It’s been stated that 98 percent of all statistics are fabricated to support opinions. Listen to Todd Snider sing about misuse of stats and check out his lyrics to Statistician's Blues. Maybe Fletcher Knebel had a point when he said ...
It's now proven beyond a doubt that smoking is a leading cause of statistics.
William Watt, a lawyer, warned people not put their faith in statistics until 
carefully considering what they do not say. Whether accurate or flawed … statistics impact daily business decisions and are well worth careful consideration.
Most would agree that numbers alone rarely tell an entire story, though. Check out how Chris Jordon expands numerical statistics … into images and photographs to picture abstracts. People stronger in mathematical intelligence may prefer numerical data … whereas people with higher spatial intelligence likely prefer graphic illustrations.
Inaccurate statistics lead to flawed outcomes. If erroneous statistics report that injuries declined in the previous year, leaders may neglect their responsibility to ensure safer settings, placing employees at risk, for instance. Statistics should be gathered by unbiased researchers. Why so?
Whenever organizations create incentives for workers to underreport any increase in accidents at work, for example, related statistics lose their reliability.
How can you make business decisions based on more reliable statistics?
1. Consider the number of cases observed. Statistics at the New York State department of Health are considered reliable only if they are based on more than 20 observations or cases. When fewer than 20 cases exist to prove a point … it is almost impossible to distinguish random fluctuation.
2. Check dates of data collection that may favor one group over another in a different era. For instance, researchers used statistics to report that overtime workers are 61 percent more likely to injure themselves on the job. That study released in 2005 is still being used to deny overtime to some 2008 workers.
3. Ensure results from data apply accurately to all related policies created. It’s usually a good idea to compare several studies – especially when new applications of a study differ from those used in the original research study.
4. Familiarize yourself with how outcomes are gathered and used. For instance, are you aware that 3 of 4 companies monitor their workers' privacy and gather data on what you do … how you communicate and with whom… and what Internet sites you visit?
5. Question statistics that work against your career or your firm’s advancement. A recent study that examined business practices, for instance, positioned women in dead end career paths … without opportunities for advancement. Have you seen it happen?
6. Bookmark reliable statistical sources … such as FedStats … which makes more than 100 agencies available to you. Another site … the Bureau of Labor statistics shows projects of the labor force by most any business category of interest … from productivity to wages.
7. Compare studies done with similar stats in other fields or countries to catch a wider picture. For instance … the World Wide Web Virtual Library shows data sources from this country as well as reliable sites from Africa, Canada, Asia, China, India, Israel, Korea and many more.
Accurate statistics tend to result in best practices. Statistical data in any business should benefit … rather than confuse workers. If, though, you feel bamboozled by statistics at work … why not enroll in professional development courses designed with business statistics in mind. How have you used statistical data in the past week?
Specific rules about what statistics to believe are always a little dangerous. In my programs I've covered some of the things to do when analyzing statistics.
Ask "Who's responsible for the study?" This involves the people who commissioned the study and the people who actually did the study. It also may involve the people publishing the study.
Find out what questions were asked. The wording of questions in a survey can dramatically influence the outcome.
Sometimes the wording leads to entirely subjective responses. On one study I evaluated, for example, respondents were asked about their drinking of alcoholic beverages and asked to choose from "none," "moderate" or "heavy." No guidance was given on how to determine those evaluations.
Favor behavioral studies over opinion polls. People often tell surveyors what the surveyors want to hear or what the surveyed people think reflects well on them. Even if they answer honestly, many people are poor describers of their own behavior because they answer based on what they intend to do rather than what they actually do.
Find out who was studied. Many psychological studies use students because they're available and cheap. That may be OK, but then again, maybe not depending on the subject.
Find out what kind of sample was used and how big it was. In many "surveys" reported in the media, the sample size is very small, often less than a hundred and is what surveyors call an "opportunity sample," meaning that whoever chose to answer was in the pool.
Posted by: Wally Bock | January 6, 2008 5:15 PM | Permalink to Comment