Z-Stats / Basic Statistics

Madelon F. Zady, Ph.D., begins a series on statistics. This first lesson is easy. All she does is explain the best way to learn about statistics (without falling asleep).

EdD Assistant Professor
Clinical Laboratory Science Program University of Louisville
Louisville, Kentucky

• Experience and attitude
• Recognition of different ways of learning
• Importance of statistics in the big picture
  • Confidence in communication
  • Quality of products and services
• Need for statistics in Clinical Laboratory Science
• References
• Author biography

[This lesson is intended for those who do not have a background in statistics or who did not profit well from their prior experiences with statistics. If you believe that your statistical background is secure, you may move on to the next lesson.]

As a 30-year teaching veteran, I have only one goal. I want you to understand this material. I will later make the argument that the future of your profession and, more importantly, the quality of laboratory test results may well depend upon your ability to deal flexibly with statistical concepts.

Experience and attitude

Since a statistics course is usually a general education requirement in many pre-professional curricula, perhaps you have sat through a statistics course before, only to be bewildered, then intimidated and then bored (out of frustration) by the concepts. Some of my students have told me that they passed the course with the idea of never again having to be subjected to this oppressive material. To say the least, these students did not get THE BIG PICTURE from their statistics course.

It is probably best to start off by meeting the problem head on and addressing the general attitude toward statistics or "stats" as they are commonly called. When I start to teach this subject matter to my students, I am met by a generally flattened affect, i.e., eyes glaze over and brains go into hibernation. I find myself having to dip in to my repertoire of novelty cues in order to provoke the slightest amount of interest. Even with that, the students are constantly monitoring for any sign of statistical opaqueness that will cause them to assume their prior "altered state." Interestingly, the students themselves are aware of their tendencies and their reactions become a source of humor in the classroom. Math anxiety reigns supreme! I pledge to you that I will make every effort to simplify the concepts for you. And I will not take any math shortcuts that leave you wondering and wandering.

Recognition of different ways of learning

Coming from an educational psychology background, I have a lot of sympathy for students who are trying to learn stats. Not being much of a behaviorist, I have examined other learning theories and found them to be more appropriate for the classroom. Two features of these theories spring to mind.

• One approach is called whole-part-whole learning [1], which means giving the student the big picture, then breaking it down into its parts and then reassembling the concept. In short - tell them what you are going to tell them, tell them, and tell them what you told them. In the end, the student is the one who should be actively constructing the concepts.
• Another interesting idea is Jerome Bruner's "learning by successive approximations" [2]. Using Bruner's instructional approach, the learner is exposed to the same concepts several times while she makes more and more out of them each time (also called cognitive reorganization or mapping). In this way, a student is not penalized for not "getting" the concept in its entirety the first time it is presented.

That said, the student also bears the responsibility to remain active in the process of learning. That will only happen if the student feels a certain amount of power over the subject matter. To that end, I find it helpful at first to acknowledge some of the limitations of statistics to students, particularly if they have not been terribly successful in their first exposures to the concepts.

Importance of statistics in the big picture

Parametric statistics are based upon the axiom that everything that is important is normally distributed. The corollary is: If an entity is not normally distributed then it is not important. Of course this assumption is questionable, for example, gender is not normally distributed. (It is considered a category-variable.)Yet this culture considers gender to be quite important. Statisticians also agree that gender is important, so they have devised a way to convert it to a measured variable and in a way to "normalize" its distribution!

Despite the limitations, statistics are necessary for at least two reasons:

• Stats are important because, in this culture, numbers can lend confidence to a communication.
• Most importantly, statistics will help us to insure the quality of laboratory testing.

Confidence in communication.

