The plural of anecdote is data

“The plural of anecdote is data.” That observation, by cognitive psychologist and “skeptical activist” Barbara Drescher, is not an apology for sloppy logic. Nor is it a rationalization for using a string of stories to prove a point without relying on hard data.

Drescher made the comment while discussing the Texas Sharpshooter Fallacy on the You Are Not So Smart podcast (which, by the way, is highly recommended overall). Her point, which was almost an aside during the larger conversation, is anecdotes and data are not opposed to each other. A dataset, really, is a large sample of anecdotes. Treating one as better than the other can leave an incomplete picture, and this observation is particularly relevant to entrepreneurship research.

There are, very roughly, three broad kinds of information about entrepreneurship. There are stories – this is the stuff of magazine articles, conference presentations, and media interviews. There are quantitative data – surveys and administrative data from the Census Bureau based on tax records and other sources.

And there is a hazy middle category that a lot of people label as data but is more or less an aggregation of specific cases and stories. It is mostly descriptive and sometimes devoid of context and analysis: the “knowing narrative,” let’s clumsily call it.

None of these is always better than the others, but everyone has their own implicit ranking of the different types of entrepreneurship information. Academic researchers, understandably, view quantitative data as vastly better than anything else – and even within this category, administrative data are preferable to survey data. The Kauffman Foundation has expended a lot of resources to improve the state of data on firm formation and business dynamics. How can you make policy or design programs without having a large sample and running regressions to unpack the internal dynamics of business creation?

Others, while agreeing that large-scale datasets are useful, see stories as more valuable. Quantitative data abstract away from the real stuff of entrepreneurship – the anxiety, the thrills, the ups and downs, and human element that makes entrepreneurship so interesting to study and write about. How can you make policy or design programs without understanding entrepreneurship from the ground up, which can only be gained through stories?

The middle category is no less valuable – not every collection of data points needs to be accompanied by an ordinary least squares analysis. And there is a limit to what discrete inspirational stories can tell us. Sometimes it is enough to pull together a lot of different stories, collect some data from them, and describe what we’ve found. How can you make policy or design programs without a coherent narrative that tries to fill the gaps in data with the color of real life?

Drescher’s comment is a reminder that, despite enormous progress, we still have lots of room for improvement in gathering information about entrepreneurship. Large-scale data is valuable, but every data point – as Drescher points out – is actually its own story, with its own context and idiosyncrasies. Stories are valuable because that’s how humans learn, but such anecdotes can also be limited because their purported lessons are sometimes useless.

Gathering information is only one step, of course – information must be communicated and conveyed in effective ways. Sometimes, large-scale data sets are the best means of communication; other times, more structured narratives – informed by data – are a better means of conveyance.

There are still gaps to fill, still knowledge to unearth, if we want to better understand – and thus help – entrepreneurs.