TICE – Turkish Integrity Center of Excellence

The Too-Small Sample

The Too-Small Sample

Topic: Marketing Research (Distortion/Falsification of Data)

Characters: Roland, statistician for a small marketing research firm Todd, owner of the marketing research firm

 

It was Monday morning and Roland, the statistician for a small marketing research house, was sitting at his terminal cleaning up the tabulation program for a telephone survey that was currently in the field. Todd, the owner of the firm, walked into Roland’s office. “Roland, I just heard from the interviewing service that the data collection for the bank study won’t be completed for at least two weeks. I guess we should have pretested the questionnaire. It’s taking twice as long to complete the interviews as we estimated. Not only are we going to lose money on this project, but the final report is due in two weeks.”

“Oh no!” cried Roland. “We need a week just to edit and code the open-ended questions before data entry can begin. We’ll have to beg for time from the bank.”

“We can’t do that. Remember how I got the project? — They had been quoted $30,000 by Jones & Wilson Research and I told them we could do it in half the time and for only 20 grand.” There was silence for a few minutes and then Todd added, “I’ve got an idea. We stop interviewing now with the 352 interviews that are complete and bump the numbers up to 500 with a multiplier, you know, a weighting routine. Roland, you can do that with just a few lines of programming. The bank will never know. Hell, they don’t know anything about survey research much less statistics. The printout will only show the bumped up numbers and they will never see the programming.

Before Roland could say anything, Todd was walking out of the office saying, “Start doing your magic with the programming now and I’ll call the interviewing service. This is great, we’ll make money and get the job done on time. Go to it, Roland!”

After Todd left his office, Roland said aloud, “Terrific, my first job as a real marketing research statistician and I’ve got an idiot for a boss.” As he reflected on the situation he thought, “bumping-up the numbers without telling the client is like stealing their money. And even if they are told, I’m not sure if the statistical routines we plan to use will be appropriate for weighted data. I don’t know what to do. I guess other companies weigh data and I do need this job. But…”

Author: Thomas J. Cosse, Ph.D. , Professor of Marketing, E. Claiborne Robins School of Business, University of Richmond

The Too-Small Sample