Fig. 2From: Time for a reassessment of the treatment of hypothyroidismDemonstration of correlation bias by combining dissimilar groups. In randomly sampled groups showing the same strong inverse correlation (blue and red regression line) but slightly shifted centres, a combined analysis (black regression line) - unlike the correct analysis by individual groups - weakens the correlation. Distinguishing between averaged and typical outcomes is particularly important for parameters with known high individuality such as thyroid hormonesBack to article page