In data analysis, it's essential to consider both direct and indirect relationships. Direct relationships occur when a construct directly influences another, but in many cases, you should also analyze indirect relationships. These are situations where a first construct affects the second through a third construct—the mediator variable. In this article, we'll explore mediation effects using two examples.
After researching the number of swimming incidents, such as drowning or emergencies at sea, our assumption was that warmer water would lead to fewer incidents. However, we overlooked a crucial factor—the number of people swimming. The warmer the water, the more swimmers there are, increasing the likelihood of an incident if we don't take this first construct (number of people) into account.
To explain what influences the intention to invest money on the Tokyo Stock Exchange, we assumed there would be positive and negative factors. We believed that trust in business partners would have a positive influence, while fear (3) would have a negative one. However, during our analysis, we found that fear (3) was statistically insignificant. This led us to investigate the possibility of mediation.
We hypothesized that trust could be connected in some way and decided to add a new connection to our model—that fear (3) has an influence on trust. As expected, we found that fear does have a significant negative influence on trust, but its impact on the intention to invest is indirect.
There are three types of mediation effects: complementary, competitive, and indirect-only mediation. In our example, we have an instance of indirect-only mediation, also known as full mediation.
When your research results don't align with your expectations, it may be due to mediation effects you haven't considered. Always be prepared to investigate potential mediator variables and their impact on your findings.
Let's discuss your project and find the best solution for your business.
