When I present the development of Tata Nano through the lens of the innovation sandbox process, some people say, “But Nano is a failure”. Why do we find it hard to accept that Nano development might have followed a good process? Because, once we know that Nano is a failure, every attribute of Nano gets tainted by the halo created by the failure, kind of a dark shadow. That is what called “Halo Effect”. It is a cognitive bias in which observer’s overall impression of a person, company brand, or product influences the observer’s feelings and thoughts about that entity’s character or properties.
Once you like Narendra Modi, you start liking other things about him – his diet, his dress, his oratory skills, his exercise regime etc. I turned to “Halo Effect” by Phil Rosenzweig because it is highly recommended by one of my favourite experts on decision making, Daniel Kahneman. Here are 3 of my takeaways from the book illustrated through the examples also from the book:
- Halo spreads through heuristic substitution: John Kotter and James Heskett from Harvard Business School did a study to find how corporate culture is related to the company performance. Their survey questionnaire had questions like: rate the strength of the corporate culture on a scale of 1 to 5, rate how the company culture fits its environment on a scale of 1 to 7, how much does the company culture values excellent leadership (scale 1 to 7) etc. In all these cases, Kotter and Heskett found high correlation between these attributes (culture, environmental fit, leadership etc.) and performance of the company. That, according Rosenzweig, is not surprising because when asked a question about, say leadership: How is the leadership of the company? We tend to answer an easier question: How is the company performing? This substitution happens without conscious awareness of the person answering the questionnaire. In effect, the variable such as leadership is no longer independent of the company performance. It is under the Halo of the outcome (in this case, performance). Thus Halo spreads through heuristic substitutions which are automatic.
- Big data can’t negate the Halo: “Good to Great” by Jim Collins is one of the bestsellers in management literature selling over four million copies since its publication in 2001. The book identified seven characteristics of companies that went from “good to great”. The research involved studying “6,000 articles, generating more than 2000 pages of interview transcripts and creating 384 mega-bytes of computer data in a five-year project”. The research started with 1,435 Fortune 500 companies then narrowed down to eleven which fit the criteria of “good to great”. The criterion was – fifteen years of stock market returns near the general average and then a transition to a period of fifteen years of stock market returns well above average. Then they thoroughly studied the eleven companies. Some of the parameters were free of Halo Effect – e.g. to manager turnover, extent of board ownership etc. However, they also interviewed managers at these “Good to great” companies and asked questions like – “How did the company get commitment and alignment?”
Rosenzweig says, “Interview questions of this nature, where managers are asked to look back and explain what happened, rarely produce valid data, since retrospective self-reporting is commonly biased by performance.” Does good leadership lead to good performance? Perhaps. But perhaps, good performance also leads to seeing the leadership as good. This gets muddled up with Halo Effect. “Good to great” does not even acknowledge that some of the data could be biased.
- 9-point delusion checklist: If Halo Effect can creep in unconsciously and not go away in spite of large data collection, then what does one do to tackle it? Rosenzweig gives a checklist of 9 delusions which can be used to check if your approach is bias prone. Some of these delusions are: the delusion of correlation and causality, the delusion of single explanation etc. Perhaps over a period these checks may become automatic.