Go woke, go broke

[SpuriusMaximus]

Not here.

Ah, ok. Fair enough. I don't know what ads they show in the UK.

 

[jt512]

Ah, I found and read the actual study: https://www.unstereotypealliance.or...INCLUSIVE ADVERTISING (Business case) WEB.pdf

It's about less than people seem to think. That inclusivity is about women. Their calculated index is the "GENDER UNSTEREOTYPE METRIC (GUM)." (And it's only in caps because I copy and pasted that from a title.) They don't even calculate their index for any other categories than men and women. I guess the UN Women part should have been a clue.

It's stuff like Mars discovering that they're selling more chocolate if they include more women in the ads....

You misunderstood how they measured the supposed diversity of the ads. You are correct that their metric for diversity was the so-called "gender unstereotype metric" (GUM). But the GUM is merely the following single question (asked separately about male and female characters):

How much do you agree or disagree with the following statement: This advertising presents a positive image of the (male/female) character/s that sets a good example for others: strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree, or strongly disagree?​

So, the GUM does not directly measure diversity at all. The study authors claim it indirectly does because they claim to have data showing that it correlates with measures of diversity "especially for females" (e.g., diversity of skin color, portrayal of women over 40, etc.), although they do not indicate how strong these correlation are.

For that reason and others, the study does not support the conclusions it purports to. Maybe more on this tomorrow...

 

[SpuriusMaximus]

So, the GUM does not directly measure diversity at all.

Pretty sure that's what I was saying too, a bit more explicitly further down in the message. And not just directly. It doesn't measure it, period.

And yes, it's just a study of GUM, nothing else.

As for the inclusivity and diversity stuff... really, it's not even about a calculated correlation. They just "deduced commonalities" in the top 25% and bottom 25% of the spectrum to "find key drivers." And it's not just that they don't show the correlation data with that diversity stuff, they don't even say what correlates with what. They just say that Kantar has 65 other metrics that they deduced commonalities with GUM, but really, what, all 65 to the same degree? I mean, they just said something about key drivers, which would suggest otherwise.

But thanks for bringing up that question, because only now it occurs to me that it doesn't actually measure what you'd normally think of as "unstereotype" either. Someone could think a character is a good example to others BECAUSE they fit a stereotype.

 

[jt512]

Pretty sure that's what I was saying too.

What you got wrong is that you said (1) that in the study "inclusivity is about women," implying (2) that the GUM is some sort of measure of the inclusivity of women in the ads. Both (1) and (2) are wrong.

And not just directly. It doesn't measure it, period.

Well, if A correlates with B, and B correlates with C, then A correlates with C and hence is an indirect measure of C, although possibly a poor one.

As for the inclusivity and diversity stuff... really, it's not even about a calculated correlation. They just "deduced commonalities" in the top 25% and bottom 25% of the spectrum.

The wording of that paragraph of the report appears (to me, as a statistician) to be saying with non-technical language that they performed a factor analysis of the diversity variables and then used regression to determine which factors best discriminated between the first and fourth quartiles of the GUM.

And it's not just that they don't show the correlation data with that diversity stuff, they don't even say what correlates with what.

They probably do know the correlations between the individual diversity variables and the GUM because they likely did the analysis that the wording of the report hints they did (or something similar that would yield those correlations as well). The fact that those data were not presented doesn't mean they aren't known. After all, hardly any data were presented.

 

[SpuriusMaximus]

What you got wrong is that you said that GUM was a measure of the degree of female representation in the ads. It is not.

Fair enough, I confess that I was a bit misled at that point. I should perhaps have been in less of a hurry to read through it.

Well, if A correlates with B, and B correlates with C, then A correlates with C and hence is an indirect measure of C, although possibly a poor one.

Yes, but at the very least it's not a guaranteed causation by C. Which is what is being pushed.

And in their case it's so trivial to debunk, that their own case studies do that. E.g., the Bayer fairy tale princesses have nothing to do with portrayal of women over 40. Or the Cadbury one also had nothing to do with women over 40, or most other inclusion criteria. So you obviously can skip some of those 65 criteria just fine.

Plus, as you said, just calculating GUM won't tell you anything about that. You could make an all white male ad for something and present them in a positive light, and it would score just as high on GUM as a mixed one. Like, you don't even need to think up a brand new ad. You could take the Cadbury ad and replace every single person on the bus with the whitest Caucasians, and it would probably score the same on GUM.

