Predicting spousal personality

---

Tomorrow morning I'm meeting Tabitha's husband for the first time. I've known Tabitha for a while: she's quiet but intense, organised and rather emotional. I'm wondering what her husband is going to be like. What should I expect? My thoughts veer to Clare and myself: if someone knew Clare very well, what could they predict about me? Let's analyse this...

Clare’s temperament sits close to the ISFP profile: introverted, attuned to sensory experience, slow to engage with strangers, low on conscientiousness, high on neuroticism. She avoids public conflict but reacts strongly to stress. If we mapped her scores onto the Five Factor Model, they would cluster in the low ranges of Extraversion and Conscientiousness, high in Neuroticism, with low-to-middling Agreeableness and Openness.

Now suppose personality traits are modelled as continuous variables, roughly Gaussian in distribution across the population, according to the Five-Factor Model. The mathematics of correlated normals tells us that if two people’s traits are correlated with coefficient r, then knowing one partner’s trait gives an expected value for the other’s:

E[Trait_Spouse | Trait_Person = x] = r · x

In plain terms: spousal similarity exists, but it is partial. The closer r is to one, the more a spouse resembles the other; the closer r is to zero, the less predictive knowledge there is of the other partner.

Empirical psychology finds modest positive spousal correlations (typically r = 0.2–0.3) for traits like education, politics, conscientiousness and religiosity; weaker and sometime negative for neuroticism; sometimes near zero for extraversion.

So what of Clare and myself?

Clare is introverted; I am more extraverted. Her conscientiousness is low; mine is high, expressed in structured projects and organised commitments. Her neuroticism is elevated; mine is much lower, I'm much more phlegmatic.

By the correlation formula, if Clare scores a high positive x on neuroticism, the best estimate for me would be r·x. With r close to zero, though, for neuroticism in spousal studies, my expected value is close to zero - consistent with the reality that her anxiety does not predict mine. 

For conscientiousness, however, spousal correlation is moderate; Clare’s low score predicts a somewhat below-average score in her spouse. Yet my actual conscientiousness is markedly higher than predicted, an example of how individual outcomes scatter around the statistical mean.

Thus the conditional-expectation model provides a sober prediction: spousal similarity is real but weak, easily swamped by variance. Knowing Clare tells you a little about me, but not much.

Back to Tabitha.

She is quiet but intense (low extraversion, high neuroticism), organised (high conscientiousness), and emotional (again neuroticism). The mathematics says: expect her husband’s scores to regress toward the population mean, weighted by modest spousal correlations.

So: he is unlikely to be as quiet as she is, perhaps somewhat more outward-facing; less neurotic than she is, but not phlegmatic; with conscientiousness somewhere near average.

If I had to guess: a calm, moderately sociable man, complementing her intensity by providing ballast rather than echoing it.

Prediction via correlation is a blunt instrument. It captures the statistical truth that spouses are neither random pairings nor mirror images. The mathematics of expectation, when mapped onto personalities, reminds us that knowing one partner yields only a dim - but not entirely unsurprising - sketch of the other.

Note: I met Tabitha's husband this morning. A very pleasant man, who in personality and life-history terms seemed very like a more concrete version of me...