New research casts doubt on claims that people have 'rose-tinted glasses' and findings suggest governments should re-examine their use of 'optimism bias' in large-scale projects.