eMarketer reported today on an Ipsos study that claims affluent people spend the most time online.  Specifically, Ipsos says that heads of households with annual HH incomes of $250,000 or more spend 6 more hours on the Web each week than do users whose HH income fall between $100,000 and $150,000.

This may in fact be true, but claims like this only tell a surface story – and confuse independent and dependent variables.

Let’s assume that Ipsos’ observation is correct.  Consider the reasons that wealthier individuals spend more time on the Internet.  After all, is it plausible that simply having money makes you more likely to surf?  It’s more likely to be tied to a mix of underlying factors that, when taken together, are more predictive of web use.

* Where do the affluent live?  Do they tend to live in more urban/web-savvy locations, i.e. are they more likely to naturally find themselves in an environment where the Web is present?

* What level of schooling do the affluent reach?  Are they more likely to have post-graduate degrees?

* Were the affluent raised in homes where the Internet was more available?

* Did Ipsos normalize for age?

* Did Ipsos normalize for profession and work environment?

* What is the ethnic breakdown of the respondents in the study?  eMarketer has separately reported that 68.6 percent of non-Hispanic whites are currently on the Web, along with 74.8 percent of Asian-Americans and 53.8 percent of African-Americans.  It matters (though this idea may also be better explained by education, place of origin, access to computers, etc.).

In November 2007, Northwestern University found statistical relevance between a college student’s choice of social networking sites (such as MySpace and Facebook) and her race, ethnicity and parents’ education.  But what if a not-so-careful study instead concluded a causal effect between attendance at a certain school and social network popularity?  It’s not so hard to imagine.

Always use a critical eye when assessing data. Look deeper.

Ipsos  eMarketer