Advice for grandparents is almost always a list of conduct: things to stop saying, habits to drop, ways to avoid becoming the relative everyone manages around. It assumes the relationship is built out of behaviour, and that the behaviour is the variable with the most room in it.

Two bodies of research have measured grandparent involvement directly. Neither found conduct to be the largest thing separating one grandparent from another.

The order nobody predicts

Harald Euler and Barbara Weitzel published a study in Human Nature in 1996 asking German adults to rate how much care they had received from each of their four grandparents. They collected ratings from 1,857 people and analysed 603 complete cases.

The order came out the same way each time. Maternal grandmother first, then maternal grandfather, then paternal grandmother, then paternal grandfather.

The interesting part is the middle. A grandfather on the mother’s side was rated as more caring than a grandmother on the father’s side, which is not what anyone predicts when asked. The line that mattered ran through the daughter, not through the gender of the grandparent.

The explanation the authors offer is about certainty of relatedness. A maternal grandmother can be sure of her connection to the child at both steps. A paternal grandfather cannot be sure at either. The ranking follows that logic exactly.

It is worth saying that this explanation is not the only one that fits. Families in many places live closer to the mother’s side, mothers and daughters tend to talk more often, and access alone could produce the same ordering without any evolutionary reasoning attached. The data show the pattern. They do not settle why it exists.

What the parish records show

The second study goes considerably further back and is one of the reasons Finnish record-keeping is prized by demographers.

Mirkka Lahdenperä, Virpi Lummaa, Samuli Helle, Marc Tremblay and Andrew Russell published their analysis in Nature in 2004, using pre-industrial church registers from Finland covering 537 women and comparable Canadian records covering more than 2,300.

Lutheran parishes in Finland recorded births, marriages and deaths with enough consistency that it is possible to reconstruct whole families across generations. That is what makes a question like this answerable at all.

The result is specific. Grandchildren were about 12 per cent more likely to survive to adulthood when a maternal grandmother was alive and under 60 at their birth. When she was over 60, the advantage fell to about 3 per cent.

There was no detectable effect between birth and age two, which points away from the obvious explanation of direct infant care and towards something operating later in childhood.

The other figure is the one that startles. For every ten years a woman lived past 50, she gained roughly two additional grandchildren, in both the Finnish and the Canadian populations.

What the two findings do to the question

Read together, they describe a pattern in which the largest measured differences between grandparents are not behavioural at all. They are positional: which side of the family you are on, how old you are, how near you live, whether you are alive when it counts.

None of that is available to be corrected by dropping a habit.

That does not make conduct irrelevant, and we would not argue it does. It means a list of behaviours is working on a smaller variable than its framing implies, and that a grandparent who feels less central than another may be reading a structural fact as a personal failing.

There is a practical reading available, though it is duller than the usual one. If proximity and contact are where the measured differences sit, then the questions worth asking are about logistics and access rather than about manner. Who lives near whom, who is invited, how often, and through which of your own children the relationship runs.

What neither study measured

Respect, for a start. That is the word in the original question and it appears in neither literature.

Euler and Weitzel measured recalled care, reported by adults about their own childhoods, in one country in the mid-1990s. Recollection of childhood is a soft instrument, and the grandchildren doing the rating were describing what they received rather than what they felt about the person.

Lahdenperä and colleagues measured survival and grandchild numbers in eighteenth and nineteenth century farming populations. That is a long way from a family in Copenhagen or Stockholm deciding whether to visit at Easter. The study also looked at maternal grandmothers only, because the authors judged that a post-reproductive lifespan cannot be defined as cleanly in men, so it says nothing at all about grandfathers.

Both are also correlational. Nobody assigned a child a living grandmother.

And the pre-industrial setting matters more than it might seem. In a society without antibiotics, where an extra adult in the household changed the odds of a child reaching adulthood, grandmothers were doing work that public health now does. Whatever the modern equivalent of that contribution is, it is not the same thing, and this study cannot tell you what it is worth.

What survives is narrower and, we think, more useful than a list. The differences between grandparents that have actually been measured are largely about position rather than performance. A grandparent worried about being liked less than the other set may be looking at a fact about family structure, and no amount of saying the right thing at Christmas will move it.