Statistics in adverts

Thread Starter

andrewmm

Joined Feb 25, 2011
1,461
Since I first noticed the small print in TV adverts , that state how big the sample was to get the percentage agree

I am shocked at how few one needs.

Any one else noticed this ,is this a new thing ?

One yesterday was some thing like 70 something percent of 24 people agree !!!

that to me is as near to just sample and sooner or later you will get the results you want,
I'm surprised how bad the sample size is

I started this in general as it was a "rant",
but I'm wondering if it should be in maths ?

Any thought on how to define what a good sample size might be
I'm thinking of raising a complaint to the advertising standards authority that the results are not statistically significant

and they should be to be advertised.
 

wayneh

Joined Sep 9, 2010
17,152
The number of samples you need to show significance depends on the size of the difference versus the variance you observe between samples.

Suppose you want an estimate of the height difference between men and women. Picture two bell curves just slightly offset. Here the variance and the difference you're looking for are roughly equal at 6". If you want to reject the null hypothesis (that there is no difference) because zero does not fall within the 95% confidence interval, you need something like 30 of each population. (That's from memory. I'm too lazy to recalculate it.) With enough measurements, perhaps thousands, you can refine the estimate and its 95% confidence interval to, for example, 6" ± 1".
 
Last edited:

Papabravo

Joined Feb 24, 2006
16,091
Assuming you've been surrounded by advertising "all your life" (dawn of television), ain't you kinda old to have any faith in it at all?
 

402DF855

Joined Feb 9, 2013
271
Personally, I don't think we should try to protect against stupidity, or more nicely, ignorance. If you don't think 70 is enough, you should still let others determine their own threshold. Protecting people from having to evaluate statements seems to be in vogue today. These efforts are doomed to failure IMO.

I saw the commercial so many times through the years, 4 out of 5 dentists recommend "gum-X" for their patients that chew gum. I've always been more interested in the 5th dentist, but that's just me.
 

wayneh

Joined Sep 9, 2010
17,152
Assuming you've been surrounded by advertising "all your life" (dawn of television), ain't you kinda old to have any faith in it at all?
Exactly. The only real enforcement against bogus claims comes from competitors, followed by angry customers. As long as you don't directly damage competitors, you're pretty safe. Customers only rarely catch fraudulent claims, like when a "foot long" is only 11".

Frankly, getting 17 out of 24 people to agree sounds like a compelling argument. Was it rigged in some way? Of course it was! The details you don't have are the important ones.
 

SamR

Joined Mar 19, 2019
3,400
Shopping on Amazon, there is always a "rating" ie popularity contest. Sounds like a good idea until you start reading the comment section. Then, the most common reaction I have is "WTF, who are these idiots".
 

jpanhalt

Joined Jan 18, 2008
11,088
Shopping on Amazon, there is always a "rating" ie popularity contest. Sounds like a good idea until you start reading the comment section. Then, the most common reaction I have is "WTF, who are these idiots".
What makes me smile in amazement is how the "average" is calculated. It's like Lake Wobegon, MN where the women are beautiful and all the kids are above average (https://en.wikipedia.org/wiki/Lake_Wobegon).
 

Thread Starter

andrewmm

Joined Feb 25, 2011
1,461
Assuming you've been surrounded by advertising "all your life" (dawn of television), ain't you kinda old to have any faith in it at all?
Thank you @Papabravo

not certain what ./ who your referring to as has faith ?

personally I always take with pinch of salt these numbers but since the TV companies have been required by law to put the sample size on screen as well as the percentage, I have become interested at just how bad some of these things look.;
 

Papabravo

Joined Feb 24, 2006
16,091
Thank you @Papabravo

not certain what ./ who your referring to as has faith ?

personally I always take with pinch of salt these numbers but since the TV companies have been required by law to put the sample size on screen as well as the percentage, I have become interested at just how bad some of these things look.;
I was referring to people who believe anything and everything they see on Television. It was after all one of the original social media platforms.
 

Thread Starter

andrewmm

Joined Feb 25, 2011
1,461
This thread seems to be a complete rehash of this: https://forum.allaboutcircuits.com/threads/65-of-woman-agree.175351/

Is the TS really that bored or does he lack a grasp of the obvious?
Thank you @jpanhalt

As I understood it , the off topic forum was the one to have a good chat about anything.

The thread you reference was the one I referred to in the first post,

That was a technical discussion, Or I tried to keep it such,
with a specific outcome, on statistics and it was very interesting ,

This is in off topic,

and a general comment on just how the advertising industry is getting
wondering if anyone else had had similar thoughts

As I said I'm narked enough that I think its distasteful, and wonder if anyone else thinks it might be under hand or something, and any thoughts on what makes a good / bad comparison ?
 

SamR

Joined Mar 19, 2019
3,400
people who believe anything and everything they see on Television
I quit believing as a very young child after a commercial for individual boxes of cereal where when it was finished the young lady model wadded it up in her hand and "It disappears like magic" and opened her empty hand. No matter how hard I squeezed I couldn't make that dang thing disappear like they did on TV! Never be fooled again...
 

402DF855

Joined Feb 9, 2013
271
My reaction is that of course advertising is full of half truths and is meant to mislead in many cases. But this applies in all human endeavors. News, politics, science, your personal lives, the auto mechanic. Much of what you are told is, to some degree, wrong.

Recalling the theories of subliminal advertising from decades ago, I've theorized that for some reason advertising executives have determined that a really effective strategy, especially for car insurance, is to annoy the heck out of the audience. This is true of both of the big names, i.e. progressive and geico. It must work for them, they keep using it.
 

jpanhalt

Joined Jan 18, 2008
11,088
In statistics, a larger sample size results in more reliable averages.
The Ns justify the means:p
1) Sample size only improves the mean or median if it is unbiased. Do you actually think star ratings are unbiased? Check out Alibaba's star ratings.
2) Ordinal data and opinions are not parametric. While a mean is often used, as it robust, something better, such as median would be more appropriate.
 

wayneh

Joined Sep 9, 2010
17,152
This is true of both of the big names, i.e. progressive and geico. It must work for them, they keep using it.
And Allstate, and Liberty, and Farmers, et al. They all advertise because it works. Sadly that's true of drug advertising also. Advertising has a huge impact on prescriptions. Both industries endeavor to sell you on something that makes big profit, and you don't really want or need.
 
Top