Last Friday afternoon, it being pretty miserable outside, I dusted off my Stats 101 prowess and decided to look for correlations. The next thing I knew, three hours had passed and I was earlobe deep in data tables and spreadsheets.
Yeah -- that’s how I roll. That’s wassup.
But I digress. What initially sent me down this path was a new study out of the University of Kansas by Tien-Tsung Lee and co-authors Masahiro Yamamoto and Weina Ran. Working with data from Japan, they found that the amount of trust you have in media depends on the diversity of the community you live in. The more diverse the population, the lower the degree of trust in media.
This caught my attention: negative correlation between trust and diversity. I wondered how those two things might triangulate with innovation. Was there a three-way link here?
So I started compiling the data. First, I wanted to broaden the definition of innovation. Originally, I had cited the INSEAD Global Innovation Index. Bloomberg also has a ranking of innovation by country that uses a few different criteria. I decided to take an average, normalized score of the two together. In case you’re wondering, Switzerland scored much lower in the Bloomberg ranking, which had South Korea, Japan and Germany in the top three spots.
With my new innovation ranking, I then started to look for correlations. What part, for example, did trust play? According to Edelman, the global marketing giant, who publishes an annual trust barometer, it plays a massive role: “Building trust is essential to successfully bringing new products and services to market.” The company's trust barometer measures trust in the infrastructural institutions of the respective countries. So I added Edelman’s indexed trust scores to my spreadsheet and used a quick and dirty Pearson r-value test to look for significant correlations.
For those as rusty as I when it comes to stats, a perfect correlation would be 1.0. Strong relationships show up in the 0.6 and above range. Moderate relationships are in the 0.3 to 0.6 range. Weak relationships are 0.3 and below. Zero values indicate no relationship. Inverse relationships follow the same scale, but with negative values.
The result? Not only was there no positive correlation, there was actually a moderately significant negative correlation! For those interested, the r-value was -0.4224. Based on this admittedly amateur analysis, trust in national institutions and innovation do not seem to go hand-in-hand. Some of the most innovative countries are the least trusting -- and vice-versa. It certainly wasn’t the neat supposed linear relationship that Edelman referred to in its press releases for its barometer.
Next, I turned to the obvious: the wealth of the respective nations. I added GDP per capita as a data point. Predictably, there was a strong positive correlation here – I came up with an r-value of .793. Rich countries are more innovative. Duh.
Now comes the really interesting part. What was the relationship between cultural diversity and innovation? If my original hypothesis was correct, there should be at least a moderate correlation here. The problem was trying to find an accurate measure of cultural diversity. I ended up using three measures from Alesina et al: Ethnic Fractionalization, Linguistic Fractionalization and Religious Fractionalization. I averaged these out and indexed them to give me a single score of cultural diversity. To my surprise, my hypothesis appeared to be significantly flawed: The r value was -0.2488.
But then I started analyzing the individual measures of diversity. Ethnic Diversity and Innovation showed a moderate negative correlation: -0.5738. Linguistic Diversity and Innovation showed a less significant negative correlation: -0.3886. But Religious Diversity and Innovation came up as a moderate positive correlation: 0.4129! Of the three, religion is the only measure of diversity that’s directly ideological, at least to some extent.
This seemed to be promising, so I pushed it to the extreme. If religious diversity shows to be correlated with innovation, I wonder how the prevalence of atheists would relate? After all, this should be the ultimate measure of religious ideological freedom. So, using a combination of results from a worldwide Gallup survey and a study from Phil Zuckerman, I added an indexed “atheism” score. Sure enough, the r-value was 0.7461! This was almost as significant as the correlation between national wealth and innovation! Based on my combined innovation scores, some of the least religious countries in the world (Japan, Sweden and Switzerland) are the most innovative.
So -- ignoring for a moment the barn-door sized holes in my impromptu methodology and a whack of confounding factors -- what might this hypothetically mean? I’ll come back to this intriguing question in next week’s Online Spin.