Why relationship advice contradicts itself

A reader attentive to available content on relationships regularly encounters opposite advice, stated with equal confidence. You should tell each other everything, you should know when to keep quiet. You should preserve your independence, you should build together.

The first reflex is to conclude that the field is worthless. That conclusion is excessive: these contradictions have identifiable causes, and some are legitimate.

A research result describes an average tendency in a sample, with often considerable dispersion. A piece of advice, for its part, addresses one person. The move from one to the other is never direct, and it is almost always made without saying so.

A recommendation that is true on average can be false for a significant share of readers, without the underlying result being contestable. The apparent contradiction then comes from an improper move from the general to the particular.

From 2011 onwards, social psychology went through a serious crisis: many famous results, obtained on small samples, were not recovered in replication attempts conducted with pre-registered protocols.

The field of relationships did not escape it. Part of the claims in circulation comes from single, spectacular and never-confirmed experiments, while others rest on hundreds of replications. These two categories are cited in the same way, which explains a good part of the disorder.

A third mechanism has to do with the chain of transmission. A nuanced result becomes a chapter, the chapter becomes a title, the title becomes a rule. At each step, the conditions of validity disappear because they harm clarity.

This is not necessarily dishonesty. A book meant to be useful must come down on one side, and coming down on one side means discarding the qualifications that made the result accurate.

Faced with a claim about relationships, four questions allow it to be situated. Where does it come from, and from what work. Has it been recovered on an independent sample. Does it bear on an average or on a case. And does it specify the conditions in which it does not apply.

A claim that withstands all four is rare, and it is generally much more modest than those in circulation. That is the price of accuracy in a field where one would like rules.