Tag Archives: sociology

Which Nation Is the Best Lover? Ask Nanaya.

Para la versión en español por favor haga clic aquí.

Let’s get straight to the chase.

Most Romantic Nations

International Net Romance Scores.

Higher score, better lovers?

Ok, well it’s awfully hard to date a whole country but we can describe the populations of a country. So how can we say which country is “the best lover” or simply “most romantic?”

We took our database of global personalities, picked out the countries with the most data, and weighted traits that impact romance to come up with different romance scores.

We came up with three types of romantic scores and found out:

  • Mediterranean countries have the most romantic populations and Latin-American countries have the least. English speaking nations are in the middle.
  • Love might be a universal language, but national language is less important than geographic proximity and shared culture.
  • The more romantic a country is, the more divorces there are for each marriage.
  • The more romantic a country is the higher unemployment is – but Nanaya needs more data to confirm this relationship.

How Did We Do This?

Nanaya will be a service that can forecast your love and social life – but running the Nanaya algorithm needs a big database of personalities. Since mid-January, Nanaya has hosted personality tests to build that database. Unlike many personality tests online, this test was built on a foundation of psychology producing scientific, repeatable results.

In the past few weeks, we’ve had well over 15,000 users around the world take the main personality test. If you haven’t taken it, do so here.

We can take those numerical personality test results to determine what makes a good lover, whether it’s for a hot fling or a long-term, stable relationship:

  • Hot Fling Score. Hot flings are all about adventure and exploration, of the world and each other. A part of this is being social and charismatic. Without these traits, no one makes the first move. Thoughtfulness is important, that way you can read each other’s emotions and respond accordingly, but it trails the others in value.
  • Stable Relationship Score. Stable relationships are a different matter. Things like reliability and thoughtfulness begin to matter more to make a relationship last. To be clear, this is more of a descriptive term than something that’s been correlated.
  • Net Romance Score. The right relationship is a mix of a hot fling and stable relationship. I add and normalized the scores to come up with the ranking at the top.

No doubt what makes for a hot fling and a stable relationship are related – but they’re not entirely the same. I designed “hot fling” and “stable relationship” scores by weighting various personality traits differently. We then calculated these scores for the distributions of personalities in the countries we had the most data on. There’s a lot more countries in our database than those in this study, but there’s not enough for “significant” results.


Best Nations for Romance

Below is a bubble chart that shows the Hot Fling and Stable Relationship Scores for these countries.

Romance Scores Grouped by Language Spoken

Bubblechart of Romance Scores with grouping by language. Red is Spanish, green is Portuguese, blue is English, and grey is Greek. The size of the circle denotes scaled standard deviation of the national personality distribution.
Bubblechart of Romance Scores with grouping by language. Red is Spanish, green is Portuguese, blue is English, and grey is Greek. The size of the circle denotes scaled standard deviation of the national personality distribution.

We expect these scores to be somewhat related. The fact they’re on a line means they’re highly correlated which makes perfect sense based on how we designed the scores. That said, they’re not totally the same. The differences emerge if we rank them separately.

Ranking of Hot Fling & Stable Relationship Scores

Ranking of studied countries for Hot Fling and Stable Relationship.
Ranking of studied countries for Hot Fling and Stable Relationship.

We’re sorry to report this, Argentina. If it helps, this is only a number and love is not a number.

However, our first guess that language was a uniting cultural feature was wrong. Let’s try a different way of grouping. Note, scores don’t change!

Romance Scores by Cultural Grouping

Bubblechart of Romance Scores with grouping by “culture.” Red is Latin American, Blue is English speaking/Protestant, and grey is Mediterranean (yes, we know Portugal is on the other side of Gibraltar). The size of the circle denotes is again based on standard deviation of national personality distribution.
Bubblechart of Romance Scores with grouping by “culture.” Red is Latin American, Blue is English speaking/Protestant, and grey is Mediterranean (yes, we know Portugal is on the other side of Gibraltar). The size of the circle denotes is again based on standard deviation of national personality distribution.

This grouping methodologies gives us much a more sensible, tighter distribution.

