FOLLOW THE CROWD Social Information and Crowdfunding Donations

  • Slides: 23
Download presentation
FOLLOW THE CROWD Social Information and Crowdfunding Donations in a Large Field Experiment ISTR|

FOLLOW THE CROWD Social Information and Crowdfunding Donations in a Large Field Experiment ISTR| July, 12 at VU Amsterdam| The Netherlands Claire van Teunenbroek & Rene Bekkers | Center for Philanthropic Studies, VU Amsterdam

Social information • The use of others as a suggestion to influence decision-making (van

Social information • The use of others as a suggestion to influence decision-making (van Teunenbroek, Bekkers, & Beersma, 2018) • Which decision would we like to influence? – Participation: yes/no – Donation amount: € • Often used by practitioners 2

Crowdfunding • Social information effect is context specific • Online context – Giving to

Crowdfunding • Social information effect is context specific • Online context – Giving to small projects – A lot of small donations • Relatively new – Popular but not very effective (<1% of the giving in NL and USA • Helping hand: can we increase donation amounts? • Research question: what is the influence of social information on online donation behavior through a crowdfunding platform? ” 3

Innovations • • • Preregistered: https: //aspredicted. org/u 5 w 9 u. pdf Natural

Innovations • • • Preregistered: https: //aspredicted. org/u 5 w 9 u. pdf Natural field experiment Sample was heterogeneous Donations are real We tested the effect over multiple projects, at multiple times. 4

Earlier research suggests that it could work Bijv: Alpizar et al. , 2008 a;

Earlier research suggests that it could work Bijv: Alpizar et al. , 2008 a; Bekkers, 2012; Martin & Randal, 2008; Shang et al. , 2012; van Teunenbroek, 2016. • Visitors to a museum donate similar amounts as previous donors 5

Hypothesis 1 • Social information hypothesis: Social information (IV) increases the donation amount (DV)

Hypothesis 1 • Social information hypothesis: Social information (IV) increases the donation amount (DV) at a crowdfunding platform. 6

Why is one suggestion enough? • Social information provides a norm which humans like

Why is one suggestion enough? • Social information provides a norm which humans like to follow (van Teunenbroek et al. , 2018). – Giving is a social process (Bekkers & Wiepking, 2011) – It could also have a negative effect • Social norm hypothesis (2): When the amount displayed in the social information condition is larger than the intended donation amount, the amount donated increases. 7

But is does not always work • Sometimes the effect is absent – Catt

But is does not always work • Sometimes the effect is absent – Catt & Benson, 1977; Kubo, Shoji, Tsuge, & Kuriyama, 2018; Murphy, Batmunkh, Nilsson, & Ray, 2015; Shang & Croson, 2009 • Or even negative – Croson & Shang, 2008; Meyer & Yang, 2015 8

Do I believe this? • Humans are driven by saving money (Sargeant & Jay,

Do I believe this? • Humans are driven by saving money (Sargeant & Jay, 2004) – If the expectation in terms of the donation amount is low, donors donate lower amounts (Bekkers & Wiepking, 2011) – A suggestion that is too high could have a negative effect (Hysenbelli, Rubaltelli & Rumiati, 2013) • Resistance hypothesis(3): Social information has a negative effect on the number of donors 9

Project funding period • Crowdfunding projects are online for several days • Social information

Project funding period • Crowdfunding projects are online for several days • Social information could work as a quality signal (Vesterlund, 2004; van Teunenbroek et al. , 2018) • Project funding period: different motives – Start: family and friends (Borst et al. , 2018) – Middle: donors without a connection with initiator (Borst et al. , 2018) • Low identification with the group > focus on quality signals (Fishbach et al. , 2011) – Final: family and friends (Borst et al. , 2018) 10

Project funding period • Project funding period (4): social information is less effective in

Project funding period • Project funding period (4): social information is less effective in the beginning as well as towards the end of the campaign, and most effective in the middle of the campaign. 11

Methods • Study context – Voordekunst crowdfunding platform (n=24. 070) – 15 September –

Methods • Study context – Voordekunst crowdfunding platform (n=24. 070) – 15 September – 16 October 2016 • Study design (https: //aspredicted. org/u 5 w 9 u. pdf) – 2 dependent variables • Giving yes / no (h 2) • Donation amount (h 1, h 3, h 4) – A/B test with two conditions • Control: no social information (n = 1. 374) • Manipulation: social information (n = 1. 283) "Did you know that the average donation amount at Voordekunst is € 82? "

van Teunenbroek & Bekkers (2018) 13

van Teunenbroek & Bekkers (2018) 13

Methods • Study context – Voordekunst crowdfunding platform (n=24. 070) – 15 September –