If a friend or a colleague told you that most adult males spend far more time reading as compared to adult females, because that was true in his family, would you accept that person's statement? Now, what if the results of a survey by a book publisher showed that males read 95% more often than females? How would you react to that statement? Of course there would have to be additional studies/surveys, but a 95% difference in means is large and meets what we call a "p of.05." Most statisticians and publishing houses would have an understanding of and confidence in this latter statement, because it is backed by the numbers. Presumably the results can be replicated and generalized to other situations.

Quality of products and services.

Suppose a hemoglobin determination on a patient's sample was performed in a doctor's office. The person who performed the procedure was trained on the job, and the total cost of the procedure including salary/benefits was $1.50 (American dollars). Now the same sample was sent to a licensed laboratory and a certified analyst performed the procedure. The total cost of this procedure including salary/benefits is $3.00 (American dollars). How do you think an insurer would interpret this situation? How would you as a trained analyst interpret this? You could say: "Oh, but the quality in the licensed laboratory would be better." How can you demonstrate this? Where are the numbers to back you up? How is your statement any different from the claim made above by the friend who told you that adult males read more than females because that was true in his family? Well the answer lies in the confidence that can be placed in a communication that is reinforced by statistical analyses.

Suppose that several hemoglobins performed in the doctor's office were compared with the values obtained in the licensed laboratory using powerful statistical tests, perhaps a real or substantiated difference could be demonstrated. When the significance of those statistical differences are communicated to laboratorians, insurers and legislators, this information could affect the quality of any future laboratory testing.

Need for statistics in clinical laboratory science

You see, even though statistics can "turn off" former students and even though statistics can be "manipulated," they are useful in providing uniformity and confidence in the communication of information and in providing for the quality of laboratory testing in the face of fiscal constraints.

Over the past few years, some of this information concerning the difference in the test results, depending upon the analyst and laboratory setting has reached legislators and made an impact on the QC programs. The upshot of the resultant legislation is that certified analysts to have statistical knowledge beyond just the mean, SD and CV. This is especially true in the areas of method validation and quality control, which are now required by US CLIA laboratory regulations. Despite these regulatory efforts, the quality of laboratory testing is now threatened by changing skill levels of those who perform tests in different healthcare settings and the pressure to reduce the costs of healthcare. As stated by Westgard in the foreword to Basic QC Practices [3]:

"The 1990s will probably be remembered as the 're-decade' in healthcare - reorganizing, restructuring, and re-engineering to reduce costs. The management strategy of 'doing more with less' has meant that laboratories end up with less staff with less education, less experience, and less training. Given the major changes in organization and processes, it would be expected that quality control efforts should have increased to guard against the dangers of doing more with less, i.e., doing more tests less well. But laboratory inspections continue to cite QC practices as one of the most frequent and series deficiencies, suggest more tests are being done with less quality control."
Laboratorians need to develop a more flexible working knowledge of statistics to creatively manage analytical quality and implement the method validation protocols and statistical QC procedures that are necessary. These basic lessons in statistics will help you take steps in that direction.

References

1. Ausubel DP. The use of advance organizer in learning and retention of meaningful verbal material. J Ed Psychol 1960;51:267-272.
2. Bruner JS. Toward a theory of instruction. London:Cambridge, 1966.
3. Westgard JO, Quam EF, Barry PL. Basic QC Practices. Madison, WI: Westgard QC, Inc., 1998.

Biography: Madelon F. Zady

Madelon F. Zady is an Assistant Professor at the University of Louisville, School of Allied Health Sciences Clinical Laboratory Science program and has over 30 years experience in teaching. She holds BS, MAT and EdD degrees from the University of Louisville, has taken other advanced course work from the School of Medicine and School of Education, and also advanced courses in statistics. She is a registered MT(ASCP) and a credentialed CLS(NCA) and has worked part-time as a bench technologist for 14 years. She is a member of the: American Society for Clinical Laboratory Science, Kentucky State Society for Clinical Laboratory Science, American Educational Research Association, and the National Science Teachers Association. Her teaching areas are clinical chemistry and statistics. Her research areas are metacognition and learning theory.