Today, you’ll (hopefully) learn something about how Nielsen Audio defines who is Hispanic and their use of language. Like so many things involving ratings, it’s not a simple “cut and dried” definition.
Nielsen attempts to collect race and ethnic information about all rating participants, even in markets where that status may not be relevant to the local market, for example, my home metro of Bowling Green, Kentucky. That’s because the individual’s data may be utilized in other services where race and ethnicity are needed. If you’re working in a market that doesn’t have race and/or ethnic weighting, that’s why Nielsen asks about it.
Due to software limitations that go back decades, Nielsen Audio can only classify individuals into three categories: Black, Hispanic, or Other. If you work in a market that has no race/ethnic weighting, then everyone is Other. If only one of the two groups is weighted, for example, Hispanic and not Black, then the Black respondents become Other.
Of course, the individual gets to decide which category they feel is closest to their identity, but Nielsen Audio’s limitations only allow you to be in one category. Some Hispanics are Black (and vice versa), but you can’t be both as far as Nielsen Audio is concerned. Typically, the hierarchy will be that you’re Hispanic first, followed by Black, and then Other. And if you consider yourself to be Asian, Native American, etc., forget it, although Nielsen does track Asian in PPM even if the service can’t weight on that dimension.
Keep in mind that Hispanic is not a race, but an ethnicity. Good survey practice requires that race and ethnicity questions be kept separate, and Nielsen does that. But good survey practice doesn’t always jive with peoples’ perceptions. In the 2020 Census, about one out of seven respondents said they were “some other race” and 90 percent of those respondents also reported they were “Hispanic”. You can see the problem.
If you’ve answered “yes” to the Hispanic question, Nielsen can move on to “language dominance”. The concept has been around for a long time concerning radio ratings having been first used by Birch/Scarborough back in the 1980s for their ratings service that competed with Arbitron at that time. Currently, 28 metros have language weighting.
The reason weighting is used is to account for differences in the sample compared to the population on variables that have something to do with what is being measured. Age and gender matter. Race and ethnicity matter. Some variables don’t matter at all. For example, I’m left-handed, but does “handedness” have anything to do with radio listening? No.
The premise behind language dominance is that speaking only Spanish makes you more likely to use Spanish language media. If you speak only English, the result is probably the opposite. Nielsen asks about language use at home, which gives different results than usage away from home. It’s a four-part question, with options for Spanish only, mostly Spanish, mostly English, and English only. If you’re truly bilingual, you have to add that response on your own.
When we’ve discussed weighting in the past, we’ve talked about universe estimates or population estimates (the terms are interchangeable for our purposes). For most variables like age, gender, geography, race, and ethnicity, census-based estimates are available and utilized.
The problem with language is that there isn’t a census-based estimate for language usage “at home”. The census does ask about other languages that you might speak and how well you speak English. If you speak Spanish, but can’t speak English, that should be pretty close to what Nielsen is measuring, but Nielsen Audio doesn’t work that way.
Nielsen creates its own universe estimates for language dominance. In the past, it conducted a major national study of language dominance among Hispanics that could be used for both national and local video and audio markets. I’ve not been able to verify whether that very expensive study is still being conducted or whether Nielsen has resorted to a less costly method.
In the past, we’ve talked about weighting and if you’re in a language-weighted metro, you should always check how the sample performed compared to the population estimate. If Nielsen misses badly, especially in the direction of English dominance — in other words, too few Spanish-dominant Hispanics were in the sample compared to the estimate — then the ratings can move wildly. This can even happen in small markets.
While at Cumulus, I had numerous conversations with Nielsen regarding Oxnard-Ventura, a majority-Hispanic market that is currently ranked 125. Invariably, Nielsen Audio could not get the sample close to the purported population for Spanish-dominant Hispanics and the result would be big numbers for the Spanish stations, while Cumulus’ English language stations lagged.
Check your market in the E-Book and see if language dominance matters. Even if you’re not in a language-weighted market, compare Nielsen’s race/ethnic sample to the population. It’s all in the E-Book.
Let’s meet again next week.
