Net Promoter Scores For Obama And Romney
September 5, 2012 13 Comments
I decided to tap into the energy surrounding the U.S. presidential election by doing a Net Promoter Score (NPS) analysis. In Temkin Group’s latest U.S. benchmark survey that we fielded in August, we asked a number of questions about the candidates and blended that data with our rich demographic and attitudinal data.
In this first post of the series, I am examining overall NPS for the candidates and the difference across age groups. I also looked at the percentage of promoters by gender, ethnicity, and level of education. As you can see in the infographic below:
- NPS scores are very low for both candidates (-57% for Romney and -33% for Obama), as the percentage of Detractors more than doubles Promoters.
- Obama has more support from all age groups 64 and younger while Romney has more support from consumers that are 65 and older.
- Obama has a more sizable lead with females than with males.
- Both candidates have about the same support from Caucasians, but Obama has more than a two-to-one advantage with Asians, Hispanics, and African-Americans.
- The gap with African-Americans is immense. With this group, Obama has an NPS of +39% while Romney’s is -87%.
- Obama’s advantage in Promoters is about the same for consumers with or without college degrees.
The bottom line: Obama leads across most demographics, but NPS is awful for both candidates.
P.S. See additional information on NPS on our VoC resource page
About Bruce Temkin, CCXP
Customer Experience Matters® 
Bruce-
I don’t buy the result. Did you conduct a random sampling? Did your sample consist of registered voters or likely voters? Was there a sample bias around Independents, Democrats and Republicans? Results should be weighted according to party representation. Right now slightly more people are registered independent, close behind that are Republicans and then there are Democrats coming in third. I contend that if you put these factors together you would likely see the NPS for both candidates much closer than your current result.
Rob
Rob Madonna, PRC | Customer Research | Plymouth Rock Management Co. of New Jersey | 908.219.5475 | 914.260.7261 | rmadonna@plymouthrock.com
Rob: The survey was a random sample of an online panel with respondent demographics matched to the US Census population distribution for age, income, region, gender, location and ethnicity. It does not include any quotas for party affiliation. There are some know “biases” for online surveys such as under-sampling the elderly (over 70), non-english speakers (largest gap is for primarily Spanish-speakers), and very poor (who have limited online access), but those would likely have only a minor affect on these results. Obviously it’s a small sample of the US, but I’m confident that these results are fairly representative of the US population.
While the sample may mirror random representation in order to do this with precision the variables I mentioned in my previous post must be considered. Many studies show a vast difference in response behavior between the total population, registered voters and likely voters. The same rules as a standard study do not apply in this case. I maintain that if everything is applied using a sample only of registered voters, weighted by party affiliation — Dem, Rep or none, we would see a vastly different result. It isn’t relevant whether the questions are designed as NPS or as a traditional poll. These results may represent the US population but I would wager serious money that they do not represent those who will vote in November
That being stated, I am a NPS skeptic. I have seen it fail too many times to be a believer. I read a study about 2 years ago that those identified as promoters tend to be “unprofitable” customers who are tribal in nature and tend to promote to others who are also unprofitable.
Rob
I love it!!
☺
It does seem like your results are far more negative for Romney compared to the national polls?
http://www.realclearpolitics.com/epolls/2012/president/us/general_election_romney_vs_obama-1171.html
Dan: Thanks for joining in as well. Our analysis was totally based on the Net Promoter Score, which is a completely different question than other polls use. A lot goes into these types of scores: sampling methodology, timing, survey design, and the calculations done with the results, etc. I can’t tell you if these NPS scores are more or less accurate at predicting voting patterns, but I agree with your observation that the gap in NPS seems to be higher than the gap shown in many of those other polls. Any thoughts about why that might be? The demographic findings, that Obama leads more with women, minorities, and young people, seems to be consistent with what I’ve heard from political pundits. The good news, for me anyway, is that I have no interest in forecasting the outcome of the election 🙂
The gap could exist for the reasons I have been stating. There are so many more dimensions to a political issue than the NPS approach can handle in my view — at least a straight read of the results. There are too many preexisting issues involved to run this as a straight NPS. I think NPS is a decent quality indicator but not very precise when it comes to modeling consumer behavior. Besides politics are a mess today which is why, when I consult I avoid political surveys. I’ll stick with the “orderly” world of products and services.
Again, i would be interested to see this study re-run using the approach I have suggested.
Funny. These are the kind of NPS scores for backing an insurance in the Netherlands in THE middel of the credit crunch.
Bruce,
I recently published two thought pieces on the concept of using CX to help our government and our country. You are spot on doing a first-ever formal evaluation. Would love to see this expanded in some way.
http://www.customermanagementiq.com/cem-multi-channel-customer-service/articles/democracy-the-ultimate-customer-experience-strateg/
http://theserviceleader.wordpress.com/2012/08/15/can-cx-help-usa-co/
Enjoy and best!
Very interesting piece of work ~ love it! The value of this work can be to test the accuracy of the NPS to predict the future; it will be interesting to match this NPS findings to the final outcome. Do you have results per state?
In terms of the larger gap between your NPS and other polls ~ my thoughts. Remember the NPS is based on a single question, but the score is a net measure of two poles of this rating. In effect you are ‘doubling’ your margin of error hence the greater variability. Keep in mind that the range of the NPS is -100 to 100 and not 0 to 100 (or 0 to 10 or %) as I suspect many polls are.
I’ve also seen in many of the research where we combine/compare NPS with other satisfaction metrics that the NPS fluctuates much more, and has a larger variance than standard satisfaction scores ~ although the correlation between the NPS and standard Satisfaction scores are very high. In other words the direction of movement is in sync, but the ‘degree’ differs.
Cant wait for the next post.
Kobus (South Africa)
One of the things to keep in mind about NPS is that it does not use an average score, which other measurements often use, but nets out the low ratings from the high ratings; isolating the most avid supporters. So the candidates could have similar scores in other polls, showing the same general level of support, but have a quite different NPS if one candidate has a larger portion of avid supporters (“Promoters” in NPS speak) or avid dislikers (“Detractors”).
Very interesting information. What NPS seems reveal is the amount of dedication and enthusiasm a person has for their candidate. It would interesting to have this score done on independents to see which candidates seem to motivate an individual to swing one way or another during an election cycle. As you pointed out, which seems to be historical rhetoric, is that the economy and a future plan holds the most significance to a person’s ideologies when the economy is a recession. Bill Clinton actually referred to this idea when he spoke of the economy improving in 1994 but individuals couldn’t feel it. It does really seem like NPS is similar to a sort of consumer sediment poll with a more specific type of questioning. Just some thoughts.
This is a great conversation everyone. This research captures the sentiment across the broad US population, but does not attempt to predict voting patterns. I’ll leave that analysis to the pollsters. I agree that NPS is not a perfect measure. I’ve worked with dozens of companies and find that many of them misuse the metric. But it has had a very positive impact on many organizations. Take a look at my post: 9 Recommendations For Net Promoter Score.