Variables lurking

A slightly different type of masquerade takes place when a single, unknown factor influences two outputs, and one of the outputs is mistaken as a factor. G. E. P. Box has given such masquerading factors the more dramatic names of either latent variables or lurking variables [Box, Hunter, and Hunter (1978), and Joiner (1981)]. 

Joiner, B.L. (1981), Lurking Variables Some Examples, American Statistician, 35, 227-233. 

Another effect is known as nonsense correlation, and is observed because inevitably we do not measure all the variables that can effect a method. If there is a latent or lurking variable that causes an effect in... 

Lurking Variable Variable outside the scope of a study that exerts an influence on the variables being studied. 

Analysis and Interpretation. In terms of statistical theory, the most complex step is the analysis and interpretation of data. When data have heen collected, described, and measured, conclusions can be drawn in a number of ways— none of which is right in every case. It is this step in which statisticians must ask themselves a few questions What is the relationship between the variables Does a change in one automatically lead to a change in the other Is there a third variable, or a lurking variable, not covered in the study that makes both data points change at the same time Is further research needed ... 

This discussion highlights some of the many lurking variables that affect the calculation of lattice energies. Calculations can give results that differ by up to 2-5 kJ moU for a number of often-hidden reasons while this is demonstrated here for atom-atom... 

It has already been established that the sin of homogenisation lurks within the field of tourist behaviour research (Galani-Moutafi, 1999). Tourists are not all the same, but it is usually inefficient when building the systematic study of a phenomenon to consider numerous individual cases in detail. It would, for example, be difficult if analysts endeavoured to document all of the variables of age, gender, nationality, economic well being, travel style, marital and family status, sexual preference, previous travel experience, attitudinal profiles and personality characteristics whenever they attempted to characterise a market. 

The missing feature or features can be best found by doing computer simulations. The model implemented on the computer is exactly that assumed for the system. Hidden variables that lurk in nature are eliminated. If there is disagreement now, it cannot be blamed on experimental inaccuracies. If there are problems in the mathematical approximations made in deriving the predictive equations, they will become visible. The theory and the computer experiments are both based on exactly the same model. Parameters of the model can be varied — to see how simulation results depend on them. Assumptions of the computer model can be varied as well - to see which were realistic and lead to sensible results, and which were in fact the root(s) of the problem(s). Improved understanding of the system and a better model with enhanced (or possibly even correct) predictive capabilities is a typical result. 

---