Now is there something unique about Mediterranean countries that leads to better lovers? Greece is not a Catholic country whereas Spain and Portugal predominantly are. One uniting feature is that all have relatively high unemployment.

Well there’s a thought, do unemployment and romantic score have a relation?  We’ll just pick Hot Fling Score as there is a fundamental correlation between with Stable Relationship & Net Romance Scores.

Hot Fling Score and Unemployment

Is this a geometric correlation!? Maybe we just need more data. Notice US & UK are almost entirely overlapping.
Is this a geometric correlation!? Maybe we just need more data. Notice US & UK are almost entirely overlapping.

Yikes! So the more amorous and passionate a nation, the less likely they’ll be at work or to have the stable institutions in place? There’s a lot of history, recent events, and other metrics that require delving into to prove this. I’m skeptical but it is interesting.

Well if there is a correlation, it’s not a linear one! Maybe we’ll revisit this when we have more data. For the curious, we got our unemployment data based on January 2015 values from here.


What about Gender?

So for a given country and trait, what do these differences look like across men and women?

Typically, men and women will have very similar distributions for a given trait in a specific country. We consider the “thoughtfulness” of Greeks below.

A histogram of the trait of Thoughtfulness of Greeks. Here we see the Greeks are very thoughtful, with a distribution skewed toward 100, which is the most Thoughtful score. Note that men and women are nearly identical.
A histogram of the trait of Thoughtfulness of Greeks. Here we see the Greeks are very thoughtful, with a distribution skewed toward 100, which is the most Thoughtful score. Note that men and women are nearly identical.

But this can be contrasted by a few cases which skew results across men and women, such as in the two different romance scores. Here, we look at the charisma of Brazilians.

A histogram of the trait of Charisma of Brazilians. Here we see that Brazilian women test as more charismatic than their male counterparts.
A histogram of the trait of Charisma of Brazilians. Here we see that Brazilian women test as more charismatic than their male counterparts.

This is just a brief overview of the difference gender makes. Stay tuned to the Nanaya Blog for future posts discussing the impact and role of gender in psychology and sociology!


Romantic Score and Romantic Success

Now all of this is for entertainment if we don’t see an actual correlation between romantic score and reality. But what’s a good reality check?

We need a publicly available indicator that tells us that people are getting into long-term relationships while tracking how fast people are leaving them. My guess is that more romantic countries are, the more they’ll be able to sustain relationships with fewer divorces for every marriage. Conveniently, there’s a name for that indicator: ratio of divorce rate to marriage rate (data from here). Unfortunately, there’s no publicly available divorce rate from Argentina, but otherwise we have the below.

Hot Fling Score Vs Divorce-to-Marriage Rate Ratio

Hot Fling Score vs. Divorce-to-Marriage Rate Ratio.  R2 is 0.61, P=0.0368, T=2.83
Hot Fling Score vs. Divorce-to-Marriage Rate Ratio. R2 is 0.61, P=0.0368, T=2.83

Hmm…Greece here is an outlier. Why? There could be a lot of reasons, maybe to be covered in another blog post down the line. That said, looking at our P-test value we do see some linear correlation. If we had a good reason to take Greece out of the picture, like some unique cultural feature invisible in this level of data (e.g.  Orthodox Christianity), we would have the following:

Hot Fling Score Vs Divorce-to-Marriage Rates Ratio (no Greece)

Hot Fling Score vs. Divorce-to-Marriage Rate Ratio with Greece removed.  R2 is 0.99, P<0.00009, T=20.52
Hot Fling Score vs. Divorce-to-Marriage Rate Ratio with Greece removed. R2 is 0.99, P<0.00009, T=20.52

So it turns out my guess was wrong – completely wrong but in the best way possible. More national data would be great, but we simply had no idea that there’d be this strong of a linear correlation.

Even with Greece, while romance score is a predictor for long-term success, the more romantic the nation the more divorces there will be. More romance and more emotional exchanges may allow for more room for cracks in the relationship to grow and romantic opportunities beyond the relationship to open up. This even applies in Catholic countries!

