3

Yes

Lab

online

3 to 5

USA

Barlow (2015, unpublished doctoral thesis): Control Group (CG)

3

Yes

Lab

online

3 to 5

USA

Barlow (2015, unpublished doctoral thesis): Experimental Group (EG)

3

Yes

Lab

online

3 to 5

USA

Bates and Gupta (2017): Study 1

5

Yes

Lab

F2F

2 to 4

UK

Bates and Gupta (2017): Study 2

5

Yes

Lab

Both

3

India

Bates and Gupta (2017): Study 3

5

Yes

Lab

Both

3

UK

Rowe (2019, unpublished doctoral thesis)

5

Yes

Lab

F2F

2 to 5

Australia

Mean or Ratio:

6.3

7:12

14:5

 

na

11:8

  1. Note: The table outlines empirical studies on collective intelligence and group performance published between October 2010 and November 2019. n = number of groups. K = number of independent samples. Effect 1 = Percentage of total variance in group IQ composite explained by first factor/component; Effect 2 = The proportion of positive correlations within the correlation matrix comprised of bivariate Pearson's correlations, r, between group IQ test items (positive manifold test); Effect 3 = Bivariate Pearson's correlation, r, between c and a criterion task; Effect 4 = Bivariate Pearson's correlation, r, between Av.IQ and c; Effect 5 = Bivariate Pearson's correlation, r, between Av.IQ and Criterion task; ns = p > .05; RA = Random Allocation to groups; F2F = Face-to-face. C1 and C2 = Criterion task 1 and 2 of a single study; r = Pearson’s correlation coefficient. Mean and Ratio scores include only primary data and therefore exclude previously meta-synthesized results from Credé and Howardson ( 2017a, 2017b), and Bates and Gupta (2017): Studies 2 and 3 (combined)
  2. aPooled data from secondary sources (≥ 2 studies) not included in the present analysis.
  3. b This pertains only to group IQ testing context and not to the group performance setting
  4. cCorrelation exists for the EG only (the c-factor was not apparent in the CG)
  5. dPaper originally added as a conference proceeding (Aggarwal & Woolley, 2014)
  6. eResult was originally reported in R2 value, controlling for intercept and team size, then transformed to a correlation coefficient using square root(.34) = .58 (see Aggarwal et al. 2019, p. 6)
Back to article page