I can see a correlation in the fact that global corporations that have the sense to not insult their potential customers, also have figured out by now that they're selling to a lot of different people. You know, part of the "global" part in global corporation.

 

[SteveAitch]

I missed all those gay couples in adverts decades back, along with the Asian couples etc.

They're still there, just depends which TV channel you watch, and possibly hte time of day - I think companies are being more selective about which channels they advertise on. Demographics or something like that.

 

[Cat Not Included]

There's a difference between inclusivity in advertising and harping on systemic racism and injustice, even if they have to invent it. "Woke" is more like the latter.

Interesting. A few typical things I've seen described as "woke madness" include:

• Changing the labels on toy aisles to no longer be 'boys' and 'girls'.
• Using female models in clothing adds that are closer than previously to average American BMI (though still under average)
• Putting very minimal facial hair on a medieval warrior woman in a video game.
• Changing an animated candy to wearing flats instead of heels.

 

[SpuriusMaximus]

Well, as has almost everything else at this point. But then the term was coined in the '30s as "staying woke" (meaning being aware) of injustices towards blacks, and then popularized again in and after 2014 by BLM to mean the same thing. And then it broadened to mean pretty much any injustice an activist was claiming to fight on their cell phone, and felt like claiming is a systemic and institutional issue, from race to gender to trans rights to body positivity to whatever.

Nowadays it's more like used pejoratively, and about almost everything, because previously it WAS used to mean almost anything and everything one feels like fighting against by its proponents.

 

[Orphia Nay]

It also doesn't go unnoticed when someone just unilaterally declares victory and pats themselves on the back, without actually even trying to have a sound argument. Because see above, the actual study supported none of those claims.

The whole thread is a sound argument. Your side lost.

 

[SpuriusMaximus]

Whatever fantasy keeps you happy.

 

[Klimax]

The whole thread is a sound argument. Your side lost.

Sound argument? What sound argument???

 

[dirtywick]

the length of the list of companies that went woke and then broke kind of speaks for itself

 

[d4m10n]

the length of the list of companies that went woke and then broke kind of speaks for itself

I mean, they weren't exactly global brands.

 

[gnome]

the length of the list of companies that went woke and then broke kind of speaks for itself

Next to the length of the list of companies that go broke without any "wokeness" involved it seems less conclusive.

 

[SpuriusMaximus]

FWIW I don't even care much if they go broke, and I sincerely doubt that any corporation actually went woke as more than a PR stunt. Small private companies like mom and pop shops may go with their ideology, multi-billion dollar corporations go for whatever PR image or stunt they think (correctly or not) will make them more money.

I was just commenting on a dishonest "study".

If they actually wanted to measure diversity in advertising, they could do just that quite objectively. Just take a marketing campaign, and count how many groups it represents, put that in your correlation. If it supports the hypothesis, cool.

But that's not what they did. They used their "Gender Unstereotype Metric", which is actually ironically named, because it has nothing to do with stereotypes, and they didn't split it by gender. In fact, they even say they averaged male and female scores for each company. And it only measures whether those characters are positively or negatively portrayed, which would only measure being non-stereotypical if the only stereotypes about both genders were negative.

ALL the correlation they actually showed was basically: ad campaigns which put their customers in a more flattering light do better. Whop-de-do. We knew that for decades. Why do you think those tampon ads showed those women swimming, riding, etc, instead of, say, being a miserable wine mom? Or why do you think car companies stopped presenting women as just some accessories for a stud with a car? Why do you think beer ads show fit young people having a party, and not, say, some loner couch potato in his underwear drowning his sorrows in the bottle?

But OK, you want to use GUM to support diversity? Again, you can actually do that in a more direct and objective way. Like calculate that GUM by race, see how much that affects the results.

But nope. What they did was literally this handwave: well, companies who did X tended to also do Y, so Y must be the cause. Even though even even their own case studies pretty much debunk that causation.

Essentially what they did is like me arguing to the effect of:
- people who order more over the internet are lazy
- a lot of people who order over the internet have a pet, since they order pet food
- therefore having a pet makes you lazy

And then that article padded it with more of their own interpretation of "woke" that neither the study, nor its press release, actually claimed.

