How To Make A Good List
I’ve been creating lists for seven years. Since I’ll soon be venturing into other corners of the Web, I was asked to write down a list of best practices.
I think that Jonah Peretti and Nate Silver have proven that the culling of data and the listing of things can be an incredibly lucrative specialty in modern “journalism.”
I’m not pretending to be nearly as fluent (or even as proficient) in the language of lists as those two are, or anyone at any other news organization for that matter. But, there was something about this exercise in trying to nail down the tactics I’ve grown partial to in the last years that I found fascinating. My rules to rank are as follows:
The lists that seem to generate the most traffic are:
-Negative. Positive lists are great and do well (i.e. the happiest colleges, the best cities for jobs, the most influential people). But for much the same reason than gossip and scandal are popular, so are negative lists. Druggiest colleges, miserable cities, fattest cities, drunkest cities, dumbest states, worst cities to get a date, most hypocritical politicians. There are caveats, of course. Money and gadgets come to mind; the wealthiest person is much more interesting than the poorest, and a list of the best apps is far more clickable than the worst.
-Simple to understand. Methodologies can be as complicated, unique and interesting as a writer wants, or as the topic demands, but the essential idea behind a list should be basic. It’s not hard to understand what a list of America’s richest people will be, or the country’s best high schools, despite the complications of compiling or digesting the data. So lists built around very elementary adjectives — best, worst, drunkest, dumbest, smartest — usually work best. Lists that rank entities based on concepts that are slightly ineffable, like innovation or vanity or humor, seem to be less popular (or perhaps they are intrinsically more niche), though sometimes are more journalistically rich.
-Universal. It’s pretty obvious: to attract the largest audience, appeal to the most people. There are some institutions that seem to hit a nerve: high school, college, summer camp, weather, and celebrity. I’m sure I’m missing some. In some ways, the “nut graph” of really popular lists is just that the data or topic at-hand appeals to basic reader curiosity. City and state lists do well, too, I think because they plays to the competitive instinct in and identity of readers, which relates to the next point…
-Ranked. I am biased because I’ve never really produced unranked lists (or, lists without a backbone of data). But, reality is chaotic, and lists are psychologically energizing.
Best Practices for Original Lists:
From my point of view, lists need to live up to their headlines. The sexier the headline, the more rigorous the data collection should be. The standards I’ve been operating under, in general, are as follows:
-Use at least three data points. Drunkest cities could simply be the city whose citizens drinks the most alcohol, but that’s a little shallow. So, think about average drinks per person per week, most binge nights on average, and the most number of people per capita in AA (just brainstorming). The idea is to give people a little more context and information. Also, avoid making readers feel cheated.
-Ask experts. If it’s a list that’s complicated, call someone who knows what you should consider. When we were developing the methodology for our best high school lists, we called experts in education, read case studies about class size, looked at research on wealth and secondary education, and looked into indicators of post-secondary success. I always think about that journo adage “we don’t print the facts, we print what people say.” This is especially true with lists. We don’t necessarily print the “truth,” we print what the numbers say.
-Always normalize data. If you’re going to rank the drunkest cities, you don’t rank cities by the amount of beer consumed. You rank based on the amount of beer consumed per person. If you do the cities with the worst drivers, you wouldn’t look at the total number of automobile deaths, you would look at the total number of deaths per licensed driver. Data needs to qualified, not just quantified.
-Be transparent. Always publish the data points used, the sources and the weighting. And, when there are limitations to the data, always clarify. A simple to understand example is crime statistics, especially rape stats. If Chicago has a higher rape rate than New York City, this may mean Chicago is especially adept at prosecuting rapists and encouraging victims to come forward, NOT that more rapes occur. Sometimes data has significant context that needs to be addressed.
-Every list is doable, but not every list is worth it. Lists are labor-intensive. They are great ways to get traffic, but sometimes they are not the best way to tell a story. And, sometimes they are too time-consuming. If data for a list isn’t available from a reputable source, create a panel of experts and poll them.