Guest blog from the LSE Behavioural Research Lab
As you may know, LSE’s motto is ‘Rerum cognoscere causas’ or ‘to Know the Causes of Things’. Although it was adopted almost a century ago, in 1922, its message still resonates today and guides the focus of LSE education and research.
Indeed, the title of the LSE100 course, compulsory for all undergraduate students across the School, is ‘Understanding the causes of things’. This not only gives a nod to the School’s motto but also indoctrinates students with the way social scientists use evidence to shape their thinking; it provides them with a broad understanding of the social sciences and teaches them that multiple-perspectives are needed to understand today’s complex problems. Students come away with an expanded understanding of social sciences and more developed critical thinking skills, providing a solid foundation for them to build upon as they continue their coursework.
In addition to being at the core of LSE teaching, the quest to know the cause of things is also at the heart of research conducted at LSE. Many methodologies contribute to this knowledge and LSE’s Methodology Department provides a wealth of expertise across disciplines; however, experimental research offers the most convincing evidence for causal relationships as it provides researchers with the evidence that characterises causal relationships between two factors (Shadish, Cook, & Campbell, 2002). Specifically, experimental research provides evidence of the temporal relationship between the cause and effect, the change in the cause corresponding with a change in the effect, and, as much as possible, the removal of other factors that could also cause the effect.
Despite being a fairly intuitive and straightforward set of criteria, as outlined below, each of these characteristics can be difficult to achieve when conducting research. Nevertheless, by employing rigorous research, like the research conducted in LSE’s Behavioural Research Lab (BRL), convincing evidence can be collected and used to support a hypothesised causal relationship.
When evaluating whether there is a causal relationship between factors, the cause should occur before the effect. Although a seemingly obvious point, its importance is sometimes lost in practice or can be difficult to determine at times.
For example, if a survey is used to examine the relationship between a leader’s level of confidence and their followers’ level of participation at work, it wouldn’t be possible to determine whether the perceived confidence of the leader occurred before the follower’s participation at work as both measures were taken at the same time.
However, if a researcher used an experiment to examine this relationship, they would be able to precisely examine the temporal relationship between the two factors. In fact, this is precisely what BRL researcher, Dr Connson Locke did in her The downside of looking like a leader: Power, nonverbal confidence, and participative decision-making paper: she manipulated the level of confidence of the leader before observing followers’ behaviour on a subsequent task. As such, her experiment provided evidence for the temporal relationship between the cause and the effect.
Another characteristic of a causal relationship is that changes in the cause are associated with changes in the effect. Essentially, if you change the level of the cause, this should correspond with a change in the effect. For example, if a policy was introduced that used an incentive to change behaviour, a change in the incentive should be associated with a change in the behaviour – that is, they vary together.
Research conducted by BRL researchers, Paul Dolan, Matteo M. Galizzi and their colleague, Daniel Navarro-Martinez, found just this in their paper, Paying people to eat or not to eat?: carryover effects of monetary incentives on eating behaviour paper. Specifically, they found that incentives that encouraged people to eat or not eat resulted in people eating more when they were paid to do so and eating less when they were paid to do so. However, rather than just observing this relationship at one time point, they examine how an incentive produced a carryover effect across two additional time points.
What they found shows that incentives produced different long-term effects: paying people not to eat continued to shape behaviour, but paying people to eat did not persist across the later time points. Hence, the covariation of the incentive and the behaviour occurred for both conditions while the incentive was present, but had a long-term effect only when the incentive encouraged not eating.
Ruling out Alternatives
Perhaps the most difficult characteristic to meet is the ruling out of alternative causes. Clearly an effect can be caused by a number of factors, but researchers are limited in what they are able to examine in each study. Much of what determines what alternatives should be considered is based on theory, previous studies, and reasoning. Nevertheless, laboratory experiments allow researchers to control the participants’ environment, ruling out additional factors that might impact the relationship between the factors of interest.
For example, in their Observing workplace incivility paper, BRL researcher Tara Reich and her colleague, M. Sandy Hershcovis, examined how observers reacted to incivility between two strangers in the BRL. This was done to rule out alternative factors that could possible shape the observer’s behaviour (i.e., a pre-existing relationship), and allowed them to specifically examine how aspects of the interaction shaped the observer’s response.
When taken together, it is clear that knowing the cause of things is a lofty goal; nevertheless, conducting rigours research produces evidence that can be used to build a case in support of causal relationships. Perhaps knowing the cause of things is less about the specific knowledge acquired but is, instead, about the process of knowledge acquisition.
Indeed, scientific research is constantly striving to improve this process as is evident in the recent effort to reproduce findings, which BRL research Heather Kappes contributed to in the Investigating variation in replicability: A “Many Labs” replication project and Estimating the reproducibility of psychological science papers and reflected on recently.
Given the capabilities and complexities of today’s environment, it is increasingly important to have well-developed critical thinking skills to be able to use and interpret evidence effectively. The specific knowledge that is known today will surely change and evolve as new evidence is acquired, and having a more refined understanding of how this process develops will certainly contribute to knowing the cause of things.
Tamara Ansons is Behavioural Lab Projects Administrator in the LSE Behavioural Research Lab.