The relation between information and variance analyses

Garner, W. R.; McGill, W. J.

doi:10.1007/bf02289132

Cited by 194 publications

(64 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For a good discussion of multivariate information concepts see McGill (1954) and Garner and McGill (1956).…”

Section: Discussionmentioning

confidence: 99%

The Epidemiology of Skin Cancer in Queensland: The Influence of Phenotype and Environment

Silverstone¹,

Searle²

1970

Br J Cancer

View full text Add to dashboard Cite

SUMMARY.-On the basis of data gathered from long term residents of 3 widely separated regions of Queensland a multivariate analysis has been made to determine the influence of a number of factors in the aetiology of skin cancer and solar keratosis. Factors considered were age, sex, susceptibility to sunburn, complexion, eye colour, ancestry, occupation, clothing habits and residential district. For both sexes, both diseases and all age groups the factor " susceptibility to sunburn " proved to be the most powerful single discriminant. On the whole it appeared that the genetically based factors as a group provided more information on susceptibility than the environmental factors. The relative importance of " occupation " remains in some doubt. In the tropical area away from the coast it appears to be of considerable importance. In coastal areas its influence appears to be blunted, presumably by factors such as sports and recreation habits.AN earlier article in this series (Carmichael and Silverstone, 1961) used actuarial techniques to estimate the cumulative lifetime risk of incurring skin cancer in various parts of the coastal areas of Queensland, Australia. The proportion of males between 20 and 80 years of age who might be expected to produce at least one cancer varied from about 5 or 60 in the sub-tropical areas around Brisbane, to 12 or 130% in the tropical areas of Townsville and Cairns some 1000 miles nearer the equator (see Fig. 1). Rates for females were about 50 to 60% of those for males.These estimates were based on recorded data at the various skin cancer clinics, and while the survey would have provided reasonably accurate results it was not possible to relate the prevalence of skin cancer to the personal characteristics of those affected. The existence of such extremely high risks makes it possible, however, to conduct field surveys of the population to achieve this latter purpose. With this in view, three separate field surveys were made in areas which had widely different climatological conditions but which at the same time provided a good cross-section of the population.Procedures have been described in detail in two preliminary reports (Silverstone et al., 1963;Silverstone and Gordon, 1966) but for convenience a brief summary follows.Regional Surveys The areas studied were located on the sub-tropical coast (annual sunshine 2740 hours; annual rainfall 44 inches); the dry north-west tropical inland (sunshine 3300 hours; rainfall 17 inches); and the wet north tropical coast (sunshine 2660

show abstract

“…For a good discussion of multivariate information concepts see McGill (1954) and Garner and McGill (1956).…”

Section: Discussionmentioning

confidence: 99%

The Epidemiology of Skin Cancer in Queensland: The Influence of Phenotype and Environment

Silverstone¹,

Searle²

1970

Br J Cancer

View full text Add to dashboard Cite

show abstract

“…It is very important to realize that the mere identification of association does not provide grounds to establish causality. Garner and McGill [3] showed that information analysis is very similar to analysis of variance. One may then extend the argument of Liebetrau [11] to information-theoretic measures.…”

Section: Measuring Attribute Importancementioning

confidence: 98%

On Information-Theoretic Measures of Attribute Importance

Yao

Wong

Butz

1999

Methodologies for Knowledge Discovery and Data Mining

View full text Add to dashboard Cite

show abstract

“…Since McGill (1954), zero quantities of Q have been interpreted as the absence of interactions (Garner and McGill 1956;Garner 1962;Matsuda 2000;Bell 2003;Jakulin 2005;Yeung 2008;Lucio-Arias and Leydesdorff 2009). This claim is demonstrably false.…”

mentioning

confidence: 99%

“…Without examining the strange behaviour of Q vis-à-vis the probability distributions to which it responds, Leydesdorff (2009), relying on Garner and McGill's (1956) observation of parallels between their accounting information for information and the analysis of variance, concludes that 'the Q-measure is mathematically sound but an interpretation from the perspective of information theory needs to be provided'. He then cites Jakulin (2005) who called, following a suggestion by Gat (1999), positive Qs measures of synergy and negative Qs measures of redundancy.…”

mentioning

confidence: 99%

Information of interactions in complex systems

Krippendorff

2009

International Journal of General Systems

View full text Add to dashboard Cite

The relation between information and variance analyses

Cited by 194 publications

References 3 publications

The Epidemiology of Skin Cancer in Queensland: The Influence of Phenotype and Environment

The Epidemiology of Skin Cancer in Queensland: The Influence of Phenotype and Environment

On Information-Theoretic Measures of Attribute Importance

Information of interactions in complex systems

Contact Info

Product

Resources

About