PHILADELPHIA: US marketers could make major savings in survey research by using smaller sample sizes without significantly increasing the margins of error, a leading academic has claimed.

Jeff Savitz, Professor of Statistical Science at Temple University and CEO of SavitzConsulting, based his assertion on the identification of so-called "inliers" within random samples.

The central limit theorem of statistics states that the average of a random sample will more closely resemble the average for the whole population as the sample size increases, regardless of the shape of the distribution.

So, for example, with a random sample of 400, with 95% confidence, all percentage estimates of the US population will be within 4.9% of their true values while with a sample of 1,000 the margin of error is only 3.1%.

Professor Savitz has developed new methods for identifying "inliers" that have predicting error rates only two thirds that of a random sample.

According to Savitz, "these inliers are uniformly present in most all demographic and psychographic groups making them useful for estimating population parameters for virtually any consumer target".

He interviewed over seven hundred randomly selected consumers from the Toluna national online panel and asked them to rate 30 different popular brands. Differences between averages given for the entire random sample and the Savitz "inliers" averaged only 2.4% with a maximum of 4.1% for McDonalds.

"This is exactly the kind of fresh thinking that marketers need as brands manage more fragmented buyers and look for more laser-focused intelligence on how to be relevant to the most valuable consumer segments," said Adam Gargani, Planning Director at Ogilvy, and Foote, Cone and Belding.

The Council of American Survey Research Organizations estimates that US researchers spend $1bn annually on samples for research. Since only two thirds as many of the Savitz Inliers are necessary to achieve the same margins of error, this could translate to a cost savings of upwards of $300m annually in the US alone.

"The savings on sample acquisition is only the tip of the iceberg," said Dr Mike Morgan, former professor at Cornell University and Senior Marketing Research Consultant in the technology and telecommunications categories.

"Inlier samples mean fewer interviews, fewer incentives, and efficiencies in data processing, potentially freeing up hundreds of millions of dollars every year."

Data sourced from Savitz Consulting; additional content by Warc staff