Economic Predictions Using Web Usage Data

Makarand writes "The Chicago Tribune has an article on the claims of ComScore Networks Inc., that it can predict major economic trends by tracking the online activity of 1.5 Million people. The company gains access to people's Internet travelogues by giving them free security software and programs that speed up their connections. Economists say that the company's models need to be tested over several years before they can be considered accurate."

Spyware

[krnlpanic]

"The company gains access to people's Internet travelogues by giving them free security software and programs that speed up their connections."

I always thought that anyone who provides programs to "Speed up your internet connection" were crooked and could not be trusted. Spyware at its finest. As far as this company providing "security software"...I won't even go there.

Sound like spyware to me.

[giel]

Tracking the behaviour of 1.5 million people. And all these people are aware they are being tracked? And they did agree?
I can't believe it...

PS. Watch out! You computer has an IP address...

security company?

[twitter]

I suppose having your every move on the web tracked and monitored by some comercial company over windows is considered secure by some. Next time these folks want security and a faster connection software, might I recomend:

• Debian [debian.org]
  
• Red Hat [redhat.com]
  
• OpenBSD [openbsd.org]
  

This would make the world a better place, even if it could not be used to forcast the next great depresion.

It begins.

[torre]

The uses of novelty things to promote the real agenda of spyware or perhaps we should call it the new reincarnation of the Trojan horse. This is only but the first public release of what they've been doing, God knows what they've done without us knowing.

but that's just my 2 cents.

Income-bracket/age-range skewed?

[mccalli]

I would imagine this data is skewed towards the mid and high-end income brackets, and also away from the older age brackets.

Given that the equipment you need to access the web is still fairly pricy, and also that the majority of people accessing it are still relatively young, I'd question the ability of this model to extrapolate to the wider world.

Cheers,
Ian

representative group?

[John_Renne]

I don't think the group theý're tracking will be very representative. Allthough there is a large group of people is using the Internet there's no way global trends can be predicted by their behaviour. There is a world out there that's hardly connected (some call it Africa) that has some influence on our world's economy but is left out completely.

Maybe the researchers should see the world is bigger than the US

Re:Income-bracket/age-range skewed?

[actiondan]

I would imagine this data is skewed towards the mid and high-end income brackets, and also away from the older age brackets.

Also, the sample will only include people stupid enough to download their spyware and let it report back on everything they do online.

Discretionary income

[ivrcti]

I really think this is only looking at a small, unscientific sample of the population, and is therefore highly prone to skewing.

It has also been my observation that most spending on-line is from discretionary funds, so this tends to skew the results as well.

Finally, it does take into account the type of information being accepted by their target audience. Those who get their information primarily from internet sources, deal with a different set of information than those who rely primarily on TV/newspapers, and will therefore make different buying decisions.

What About Credit/Debit Card Co's Right Now

[jck2000]

Does anyone know what Amex, Discover, Visa, MasterCard, etc. are doing currently with the data they accumulate? I am not suggesting anything nefarious, rather I think these institutions, by having much larger (and probably more representative) sample sizes, would be able to accomplish much more than this smaller company would.

Re:Registration required Complimentry copy below

[mr_z_beeblebrox]

My own thought on this (my fiance is a biologist and we do a lot of 'those' stats, that is my only qualification) is that internet users are a subset of the people who make up the economy. As more research goes into this they will find that trends indicated by internet usage will have a relationship with actual trends of the 'whole picture' and will later develop the internet /society ratio as a valid indicator. However, until they have this backlog of data to compare it to it will not remain valid.

They could of course mine the net for data on past years and produce 'predictions' for those years to compare with actual data on those years and deduce the ratio from there.

Re:Income-bracket/age-range skewed?

[Reinout]

So all the smart people get filtered out, leaving only dumbasses. Well, the sites they visit often must belong to stupid companies. Tada, instant stupidity-rating! :-)

Reinout

Another way to manipulate the dog races...

[Zarf]

After reading the article and deciding to ignore the already made point that this is an invasion of privacy:

This kind of data if collected well could very well help in profiling online trends and giving subscribers to this data a "leg-up" on their competitors. That's True. BUT, I doubt that this data can be used for predicting meat-world trends. The only people you are dealing with are the ones who are both willing to buy online and are willing to allow spyware on their boxen. I'd guess that the fact that you're talking about a select group of unsophisticated users, who are yet sophisticated enough to research and/or purchase online, would mean that the data is self-censoring.

It's sort of like surveying people who hate telemarketers over the phone. You'll only talk to a very few people and likely have a useless data set. It would be like a survey on invasion of privacy issues only from people willing to report to the surveyor thier SSN.

Catch my drift?

The resultant data would influence an investor house to make an unwise decision and bet on the wrong dog in the grand dog-race called the Stock market. The data provider can dope up the right dogs to make itself some money. That's what I think of whenever I read about these "trend" predicting companies. That's just me though.

Re:Spyware at its finest

[tmark]

Why should it be illegal if the users have agreed to a valid disclaimer ?

Re:Income-bracket/age-range skewed?

[Anonymous Coward]

I think the above argument one is more relevent. You are getting a subsample of the population who's behavior may or may not mimic that of the general population. Further this data only has real value as long as you have a continuing time series availble. If technology/the sample charecteristics/sampling techniques change in the next 3 years the data doesn't have alot of value, as its predictive power has not been established with out of sample testing and the ability to continue to use a consistent dataset to continue to use it to forcast once a predictive model is built. And although I'm a PC user, do the have a MAC version? MAC users on average have highter incomes (FLAME ON!). This alone skews the sample from the general population. Is that enough to make its predictive power worthless? Probably not, but only time will tell.

Re:Economic Predictions Using Slashdot +5 Rated Po

[fishbowl]

>Please stop badmouthing Free Software as if it
>had anything to do with "Communism".

That became a lost cause when Stallman first used the term "Manifesto" to frame his agenda.

The best part of their theory

[T.E.D.]

...is that there isn't enough historical data on past internet usage to prove its all baloney yet.

By the time there is, they ought to be able to pump out quite a few subscriptions, books, and speaking tours.

On the bright side, at least now we know where the "Y2K" baloney purveors went. Even better, they are leaving us coders alone this time.