After taking on a behavioural economics project, Amy Mitchell and I chose to test the relative strengths of the bandwagon effect and altruism bias within PGS in a donation scenario. The bandwagon effect, sometimes known as herd behaviour, is where our behaviour copies what the majority has done before. For example, buying a new product as it is popular with consumers. Altruism bias, however, is where our behaviour gravitates towards the fairer or more just option.
Our main hypothesis was that the bandwagon effect would override the altruism bias and be the main dictator of test subjects’ behaviour. The bandwagon effect is a powerful cognitive bias, which drives much of our behaviour based on strong egos or social norms - effectively our entire understanding of what is appropriate behaviour. However, with donation framing, altruism bias could be emphasised to the extent that it would override this bias.
Before school on a Thursday morning, we followed a scientific method, using both a control and test group to check that our results would be valid. Taking a control group of 12 (from different year groups and genders), we offered each subject a choice of two jars, one labeled “John” and the other labeled “Jeremy”. We framed each trial with the following scenario:
“We have been raising money for the homeless in Portsmouth and would like your help to decide who we should give it to. John and Jeremy are both from Portsmouth and are equally in need. Please cast your vote.”
The subjects then took a counter to place their vote in either jar, removing it afterwards to reset each trial. We repeated this method with a test group of 10 (from different year groups and genders), now with the jar labeled “John” ⅓ full with extra counters. We asked each test subject to cast their vote. When subjects were in groups, we gave the briefing together, but kept each vote anonymous by asking those not immediately taking part to turn around. Through this, we ensured each vote was genuine, independent, and not affected by peer-pressure or additional bandwagon effect.
These results show that altruism bias is much stronger than the bandwagon effect in this scenario, likely due to the heavy altruistic framing. 100% of test subjects chose the jar with no counters in it, showing a negative nudge from the ⅓ full jar. This was caused by a greater perceived unfairness compared to when both jars were equal, as each vote represented greater wellbeing from the same initial welfare.
Of course, no experiment is without its limitations. These results are only relevant for PGS as there was no variance in this aspect - meaning that the data are unlikely to be replicated exactly in other scenarios. However, within our samples we incorporated a variety of ages, genders, and cliques. Furthermore, there could be biases based on the names used in the test. “John” or “Jeremy” could have emotional value for some subjects (for instance, being the name of a family member) that would affect their choices. However, I believe the increased humanity gained by using names within the experiment led to a more realistic representation of the scenario, and the data does not appear to have been skewed to a large extent. The main limitation of this experiment was the small sample tested. Altruism bias is likely to affect a smaller percentage of subjects’ behaviour if a larger sample is taken, due to the fact it is unlikely that the bias would consistently change 100% of behaviour. This calls for increased testing beyond our means as students (with a lack of time and ability to travel) in order to find a more representative result for both sample groups. Finally, it is true that the probability of choosing “John” in the control group was not 0.5, however this is likely due to the small sample size. Nevertheless, through statistical analysis, it can be seen that the raw data is not significantly different from the expected result and so it cannot be said that the true probability is not 0.5.
P = probability of choosing “John”
H0: p=0.5 H1: p≠0.5 α = 0.1 / 2 = 0.05 Actual Critical Range = x<2, x>9
Therefore, 5/12 is not significant, thus the null hypothesis cannot be rejected. The result 5/12 is not within the critical range, meaning that it cannot be asserted to a reasonable degree of certainty that the probability of choosing “John” over “Jeremy” was not 0.5.
Overall, this project has shown that, within the PGS community, altruism bias overrides the bandwagon effect when a situation has an altruistic framing. Due to the limitations of our sample groups, this may not be the case for other samples from different areas and suggests that further testing should be undertaken. Furthermore, it is important to compare this to real-life examples of similar situations to evaluate this finding’s relevance. For example, we should look at how this altruistic leaning affects supermarket counter voting systems when there are varied causes and very personal attachments to different causes. A point to consider in this is that: when the time taken to make the decision isn’t the same, different skews may occur. For instance, as these supermarket charity voting boxes are placed on the way out of the shop, consumers are moving faster and will likely not stop to read the brief descriptions of each option. Therefore, it is probable that the bandwagon effect is stronger there as public opinion is the easiest information to comprehend. Following the proximity bias, if perfect information is difficult to attain, consumers will revert to the most obvious or prominent piece of information (which in this case is the visually displayed public opinion). Nevertheless, the findings of this experiment show that altruism bias is far stronger than the bandwagon effect when a scenario is framed altruistically. So, our ingrained instincts to follow the herd really can be overridden by perceived unfairness.