Some combinations just work: a playlist of the year’s top techno tracks; a compilation of short stories about famous entrepreneurs; a flight of Japanese whiskies. Then, there are others that may seem a little strange, like a lone oil painting in a sculpture exhibition or a bright green ensemble in a fashion collection of otherwise muted hues.
In a paper recently published in Organization Science, we investigate how the way elements are combined affects the evaluation of the individual parts. Building on the idea that audiences cannot fully disentangle the quality of “fit” between elements from the quality of the elements themselves, we posited that audiences are likely to direct their dislike of a product with misfit features to the individual elements due to their inclusion in a novel combination.
The whole vs. its parts
We focused our analysis on combinations that are akin to “salad bowls” rather than “melting pots”. While the latter fuse together different components to create something new – like romantic comedies that mesh aspects of romance and comedy genres – the former contain elements that can be independently identified, consumed and evaluated, such as an album composed of discrete songs.
The quality of any creation made by combining different elements is affected by at least two factors: the quality of each element and the extent of fit (or misfit) between them. We hypothesised that individual elements of a product will be rated less favourably when the fit between its different elements is poorer. This effect may be exacerbated when people first conduct a holistic evaluation of a product before assessing its components, as this prompts them to see misfit more saliently.
While individual quality and fit are in principle independent from each other, evaluators may not fully distinguish them. This is influenced by the halo effect – when evaluators fail to assess distinct features independently. When this happens, the assessment of one attribute can bias the evaluation of other attributes either through direct spillover or by affecting the general impression of the product.
Mismatches in the music realm
In our first set of analyses, we examined a sample of over 350,000 songs on just under 32,000 albums released between 1998 and 2005. We wanted to examine how the extent of misfit of tracks on an album influences the success rate of individual songs.
We expected that tracks included on albums with poorer fit would face a penalty and therefore perform worse in the market than those on albums with better fit. Furthermore, we predicted that as albums became less important as units of consumption, this penalty would decrease.
To test the first argument, we constructed a measure of misfit at the album level using the sonic features (e.g. valence, acoustic measure, energy and tempo) of tracks included on the album. We then used these sonic features to calculate the fit between tracks.
We found that individual songs were more likely, on average, to become independent hits when the albums they appeared on had high levels of fit. Meanwhile, songs on albums with poorer fit were less likely to appear on the Billboard Hot 100 charts, as well as to remain on the charts, compared to songs from albums with better fit. Incorporating data on sales, radio plays and streaming, the Billboard Hot 100 charts are the standard record charts for songs in the United States, commonly used as a measure of popularity and performance.
To test the second argument, we leveraged an important shock in the music industry: digitisation that led to a dramatic shift in consumption beginning in 2001, from predominantly CD albums to single tracks. We found that the positive effect of album fit on chart success dropped significantly after 2001, suggesting that fit did not matter as much once consumers changed their listening habits.
One of these things is not like the others
We also conducted an online experiment in which 2,330 participants were presented with a gallery of three photographs. Participants were asked to rate how much they liked the gallery and each of the individual images. In the “fit” condition, they were shown either three black-and-white portraits or three wildlife photographs. In the “misfit” condition, they saw either two portraits and a wildlife photograph, or two wildlife photographs and a portrait. We also asked participants to assess the fit between the photographs.
Galleries with three photographs of the same type were perceived to have considerably better fit than galleries that included an outlier photograph, and misfit galleries were rated lower overall than fit galleries. What’s more, individual photographs were evaluated less favourably when they appeared in misfit galleries compared to fit galleries, even when excluding the ratings of outlier photographs in misfit galleries.
Implications for innovation and organisations
Our results suggest that penalties stemming from misfit combinations can spill over from the whole to its individual elements. When assessing such products, consumers will not only react more negatively to the product itself but will also evaluate its individual components more harshly.
This has important implications for innovation. For decades, research has emphasised the critical role of recombination in the development of new ideas, products and services. Bringing together unrelated, distant elements is a difficult process but is also a key driver of breakthroughs. Our research highlights a previously unrecognised cost of such misfit combinations – they may not just lead to the failure of the whole; they may also lead to the failure of its parts.
Beyond the evaluation of products, our findings might also apply to a broader set of organisational contexts. The misfit bias applies to any setting where three conditions are met: individual elements are brought together into a larger combination; the performance of the whole depends on the quality of individual elements and the fit between them; and the elements themselves face subjective scrutiny.
This dynamic is present when assessing how individual firm resources or routines contribute to overall firm performance, as well as evaluating how individual team members contribute to overall team performance. In both cases, the performance of the “whole” may be unambiguous – for instance, the amount of firm profit or total team output – and a function of the quality of each element. Yet, outsiders, and even insiders, are often unable to fully understand the specific relationship between each element to overall performance. As a result, the misfit bias could come into play.
Our theory suggests that evaluators are likely to undervalue specific firm resources when they are observed in misfit combinations. Similarly, the allocation of credit is shown to be notoriously difficult in teams. To overcome this, managers could evaluate an individual employee’s contribution to a team by observing the changes in team performance when specific individuals are present vs. absent.
In many settings, this type of experimentation can be a helpful approach to overcoming the misfit bias and forming more accurate assessments of the quality of individual elements – be that an artwork in a gallery, an organisational resource or an individual team member.
