would have to be explained by items 4 to 7, which have relatively low weights!
Item 4 is the killer. It is actually heavily weighted in the paper you link, though not in all versions of the SOI. You can score a 50 without ever having had sex or ever expecting to have sex, and up to 32 of those points can come from a response to 4. 4 is obviously not zero sum and need not average out.
There is very good reason to believe that men fantasize about sex more often than women (who buys most pornography?), and there is also reason to believe men are more likely to fantasize about persons other than their current partner. It's also possible that "once per day" is actually relatively infrequent given how often men think about sex. Put these together and it can explain a lot of the variance. The paper does not show breakdown by answers, at least not that I saw.
It's also worth noting they had some significant response-rate issues (RR < .5 for some demographics). If women who like casual sex, are promiscuous, or are prostitutes are less likely to respond than all other groups, that could explain the variance.
Also, there are other versions of the ROI that don't need to average out, as they lump large numbers together (i.e. "More than 8").
There is very good reason to believe that men fantasize about sex more often than women (who buys most pornography?)
I don't see that as obvious. The obvious counter-argument is "who buys most romance novels?", which are basically a different medium for porn.
I would love to see breakdown by question, if anyone can find a paper with that.
I just did some reading about "Sociosexual Orientation Inventory", a simple 7-item test designed to measure one's openness to sex without love and long term commitment.
Here are the questions. How long will it take you to spot the huge problem ahead...
Score is: 5 x item1 + 1 x item2 (capped at 30) + 5 x item3 + 4 x item4 + 2 x (mean of item5, item6, and reversed item7)
Do you see the problem already?
Quite predictably, researchers report that men have much higher SOI scores than women in all countries. But the first three questions (ignoring non-1:1 gender ratios, differently biased sampling for different genders, different rates of homosexuality between genders, different behaviour of homosexuals of different genders, 30 partners cap on the second item, differently biased forecasts of the second item and other small details that won't affect the score much) - simply have to be identical for men and women, so the entire difference would have to be explained by items 4 to 7, which have relatively low weights!
The differences between men and women can be really extreme for some countries, Ukraine has 50.79±28.92 (mean±sd) for men, and 17.36±8.65 for women, which means that either Ukrainian men, or Ukrainian women, or both, are notoriously lying when asked about past and future sex partners. In most countries the differences are more moderate, with total 48-country sample's scores being 46.67±29.68 for men, and 27.34±19.55 for women. Latvia leads the way with smallest difference, and so most likely greatest honesty, with 49.42±23.61 for men, and 41.68±26.68 for women, what can be plausibly explained by just differences in attitude. (fake lie detector experiments have shown it's almost exclusively women who are lying when answering questions like that)
SOI seems to be considered quite useful by psychologists, it correlates with many nice things, not only other questionnaires, but country SOI averages correlate with various demographic, economic, and health scores in quite systematic way. Still, I cannot read papers about it without asking myself - why didn't they bother to perform this basic sanity check - which would detect huge number of outright lies in answers. And more importantly - what proportion of "serious science" suffers from problems like that?
References: The 48 country SOI study, fake lie detectors shows which gender lies more.