"White Hat Bias" - fixing results to support what you think is righteous

When we look at research in public health fields - those matters relating to lifestyle choices especially - we often see what we suspect is a search for evidence to substantiate an ideologically-predetermined position. So, for example, the ideology decides that standardised packaging would be a way to reduce the uptake of smoking among the young and this is followed by a set of studies that appear to show just how much packaging influences the decisions of young people considering taking up smoking. Except that, when we pull these studies apart, we find that they show nothing beyond the (rather obvious) fact that people prefer attractive, bright colours to drab unattractive colours.

This is "white hat bias" - a term coined by US biostatistician Professor David Allison:

It was Professor Allison and Dr Mark Cope who coined the phrase in relation to obesity research. They defined it as: “bias leading to the distortion of information in the service of what may be perceived to be righteous ends”*

 

It seems to me that this problem is endemic in evidence-gathering for the development of public policy and especially social policy. Examples would include starting with the presumption that inequality is a negative indicator of health - there isn't any consistent evidence to support this contention yet is remains an assumption in healthcare planning and, in particular, planning around public health.

 

Alongside this prejudiced approach to evidence we also find the use of 'evidence' that is not supported by the facts -  so one of Bradford Council's public health consultants presents as 'fact' that, without further intervention, hazardous drinking will increase in the City. This ignores another fact - that alcohol consumption (especially among the young) has fallen significantly. It seems odd, given this change, to make an argument stating that 'hazardous' drinking will increase unless 'something is done'. We see an intelligent, well-informed man allowing his presumptions about health to run ahead of the actual evidence simply because he believes that drinking is something to be discouraged.

This problem would be fine if it wasn't the case that lay people can be easily dismissed when they challenge this 'white hat bias'. From sweeping statements such as 'we have evidence to support this from the literature' (I got that one when I challenged some facts), to simply ignoring the challenge and falling back on the logical fallacy of 'I'm the expert so I'm right'. We also get 'Macclesfield is in Greater Manchester' answers where the 'white hat bias' allows the person to simply ignore a fact. A good example here is the fact that consumption of non-dairy extrinsic sugars (that's all the sugar we add to stuff plus honey) has fallen significantly over the past 30 years. Meaning, of course, that sugar cannot be isolated as the guilty party in rising rates of obesity.

The problem of 'white hat bias' isn't limited to public health - I've seen it in work around regeneration, local economic development and retailing - but it is that case that such bias is now so ingrained in public health research as to cast doubts on the entire corpus. As the researchers at the University of Alabama pointed out:

UAB researchers examined ways in which scientists writing new research papers referenced two studies reporting the effects of sugar-sweetened beverages on body weight. They found that less than one-third of the papers that cited the beverage studies accurately reported the overall findings, and more than two-thirds exaggerated evidence that reducing sugar-sweetened drink consumption reduced weight or obesity. The UAB researchers also found several examples in breastfeeding studies in which the authors selectively included some data and discarded other research to support the theory that breastfeeding decreases the risk of obesity.

 And the result of this is that we get public policy initiatives that are not appropriate to the problem - plain packaging for cigarettes is unlikely (indeed one year on in Australia evidence now supports this argument) to make any difference to either the take up of smoking by teenagers or the overall consumption of tobacco. Yet the bias of researchers means that the only studies considered are those limited, qualitative studies showing that kids like bright colours.

The same problem can be seen with policies to limit or control fast food takeaways despite there being little or no evidence linking fast food to obesity. However, we still get Councils pushing through planning controls (banning takeaways near schools, for example) that won't change anything. 

We went through a time when 'evidence-based policy-making' was all the rage. What we should now realise is (as those cultural studies students could have told you from the start) that scientists and researchers wedded to a particular position will be selective in their interpretation and presentation of evidence so as to provide support for that position. 

 

The researchers who coined the phrase 'white hat bias' should be commended for their efforts and for the warning. But I fear it will fall on stony ground - the ideologues of public health are simply uninterested in any evidence that questions or contradicts their prejudged policy prescriptions.

....