We only have a few countries of statistically significant size, but hopefully we can redo this study later to see how this indicator may change. Even with Greece, our P-Test value still indicates correlation.


 Odds & Ends

There’s a sample bias going on across all countries that’s pretty uniform: they were interested enough to take a personality test on Nanaya. This crowd is a bit younger and more urban than the whole national population. Maybe a later post will discuss how we can statistically control Nanaya studies.

Don’t like what you see? Want to see your country included? Take the personality test and make data big. The more people take the personality test, the sooner Nanaya will be available.

If you want to complain, send invectives and curses to info@nanaya.co, we’ll still love you. Through the power of the internet, that very same email also works for positive feedback and questions.

Hot Fling, Stable Relationship, and Net Romantic Scores are synthetic metrics and aren’t a part of a Nanaya algorithm.

Data was pulled from the Nanaya dataset as of January 31, 2015 at around 15,200 users. It’s grown quite a bit since.

Bubble charts generously plotted by Jackie Wisniewski. Histograms & analysis were done in Mathematica.

“Should I Dump My Girlfriend? Will I Find Another?”

Well let’s not get too hasty.

Earlier this week, I posted the Nanaya white paper to ArXiv: “Should I Dump My Girlfriend? Will I Find Another?” Or: An Algorithm for the Forecasting of Romantic Options. The title is a bit melodramatic but it’s a reference to Peter Backus’s “Why I Don’t Have a Girlfriend.” Apologies to everyone, especially Dr. Backus.

So this paper summarizes how the Nanaya algorithm works. I genuinely wonder who will have actually read it. Theory papers are hard enough to get through on their own and then harder yet when it’s an interdisciplinary mush. So we’ll go over some of the key points here.

The first major issue is uniqueness. Did we do anything academically worthy? Well, a lot of people have worked on the Secretary Problem (to be covered in a future post), but no one has really tailored it for a situation like romance. There are simplifications that do little justice to the complexity and information available to people – but those aren’t useful. Let’s just call it overzealous journalism…

We also do something new: comparing a specific relationship to being single and the chance of any other relationship. This is options analysis and rather new to the problem.

The Nanaya process.
The Nanaya process.

The rest is explaining what’s going on in the algorithm. Let’s check out the first figure:

So let’s break it down by each step:

Assessment of Inputs

When the Nanaya Beta is up (pending funding, be sure to call your local, friendly Venture Capital fund and request them to fund Nanaya) we’ll be asking a bunch of questions to understand your life circumstances:

  • What is your personality and social and romantic history. This gives us an idea of how likely you’ll be successful in finding a match, attracting a match, and how happy you are when you don’t have a partner in your life. We also need to know how long you’ve been in the city and job you’re in as that has a huge impact on how many people you’ll be meeting. You can check out this blog post that goes over this detail.
  • What is the ideal This is a mix of personality and specific values that are shared.
  • If you’re with someone, how do you feel about them.
  • The groups you interact with, like at work, geographically, and in socializing.

Determining Match Probabilities

With all of that data we can use our database and others available to us to figure out what the chance is of finding a match in any given encounter, depending on the group.

These numbers are typically low. Something like 1-in-10,000 is not unreasonable. This is where statistics comes in to play and you have something like the birthday paradox to offset those low odds.

Sociological Modeling

This is where we mix the match probabilities with your social behavior to figure out what are the chances of actually meeting someone. This is a bit of sociology and a bit of statistics to solve a problem no one’s really touched before, at least in this context (if I’m wrong, let us know). The sociology is based on personality, group types, and the results of some of our prototype experiments. The probability is based on binomial distributions and a variation of the Urn Problem. This is how we describe our social interactions…with urns!

The Nanaya implementation of the Urn problem
The Nanaya implementation of the Urn problem

We admit we make some simplifications in the paper, namely use of binomial probability distributions and assumptions that populations are very large. For almost everyone, these are pretty good assumptions but not for people in small populations. This gets into a lot of theoretical combinatorics which is well understood but not yet implemented.

Utility Function Valuation

This is the ugly philosophically ugly part where we literally put values to things that ordinarily don’t have numbers associated with them: like compatibility and happiness.