 

[SpuriusMaximus]

Also, let me introduce people to this funny concept: articles and studies written by PR. Unlike ads, these don't tell you "go buy X". They masquerade as serious articles telling you stuff like "the era of IT people going to work in jeans is over, companies prefer a professional suit now" and incidentally mention Men's Wearhouse. (Actual PR campaign back in the early 2000s, after the dotcom bubble had burst, and yes, paid for by Men's Wearhouse.) Or that some professor calculated the formula for procrastination... without defining the units or how you measure any of the stuff in the formula. And, oh, look, he has a consulting firm for hiring, so *nudge* *nudge* *wink* *wink* you know what I mean. (Yep, another actual case.) Or that whole small villages on the mountain only have at most a dozen schoolkids, means there won't be enough people to pay for your retirement unless you pay for a private pension, and oh, look, they actually mention an insurance company that offers such pensions. (Ditto.)

And the phenomenon is ubiquitous. In local newspapers half the articles can be written by PR companies. They're professionally written and they're free, so hey. But they get into major newspapers, TV and even gaming sites too. And yes, it includes more (non-peer reviewed) studies and papers than you'd think. They just go shopping around for some academic who'll put his or her signature on it, in exchange for adequate remuneration. Invariably they eventually find one.

And frankly this whole study reeks of a PR op funded by Kantar. And not even the subtle kind. Most of the data comes from them, it's their criteria for inclusion, and they're mentioned dozens of times in it. And oh, look, they offer seminars for inclusion and whatnot


Which brings me to: was the list of companies chosen randomly or cherry-picked by Kantar or how? The study doesn't even mention the word "random".

 

[jt512]

If they actually wanted to measure diversity in advertising, they could do just that quite objectively. Just take a marketing campaign, and count how many groups it represents, put that in your correlation. If it supports the hypothesis, cool.

But that's not what they did. They used their "Gender Unstereotype Metric", which is actually ironically named, because it has nothing to do with stereotypes, and they didn't split it by gender. In fact, they even say they averaged male and female scores for each company. And it only measures whether those characters are positively or negatively portrayed, which would only measure being non-stereotypical if the only stereotypes about both genders were negative.

Indeed for each ad in the study they had measures of 65 characteristics of the characters portrayed, including the "inclusivity" variables age, ethnicity, skin tone, body type [and] sexual orientation.... So, if they wanted to test the hypothesis that "inclusivity" in advertising boosts sales, why did they use GUM, a measure of positive portrayal that only correlates—to an undisclosed degree—with these direct measures of inclusivity, rather than do the obvious thing: use the direct measures of inclusivity themselves? We can't say for sure, but one obvious possibility is that the direct measures of inclusivity didn't show the positive results they hoped for, but GUM, a measure of positive portrayal, did. So they used GUM and argued (poorly) that it can be considered a surrogate measure of inclusivity.

 

[SpuriusMaximus]

Indeed for each ad in the study they had measures of 65 characteristics of the characters portrayed, including the "inclusivity" variables age, ethnicity, skin tone, body type [and] sexual orientation.... So, if they wanted to test the hypothesis that "inclusivity" in advertising boosts sales, why did they use GUM, a measure of positive portrayal that only correlates—to an undisclosed degree—with these direct measures of inclusivity, rather than do the obvious thing: use the direct measures of inclusivity themselves? We can't say for sure, but one obvious possibility is that the direct measures of inclusivity didn't show the positive results they hoped for, but GUM, a measure of positive portrayal, did. So they used GUM and argued (poorly) that it can be considered a surrogate measure of inclusivity.

Honestly, that was my suspicion too.

 

[SpuriusMaximus]

Just to add, there is actually evidence to the contrary: too much of some kinds of diversity in ads can lower the effectiveness, including from minorities. For example, mixed-race couples are perceived lower than same-race minority couples or (depending on who you trust) even white couples. In fact, they can even attract more negativity towards your brand. You have better results just showing a black couple or an Asian couple, than having the more diverse alternative of showing an interracial couple.

Source (among others):
https://develop.sfsu.edu/mixed-race-couples-still-controversial-many-sf-state-study-finds

I eagerly await someone telling me that the San Francisco State University and its students (who were the lab rats, so to speak) is some fascist trumpist organization who can't be trusted, because the results don't match what they want to hear

 

[novaphile]

I can remember an Australian advertising study, from waaaaay back in the day, that pointed out that black African males only appeared in Australian adverts as bar tenders, musicians or athletes.

For a while, I was noticing it all the time, but less so now.

I think this is no longer true here, but mainly because we now show so many adverts that are imported from the UK and the USA (with Australian accents dubbed over the vision).