Methods • Study context – Voordekunst crowdfunding platform (n=24. 070) – 15 September – 16 October 2016 • Study design (https: //aspredicted. org/u 5 w 9 u. pdf) – 2 dependent variables • Giving yes / no (h 2) • Donation amount (h 1, h 3, h 4) – A/B test with two conditions • Control: no social information (n = 1. 374) • Manipulation: social information (n = 1. 283) "Did you know that the average donation amount at Voordekunst is € 82? "

Results (1) • Donors in the manipulation condition donated about € 11 higher (F

Results (1) • Donors in the manipulation condition donated about € 11 higher (F = 4. 72, n = 2, 638, p =. 030) – This is about 16% and in line with previous research 15

Results (2): social norm • There is a difference between the categories (X 2=9.

Results (2): social norm • There is a difference between the categories (X 2=9. 15, p=. 010). • Treatment condition had donations less around 25, but more high donations (>500) 16

Results (3): resistance • There was no difference in the number of donors between

Results (3): resistance • There was no difference in the number of donors between the two conditions – 9% (X 2 = 1. 45, p =. 221), Manipulation Control Total Site visitors 11, 973 12, 097 24, 070 Donors 1, 283 1, 374 2, 657 17

Results (4) project funding period • Social information has mainly an effect in the

Results (4) project funding period • Social information has mainly an effect in the middle 18

Results (4): project funding period • Significant effect in the beginning (F = 1.

Results (4): project funding period • Significant effect in the beginning (F = 1. 54, n = 1, 539, p =. 035) • No effect in the middle (F =. 89, n = 484, p =. 346) • Largest effect in the final stage (F =. 09, n = 615, p =. 769) Results are from a non-parametric Kruskal-Wallis test. 19

Conclusion (1) • 16% increase in the donation amount • Social norm (h 2):

Conclusion (1) • 16% increase in the donation amount • Social norm (h 2): no support • No smaller SD at the treatment condition • Possible that 82 is too a-typical to function as a norm • Resistance (h 3): no support • No difference in the number of donors (9%) • Funding period (h 4): no support • Largest at the end of a campaign • Donors at beginning and end uncertain 20

 Thank you Claire van Teunenbroek, MSc. https: //crowdfundingpscvt. wordpress. com/ @PSCTeunenbroek p. s.

Thank you Claire van Teunenbroek, MSc. https: //crowdfundingpscvt. wordpress. com/ @PSCTeunenbroek p. s. c. van. teunenbroek@vu. nl Prof. Dr. René Bekkers https: //renebekkers. wordpress. com/ @renebekkers r. bekkers@vu. nl

Table 2. Differences in number of views, number of donations, donation amounts and number

Table 2. Differences in number of views, number of donations, donation amounts and number of outliers by condition Treatment Control All Number of viewers 11, 973 12, 097 24, 070 Number of donors 1, 283 1, 374 2, 657 Mean amount donated 97. 53 94. 33 95. 87 (SD = 310. 24) (SD = 347. 20) (SD = 310. 24) 35. 00 30. 00 35. 00 (SD = 146. 96) (SD = 347. 20) (SD = 310. 24) Number of outliers 9 10 19 Mean amount donated excluding outliers 69. 54 80. 73 74. 95 (SD = 116. 74) (SD = 146. 96) (SD = 132. 29) Median amount donated excluding 35. 00 30. 00 35. 00 outliers (SD = 146. 96) (SD = 116. 74) (SD = 132. 29) Median amount donated 14/06/2018 van Teunenbroek & Bekkers (2018) 22

Table 3. Overview of the donation amounts per condition and project funding stage Outliers

Table 3. Overview of the donation amounts per condition and project funding stage Outliers Condition Begin stage Middle stage Final stage M SD n M SD N Yes Control 65. 71 194. 50 817 184. 56 697. 24 240 123. 33 283. 66 301 Yes Treatment 72. 96 174. 72 727 137. 12 410. 43 255 99. 75 225. 97 317 Yes Total 69. 13 184. 87 1, 544 160. 12 567. 78 495 111. 24 255. 76 618 No Control 55. 21 78. 73 814 91. 00 152. 58 234 90. 57 156. 33 299 No Treatment 66. 27 116. 63 725 89. 94 154. 77 250 108. 10 194. 58 316 No Total 60. 42 98. 54 1, 539 90. 45 153. 56 484 99. 10 176. 04 615 14/06/2018 van Teunenbroek & Bekkers (2018) 23