Replications – Armstrong, Hubbard, Rossiter et al.


Anthony Pecotich adds to the thread with statistics, theory and measurement

 ARC: Community: ELMAR: Posting

areas: methods: dialog

As always Scott Armstrong’s remarks are highly perceptive and on target. I thank both Scott and Ray Hubbard for raising and explaining a very important basic, but too often misunderstood issue in marketing research.

Replication of studies and or repeated independent, sampling is the foundation on which statistical theory is built. We use a statistical theory as a tool to help as to make difficult decisions concerning the results of our studies in the presence of imperfection and uncertainty. Indeed if our results were purely deterministic (occurred every time) we would not need the statistical theory as all the power of our conlusions would come from replications and or exhoustive observation.. Further, this statistical theory is based on assumptions that are not necessarily an empirical reality. Philosophers of science and statisticians such as Fisher, Box and Tukey all stated, in one form or another, that the results of only one study are provisional on further research and that data should be analyzed in many different ways for a complete understanding. The truth content of events in the real world does not depend on statistical probabilities but on repeated occurrences. This is self evident, and to suggest that there is no need for replications is somewhat like stating that once a lottery has been drawn and a winner decided there is no need for further draws of the lottery, we simply give all the future prizes to the same person. I would most certainly not buy tickets in such a lottery and would be happy to take the money from anybody that is foolish enough to do so.

Although there are philosophical issues associated with the separation of measurement and theory and strong arguments have been made by Norman Anderson that "maesurement theory is an organic component of substatntive theory" it is clear that when conducting research we assume among other things:

  1. That a real world exists independent of ourselves and the method of observation.
  2. That the nature of our measurement and or observation is such that in the absence of a theoretical explanation we should get the same result on repeated observation (this is the basis of measurement validity). Michael Jordan was 2 foot 6 inches when he was say 4 years old now he is say 6 foot 6 – we all know and understand the reason for that. I am sure that serious questions would be asked of anybody who claimed that yesterday Jordan was 2 foot 4 but today he is 6 foot 6.
  3. That the results of our observations (measurement) may be coded so as to corespond to some properties of the mathematical number system, However, this is problemmatical and has to be established by replication. For example, height and weight have an empirical anaologue of addition (not always but within a range) while there seems to exist no such procedure for the addition of psychologica properties. Fechner, Thurstone, Stevens, Torgeson, Guilford, Anderson and so on have all made this clear in their work.

The appearance of this controversy in ELMAR should be of concern to the marketing community as it is likely to mislead the discipline and editorial policy. It seems that something is dreadfully amiss with the editorial reviewing processe given some of the nonsense being published in recent times.

Dr. A Pecotich
Business School
The University of Western Australia
Facsimile (08) 6488 1004,
Telephone (08) 6488 2892
International code 61 89 6488 2892