Using personality test results for who you are, who you want, and who your partner is we can estimate how happy you’ll be in any given relationship. We can also use your personality and romantic history to gauge how happy you’ll be in time as you’re single.

To simulate a random match you meet through the course of your life, we run a Monte Carlo simulation based on the groups you interact with. Some groups will have their own types of traits that lead to specific archetypes or segments (in demographic-speak) – we want to respect what’s observed in reality so we use these archetypes in seeding the randomness of the Monte Carlo.

Reporting the Results

With all of the above done, all that’s left is showing the numbers in a way that’s easy to digest!

If you sign up and take our personality tests, we’ll be inviting you to join our Beta program when it’s ready! So be sure to take your personality test and start today!

 

Does Online Dating Actually Boost Your Odds?

This is a bit of a no-brainer: online dating increases your exposure to more people than you ordinarily would run into each day. Sometimes that’s not so good:

But let’s face it, it’s not always easy to meet people you want to meet and the one might be out there. There’s pretty minimal investment in online dating, for better or worse.  If you use a site like eHarmony or OkCupid, they do a bit of work in figuring out your personality and finding matches that you’d likely be compatible with. That makes things even easier.

But is it worth that minimal investment? The costs of premium services, frustration of having your messages ignored, objectifying remarks, and simply the wherewithal needed to maintain yet another online account?

If you have thick enough skin, yes. It’s all in the math.

We can use a few portions of the Nanaya’s algorithm to show the impact. At a high level, there are two major factors that govern whether or not you will run into a match: how you interact within groups you belong and the chance that, upon any given encounter within a group, that a person will be your type.

I’ve used values from a volunteer who lives a rather normal, post-college life in the same city she was raised. She isn’t too picky but because there isn’t too much variation in social habits, the odds of someone new popping up aren’t distinctly high. In fact, the ideal environment for meeting compatible matches, work, is also one of the worst because of low turnover rate. She isn’t very picky but just enough that people she’ll run into out on the town will probably not be her type but a group of close friends has a different distribution of personalities than the city itself. Distant friends, family, and strangers met traveling are the smallest contribution.

Let’s look at her odds without online dating:

Volunteer's odds of finding a match without online dating broken down by group.
Volunteer’s odds of finding a match without online dating broken down by group.

The relative probability is shown for each group normalized by the “all groups” probability for that given point in time. You can compare this to the “stacked line” plot on the main page of Nanaya.

So what happens if she starts to go on a few online dates from people online starting in a year at a frequency of about one date every two months?

Volunteer's odds of finding a match with and without online dating.
Volunteer’s odds of finding a match with and without online dating.

We can see the group breakdown for the real effect of online dating. It’s pretty dramatic, especially considering how many more people she meets elsewhere!

Volunteer's odds of finding a match with online dating broken down by group.
Volunteer’s odds of finding a match with online dating broken down by group.

Some people might notice that it’s seems incorrect to separate potential matches from the local area from other, similarly local places (i.e. work and online). There’s undoubtedly validity to that argument but biased sampling, i.e. real human interactions, indicate otherwise. There are different personalities demographic, especially considering the size of the local population and the way data is aggregated. We also assume that there exists a sufficient sample of compatible men online for the volunteer to date, but most metropolitan areas are well populated when it comes to online dating and most sites do a good job of filtering. It’s that filtering that really helps the “signal to noise” problem in finding compatible partners.

Fortunately for my volunteer dating, let alone online dating, isn’t an issue. She’s actually happy in a relationship! After running the prototype, Nanaya predicted she should stick with it. A decision she had already confidently made. Good news for the volunteer and good news for continuing validation!

But the impact of online dating isn’t exclusive to that volunteer. All validation studies have shown it to be a rather significant part of improving romantic outcomes, ceteris paribus, anyway.

If you are interested in increasing your odds in finding a match, I recommend having an online dating account. Even better if the site is well trafficked and has personality testing. Log in now and again to search for new people and check on your inbox, but don’t take it too seriously. Why? That gets into ceteris paribus and that’s the subject of a later post.

