Background

I've been meaning to test the EQ-SQ theory for a while, but it got put on pause while I'm asking Simon Baron-Cohen to comment on my test idea.

However, I've noticed that to many people, the point of EQ-SQ theory seems to be that it provides a theory for how sex differences in technical occupational interests could arise, and I suspect people mostly wouldn't care about a debunking of the specific claims of EQ-SQ theory but instead more about whether men are inherently attracted to computers due to something vaguely along the lines of the rationality or objectivity associated with them.

So that made me wonder if I could find out where occupational interests actually come from, and in turn make progress on understanding the answer to this question. And that made me think about my own path into tech. I could come with multiple models, but for now, let's focus on one of them: ideology.

I think I've always had a strong appreciation for invention as a driver of progress, and have felt that computers are one of the most general ways of inventing things, with the potential to become artificial general intelligence which could solve many problems (though I have become more negative in my attitudes towards AGI over time). This attitude - an optimism about technological progress - could be interpreted as an ideology that I will call technophilia.

It seems intuitively plausible that technophilia might get people into tech. Both logically, and in my own case, and in certain important local community cases (MIRI opposes capabilities research due to fear about AGI, but on the other hand some people credit Yudkowsky's singularitarianism to motivate them to working on AI).

Furthermore, there might be some other indicators of this. There's a stereotype of shady finance people explaining why finance is the most important thing while ruining everything, which would seem to reflect a disagreement between finance people and the general public about the importance of finance, and therefore probably also a correlation between finance interest and capitalist ideology.

So between technophilia and capitalism, it seems like we already have some good potential examples. Of course, correlation != causation, so I don't think we can conclude anything just from showing such correlations. But the evidence is suggestive enough that it motivates me to get started on this.

Before we can even quantify the correlation or start robustly investigating the causation, I believe we need a strong measure of the phenomena under investigation. I'm currently reasonably satisfied with the measures of vocational interests that are available, but I would like to have a proper measure of vocational ideology.

To achieve this, I have recruited a sample of around 200 people on Prolific and asked them about 20 vocational interests. For each of the vocational interests, I have taken those who score high and low on the scale, and invited 3 high scorers and 3 low scorers to describe their opinion on the field, using a question like the following:

Opinions on carpentry*

What are your opinions about the role of carpentry? This could be things such as:

* How important/helpful/harmful is carpentry to society?

* What is the future of carpentry?

* Does society have too much or too little respect for carpenters?

* What is the character/virtues/vices of carpenters?

* Do you have any personal experiences or relationships which deeply affect how you view carpentry, and if so, which?

You don't have to address these specific questions; you can also give you own opinions on carpentry, separate from these ones.

(Minimum 300 characters)

I am getting around 100 responses to different jobs, and based on those responses, I identify statements that might be ideologically relevant, which I intend to submit to factor analysis.

Market

I grabbed a random measure of marketing interests that I saw while scrolling through twitter one time:

• Persuade customers to try a new product

• Increase sales for a company during a promotion week

• Sell services to a target group of people

(This had a Cronbach alpha of 0.8 in my data, but that might not matter because if it seems justified I will adjust for attenuation.)

Based on people's responses, I have the following items assessing marketing ideology:

• Advertisements can be useful for alerting and informing customers about products of interest.

• Marketers often have a morally shady character, and might use manipulation or psychological tricks.

• Some shady companies use targeted advertising to search for people who are financially desperate and offer expensive loans.

• Politicians use targeted advertising to trick passionate people into supporting them.

• Plastic surgeons use targeted advertising to trick girls with poor body image into harmful treatments.

• Some online advertising is a harmless way to fund free access to websites.

I might adjust the items, e.g. adding more items as I get more responses to the survey, but you can think of these as indicative of what I mean by asking people about marketing ideology. "Ideology" is meant to cover abstract value-laden questions about bigger societal relations, as well as views about the nature and virtues and vices of things.

The simplest way to score them might be to create a sum score of positive vs negative opinions, but I might do something more elaborate, e.g. factor-analysing or adjusting for attenuation, if it seems appropriate. If the marketing factor seems part of some other more general factor that includes opinions about other jobs than marketing too, then I might also include items from those other jobs.

Resolution criteria

This market resolves to YES if I get r > 0.4 (or R>0.4) with an IMO appropriately adjusted measure of marketing (or other ideology).

This market resolves to NO if I get r < 0.4.

If there is no meaningful ideological variation in these questions, then the market still resolves to NO unless I somehow get meaningless ideological variation to go r > 0.4.

If something stupid happens along the lines of my measure being too biased around progressive ideology rather than vocational ideology to capture a proper r > 0.4, then I guess the question resolves NO, rather than me doing some convoluted followup thing.

If I can't decide what types of statistics I feel are best for the analysis, and this leave the answer ambiguous, I'll probably resolve PROB 50% (or maybe some skewed thing, e.g. if 3 out of 4 methods lead to r > 0.4, then I will may resolve PROB 75%, but it depends on whether the 3 methods are "too similar" to each other).

TO EMPHASIZE: A YES RESOLUTION TO THIS MARKET IS NOT STRONG EVIDENCE THAT INTERESTS ARE UPSTREAM OF IDEOLOGY. Anyone saying so is probably lying or bad at inference. Rather, a NO resolution is IMO strong evidence against interests being upstream of ideology.