The Continuing Need for Human-Based Media Analysis
Thursday, February 16, 2012 at 9:14AM 
By Albert J. Barr, Chairman & CEO, CARMA International, Inc.
Using computers to analyze media coverage is useful and amazing. Technology really has come a long way since I got into the media analysis business in 1984.
Back then, the challenge was how to convert information captured from an actual hard copy of a newspaper article into data that we could enter into a computer database. To accomplish this, I designed a system where all of our analysts used typewriters with paper forms. The fields being researched were typed on 8 1/2 x 11 inch sheets of paper that were printed with orange ink. The data gleaned from each article was typed, of course, in black. There was enough room on one sheet of paper for three articles.
image by Mustafa Khayat
To save time, and automate the process, we fed thousands of these forms into a scanner. The scanner could read the typed data but it could not see the orange ink that had the field names and boxes (media name, favorability, issues, messages, etc.) I can remember clients visiting our offices in Washington and being wowed at this creative, state-of-the-art idea while they watched the scanner input information from thousands of researched articles, thus replacing human typists for data entry.
We've advanced light years since then. In the early days, I don't think you could count the number of media analysis companies in the U.S. on one hand. Since the Internet, however, and tremendous advances in computer hardware and software, this has become a large and highly competitive business.
At CARMA International, we believe in technology. Computers can handle huge amounts of information efficiently. There is no way that humans can keep up with that kind of pace.
However, I believe that while computers are fast and relatively accurate, they still can't pick up sarcasm and all kinds of nuances that appear in media coverage. This is why I believe a strong need still exists for some kind of human intervention both in measuring and interpreting what all this coverage means to companies, governments, and all organizations who need to know what's been said about them.
I believe, at least for now, there has to be some form of compromise between using computers to digest millions of bits of information and humans to help analyze and interpret their meaning.
A good way to do this is to use the same approach that survey research firms have been using ever since they started polling. If you are using an automated service to "analyze" thousands, or even hundreds of thousands of mentions in both traditional and social media, it still makes sense to get a good statistical sample from this base and have real people analyze, measure, and interpret the sample. This way you can get dynamic results along with professional advice about what's being said, emerging trends, and a much more accurate measure of media sentiment.
George Fueschel, an IBM technician and instructor in New York, coined the term "GIGO," garbage in, garbage out. Wikipedia says the term "is used primarily to call attention to the fact that computers will unquestionably process the most nonsensical of input data ('garbage in') and produce nonsensical output ('garbage out'). It was most popular in the early days of computing, but applies even more today, when powerful computers can spew out mountains of erroneous information in a short time."
Quality Control is our mantra at CARMA. I would strongly advise anyone using computers for media analysis to include a serious element of human intervention. It's insurance so organizations don't waste precious funds on information that may prove to be of little to no use.
human analysis, media analysis, metrics, sampling 
The World's Leading Media Measurement & Analysis Company
