On 2014 Nov 30, Harri Hemila commented:

Believing that the 2% average effect of vitamin E on mortality is valid for all ATBC participants is an example of ecological fallacy.

The ATBC Study, Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group., 1994 reported that the average mortality in participants administered vitamin E was not different from the non-vitamin E group with RR = 1.02 (95% CI 0.95, 1.09).

However, a subsequent subgroup analysis found that the effect of vitamin E was not uniform over all the ATBC Study participants. The combination of age and dietary vitamin C modified the effect of vitamin E so that there was strong evidence of heterogeneity in the effect of vitamin E over 6 subgroups (P = 0.0005) Hemilä H, 2009. Furthermore, given that baseline age modified the effect of vitamin E, a further analysis by the participants’ follow-up age was done and a significant difference in the effect of vitamin E on mortality was observed before and after the age of 71 years (P = 0.03) Hemilä H, 2011.

Ecological fallacy occurs where data about a group is used to conclude information about an individual. Thus, assuming that the +2% average effect of vitamin E on mortality in the ATBC Study is valid for all ATBC participants is an example of ecological fallacy. A substantial proportion of the ATBC Study participants are inconsistent with the average effect.

On 2014 Nov 30, Harri Hemila commented:

Believing that the 2% average effect of vitamin E on mortality is valid for all ATBC participants is an example of ecological fallacy.

The ATBC Study, Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group., 1994 reported that the average mortality in participants administered vitamin E was not different from the non-vitamin E group with RR = 1.02 (95% CI 0.95, 1.09).

However, a subsequent subgroup analysis found that the effect of vitamin E was not uniform over all the ATBC Study participants. The combination of age and dietary vitamin C modified the effect of vitamin E so that there was strong evidence of heterogeneity in the effect of vitamin E over 6 subgroups (P = 0.0005) Hemilä H, 2009. Furthermore, given that baseline age modified the effect of vitamin E, a further analysis by the participants’ follow-up age was done and a significant difference in the effect of vitamin E on mortality was observed before and after the age of 71 years (P = 0.03) Hemilä H, 2011.

Ecological fallacy occurs where data about a group is used to conclude information about an individual. Thus, assuming that the +2% average effect of vitamin E on mortality in the ATBC Study is valid for all ATBC participants is an example of ecological fallacy. A substantial proportion of the ATBC Study participants are inconsistent with the average effect.