February 3, 2015

Social media preferences vary by race and ethnicity

Latinos, blacks and whites use social media networks about equally, but there are some differences in their preferences for specific social media sites. For example, Instagram is more popular among Latinos while Pinterest is more popular among whites, according to a late 2014 Pew Research Center survey.

Today, about eight-in-ten Latino, black and white adults who are online use at least one of five social media sites – Facebook, Instagram, Pinterest, LinkedIn and Twitter. Among these, Facebook stands out as the most widely used platform, regardless of race or ethnicity: About seven-in-ten adult internet users (71%) say they use the site.

Latinos', Blacks' Use of Social MediaBut there are differences by race and ethnicity in the use of other social media sites, in particular Instagram. The photo-sharing site is more popular among Hispanic and black internet users than among white internet users. About one-third (34%) of online Hispanics use Instagram, as do 38% of blacks. By comparison, only 21% of whites use the network.

Instagram’s popularity among younger adults is notable. For example, roughly half (53%) of online adults ages 18 to 29 use the service, compared with 25% of those ages 30 to 49, 11% of those ages 50 to 64 and 6% of those 65 and older. This is worth noting when looking at social media use by race and ethnicity since Latinos are significantly younger than other groups. More than a decade separates the median age of Latinos (27) and whites (42), while the median age for blacks is 33.

By contrast, the social media site Pinterest is more popular among white internet users than among other groups. About one-third (32%) of whites use Pinterest, compared with 21% of Hispanics and just 12% of blacks. LinkedIn differs in that about equal shares of whites (29%) and blacks (28%) use the site, compared with 18% of Hispanics. Meanwhile, Twitter has a more equal distribution. About one-in-four Hispanics and blacks use the site, along with 21% of whites.

Topics: African Americans, Internet Activities, Race and Ethnicity, Social Media, Social Networking

  1. Photo of Jens Manuel Krogstad

    is a writer/editor focusing on Hispanics, immigration and demographics at Pew Research Center.


  1. Anonymous5 months ago

    Dear Pew Research… A couple of years ago I asked why Asian/Pacific Islanders are not included in your research and the response was you don’t have the capacity to translate your surveys into all the other languages. My reply was we are literate in English and you can present the data that the survey was done in English. It is 2016 and still, Asian/Pacific Islanders are not included. WHY?

  2. Anonymous7 months ago

    Without knowing how “use” is defined – the study is kind of meaningless. We know that FB is constantly sending out updates and notification and other unsolicited information – which as soon you engage any of these pieces of info you are counted as FB “user.” Even though I probably don’t use FB ten minutes a month, I still get daily communications from them. I don’t consider myself a FB user more of someone used by FB.

  3. A Guy2 years ago

    Yea, why are Asians not included in this survey? This is not the first time I’ve noticed that Asians have been excluded from Race & Ethnicity surveys at Pew and it’s disheartening.

    1. Jordan Blaza5 months ago

      here’s their response: pewresearch.org/fact-tank/2016/0…
      “U.S. Asians are more likely than whites and blacks, but not Hispanics, to lack proficiency in English. Just 62% of U.S. Asian adults speak English proficiently, while 38% speak English less than very well. By comparison, 38% of Hispanics, 3% of blacks and 2% of whites speak English less than very well, according to self-reported data among non-institutionalized adults in the 2014 American Community Survey.”
      they need to do a better job reaching English speaking Asians because this is not acceptable.

  4. Pew2 years ago

    And again, Pew shows us Asian Americans that we don’t matter.

    Thanks, Pew!

  5. Valerie Lambert2 years ago

    Why didn’t you show the splits based on gender? For example, Pinterest is much more preferred by women. And for those of us data geeks, you could have sent us to a deeper link that splits out black men/women, latino men/women, white men/women for each media channel.

    I also agree with the other commenter – why did you leave out Asians in this study? It’s not like Pew to disregard a significant data point when researching.

    1. Jim Ewins7 months ago

      Do those who use social media matter?

  6. Martin Vega2 years ago

    I would be cautious about interpreting these results as a nationally representative sample, given the study’s myriad issues:
    * There is an over-recruitment of Hispanic cell-phone sample significantly above the levels cited by the CDC for Hispanics with cell-phone only households: PEW: 76%; CDC: 56%
    * The Hispanic portion of the sample as a percent of the total sample is just 9%, well below the percentages Latinos comprise of the total U.S. population as well as those online
    * Some of the social media metrics are based on sample sizes that are unstable, thus, should only be used directionally: LinkIn sample: 33; Pinterest sample: 39

    My recommendation would be that if you are going to invest in a study of this nature to have actionable results, you should allocate sufficient sub-samples so they are stable enough to ensure projectable results. In addition, your interview modes by telephone (i.e., landline vs. cell) should mirror or approximate the CDC’s incidence levels.

    1. Alan Rosenblatt2 years ago

      Martin, regarding your comments, while I was not involved in the study nor am I connected to Pew in any way, I do have extensive graduate school training and professional experience as a research methodologist. Given this, I find your critique uncompelling.

      First, as far as I know, “unstable” sample sizes is a meaningless phrase. A sample size is what it is. The question is whether or not you can infer from the sample to the population. The Central Limits Theorem assumes a normal sampling distribution. Samples of N=30 is sufficient to make this assumption (note sample sizes on any Z-table). More importantly, the application of the CLT requires a sufficiently large sample overall, not within individual cells. In this case, N=1445, which is more than adequate (by a long shot).

      As for what seems to be a contradictory concern on your part about oversampling of Hispanic cellphone users and an undersized sample of Hispanics… that is the reason for oversampling… to ensure a sufficiently large sample. Once oversampled, the appropriate weighting is applied to correct the data.

      As for the 9% of the sample being Hispanic, it is inevitable that specific breakdowns in any probability sample will yield some oversampled and some undersampled subgroups. That is why analysis is generally based on the CLT, which looks at the specific sample within the context of a sampling distribution.

      So while I would never claim this is a perfect sample, there is no indication in the report, nor in your critique that it is a bad sample.

      When it comes to making strategic decisions about targeting and channels in a social media campaign, we use the best data available. And better still if we use multiple data sources. This data is a good basis to work from. Corroborating its findings with other studies would be a plus, but it would not negate these data.

  7. B NGTM2 years ago

    Why in this survey Asians are not included?

    1. wei lu2 years ago

      Good question 🙂 I am going to ask the same thing! Did you look into the report details?