Note on bias, prediction and utility from a Critical Realist perspective.

We all live by strong biases, egocentric, in the sense that we have only a first person perspective from our own standpoint (assumption 1 & 2 and there are plenty more where those came from! Also, egocentric is used without judgemental value in this context). Naturally, we can observe and evaluate, sympathise and empathise with others but my argument does not revolve around the observer's perspective. Rather, it is about the 'data' given, that which is worked from, from the egocentric individuals' perspective. In real life (IRL for my fellow computer-geeks), we assume, interpret and justify actions of others and oneself (most oftenly) in response to situations, actions and relations. While I could turn this OP into a discussion of awareness and applied criticality to one's preceding actions in a specific situation which would affect it and account for parts of one's response-actions -it is beyond the scope of this specific piece (nevertheless very interesting). The issue is that we are not good with assumptions of others' behaviour -even folk psychology requires more than one perspective to account for a situation and these are most often generalisations beyond justification. We are subject to environmental influences in decision making, intraindividual bias both from biological limitations to attention, interpretation and retention and from previous experience, anecdotes, interindividual bias from differing norms depending on who you are interacting with and lastly a cognitive bias in which we reinterpret, change, remove and add information in retrospect to a given situation (this is, by the way, not an exhaustive list of our phallacies). Psychological research on the subject of bias is overwhelmingly in favour of our poor ability to accurately account for situations and others' part in it, let alone our own involvement in the situation (excuse the lack of references, what I have in mind is mostly eye-witness and implicit priming research).

Indeed it is a central reason as to why we are told to write dispassionately and technically when presenting our experiments. It has several important consequences; the first is that it forces one to distance oneself from one's own interpretation (if one is given) and secondly, the observer (the reader) is given a value-free account of someone's interpretation of their data. This contributes to allow the viewing of an experiment critically, without having to be bound to criticise the discourse itself for bias. An example would be selective portrayal of results, choosing to present only results supportive of one's hypothesis when there are other parts of it that would speak against or support a null hypothesis; this is a bias not controlled for by writing dispassionately, but, it becomes much clearer to the reader if this is what the author has done. It is thus a safeguard, but does not guarantee, against portraying instead of presenting and is the best way we have been able to come up with. So far.

This is to me reason alone why reliance on accounts from only a few individuals is worth very little as basis of understanding for social phenomena, in scientific discourse. It does have worth -and can have worth to others- but all too often the consequences of which are overlooked or not mapped out. For my own private empathy/sympathy register, it is fascinating to be allowed entry into an individual's world -a perspective one would not have unless one was either a member of the same group or the individual oneself. It fosters understanding for something I previously dismissed or was ignorant of. One gains a larger perspective -but(!)- only in the realm of one's own assumptions and cognising when observing social phenomena from one's individual, private/personal, view of the world. It also adds another bias to one's framework; new assumptions are made upon the limited perspective of this, one, or a few individuals -which does not necessarily account for other "similar" individuals.

Understanding social phenomena is a different thing when we are discussing the basic scientific principle of prediction (and its utility). Scientific research demands predictability and an unspoken assumption is for it to have utility, i.e. the consequences of any prediction must aim to explicate and elucidate a proposed relationship -it must hold without our intervention/involvement/observation of it. This prediction is then evaluated for its utility by how well (or not) the proposed relationship predicts a future social phenomena.

We assume there is a mechanism by which an observed action is caused but we are well aware of that, because of our biases and limitations, we may perhaps only come as close as these biases and limitations allow us. We have however still an excellent way of understanding how close we actually are to explaining 'what actually happens' because of the consequences of our prediction's implementation. If X then Y. X. Y? Y'? Y''? Z? While social phenomena are far from as simplistic as an X and Y relation, social scientists are creative in testing these predictions -sometimes they lead us down a garden path, sometimes they reveal more or less solid predictions. The point is, of all the assumptions made in this OP, being aware of one's shortcomings is a crucial way of both understanding why one's research showed the results it did (regardless if it fits or not with what one _wants_ to find), communicating a _needed_ skeptic perspective, guides one to see potential bias in others' research as well as leaves one open to (and happy for(!)) criticism from others. We cannot afford to 'believe in' our theories and hypotheses, they introduce a bias that can be veiled by scientific discourse and very concretely limits a potentially wider/better (in regard of predictability) perspective of a social phenomena.