All’s Nerdy in Love and War

Time to take a break from writing about relationships and psychology and write about something close to home: engineering. Specifically, this post connects fighting wars to forecasting love though an engineering history lesson.

If there’s something humans are born to do it’s stay alive. While almost all creatures have the biological urge to “fight or flight” in dangerous situations, humans are pretty unique in our ability to plan to survive. Planning is hard though. It’s instinctive to “run away from the angry wild animal we just poked with a stick,” but as soon as you add more wild animals or people the complexity increases along with the ability of a group to follow through with the plan. Planning also takes knowledge, but knowledge one day might be obsolete the next. People can spend years planning for business, war, and love only to have them fall through at the onset. To make a good plan takes knowledge of all the things involved in the plan, the system, and flexibility to account for all possible outcomes.

Arguably, people weren’t very good at planning till the 20th century. By about World War II, technology finally caught up such that we could get information from distant events in real time. The scale and interconnectedness of the war forced governments to accept how mind-bogglingly complex planning is. The destructive power of modern weapons and speed of battle in World War II also gave a bit of urgency in formulating good planning.

There’s nothing like self-preservation as inspiration. What resulted was a new discipline, systems engineering.

From a system engineer’s perspective war isn’t just a series of battles. The system engineer zooms out to see a web of interconnected parts: birth rates, the time to heal an injured soldier, grain harvest yields, tank production, and even innumerable things like morale. An example in war planning was the German Tank Problem where statistics helped Allied war planners determine the need for tanks on the ground by estimating the number of German tanks.

After World War II, things got even deeper. The destructive power of bombers and tanks gave way to nuclear weapons that can level nations in less than an hour. To prepare for nuclear war required rapid decision making based on limited information. Thus, the fields of game theory and decision analysis emerged.

As computers were integrated into war, it became vital to understand how information can be gathered and communicated consistently and meaningfully. An effort was undertaken by America’s Department of Defense to create such a framework. In the 1990s the Department of Defense Architecture Framework (DoDAF) was created to help understand, assess, and communicate the complex systems in war. DoDAF is a language to communicate how complicated things work but it’s also a framework of thinking about them.

An example of a DoDAF "Operations View."
An example of a DoDAF “Operations View.”

NASA isn’t a part of the DoD but space travel can be just complicated as military operations. Years ago, I used DoDAF to understand how to plan and cost the construction of Moon and Mars bases over the course of decades. Unfortunately, it’s not as easy as putting things on a rocket and launching it. Finding the cheapest way of doing it takes a bit of effort and worrying about the details, such as how much cheaper a certain module is to build after building the first one and the weight of food.

Snapshot from the Exploration Architectures Operations Cost Model. This brings back memories.
Snapshot from the Exploration Architectures Operations Cost Model. This brings back memories.

But women are not from Venus nor are men from Mars. Assessing the dynamics of dating and romance actually takes some rocket science. The Nanaya algorithm was designed much in the same way that decisions on the battlefield are made.

We can all do simple calculations to find out the chances of finding a partner much like I discussed previously, but that’s a single, meaningless probability that isn’t descriptive of the process involved in finding that person. If we want something meaningful, we need to address the entire process.

Specifically, with the help of DoDAF we can build something like Nanaya. Specifically, building a systems model, such as Nanaya, requires the following:

  • Defining the “building blocks” of a system
  • Determining how the blocks are connected
  • Understanding the values and qualities of each block and connection
  • Generating the rules of how all these parts change in time
  • Assessing the errors in knowledge and estimation

Once we follow these steps, we can get the equations to determine what actually happens in the system. With these equations, we can develop an algorithm for decision analysis.

Overview of Nanaya "systems."

Overview of Nanaya “systems.”

Strange but true: the very tool developed to help fight wars was used to make Nanaya, a service to help people make sense of their love lives.

But it makes sense if you ask me. If the first thing people worry about is self-preservation, the next is intimacy and love. If we’ve developed the tools for self-preservation, shouldn’t we develop tools for love? It’s time to make modern love just as nerdy as modern war.