Sound evidence and reason for god (s) free

[ProfessorDavey]

'Objectivity is a philosophical concept of being true independently from individual subjectivity caused by perception, emotions, or imagination'

The issue is that we can't avoid the issue that our perceptions can be anything other than subjective. The reason for mentioning hard solipsism is that all external validation such as scientific tests are things we perceive as a subject.

But that isn't a relevant concept with regard to science, where objectivity is about eliminating personal bias in the conducting of scientific studies that may ultimately bias the results of that study:

'Objectivity in science is an attempt to uncover truths about the natural world by eliminating personal biases, emotions, and false beliefs. It is often linked to observation as part of the scientific method. It is thus intimately related to the aim of testability and reproducibility.'

But in reality well constructed scientific studies meet the scientific objectivity test, but arguably also meet the philosophical definition - again I think you are confusing subjective bias (which emanates from 'perception, emotions, or imagination') and error, which is not purely restricted to human observers - there is error from automated systems just as much as human operators and observers. But error isn't subjectivity unless there is a bias rather than an error.

If I want to measure 100 microlitres of a reagent as part of a scientific experiment I will want to use accurate measuring equipment - that might be operated by a human or entirely automated. Both will have error associated with them, but it only impacts on subjectivity if the operator biases the measurement in a particular manner.

[Nearly Sane]

But that isn't a relevant concept with regard to science, where objectivity is about eliminating personal bias in the conducting of scientific studies that may ultimately bias the results of that study:

'Objectivity in science is an attempt to uncover truths about the natural world by eliminating personal biases, emotions, and false beliefs. It is often linked to observation as part of the scientific method. It is thus intimately related to the aim of testability and reproducibility.'

But in reality well constructed scientific studies meet the scientific objectivity test, but arguably also meet the philosophical definition - again I think you are confusing subjective bias (which emanates from 'perception, emotions, or imagination') and error, which is not purely restricted to human observers - there is error from automated systems just as much as human operators and observers. But error isn't subjectivity unless there is a bias rather than an error.

If I want to measure 100 microlitres of a reagent as part of a scientific experiment I will want to use accurate measuring equipment - that might be operated by a human or entirely automated. Both will have error associated with them, but it only impacts on subjectivity if the operator biases the measurement in a particular manner.

It's the relevant concept on this thread as it was a metaphysical discussion of what is reality that started this part of it.

[ProfessorDavey]

Except it's not useful in a conversation about 'objective reality' which is when I first used the term.

But 'objective reality' and science go hand in hand as we typically define the former as that which is demonstrated by the latter.

[Nearly Sane]

But 'objective reality' and science go hand in hand as we typically define the former as that which is demonstrated by the latter.

No, because objective reality is a metaphysical concept.

[ProfessorDavey]

No, because objective reality is a metaphysical concept.

And how do you test that reality, to actually be ... err ... reality. I think that would be via science.

But the broader point is that if you claim that science is not objective you need to use the appropriate use of objective within a scientific concept - in other words the accepted definition of objectivity in science.

[Nearly Sane]

And how do you test that reality, to actually be ... err ... reality. I think that would be via science.

But the broader point is that if you claim that science is not objective you need to use the appropriate use of objective within a scientific concept - in other words the accepted definition of objectivity in science.

Why would I need to use the meaning of a term in a metaphysical discussion that it's inappropriate for?

[bluehillside Retd.]

NS

I am just using their posts about reality. I am uninterested in the problems of 'some theists'. Don't you think that the idea that science is objective as PD and jp seem to think is hugely problematic?

Yes, in the sense that the theoretical and the certainly actual are at odds. At an ontological level science gives us a reality, but there’s no way to know whether that’s just what you get when science looks in the only places it’s capable of looking. At a practical level too science is done by subjective beings, and no matter how much it seeks to eliminate our biases there’s no way to know for sure that it actually does that – in other words it’s ideal in principle can be misapplied in practice.     

PD

But that doesn't mean that science, in its theoretical state, isn't objective - merely that scientist are sometime not able to meet that goal.

I’m not so confident about that. Axiomatically science even in its theoretical state can only investigate that which its tools and methods are capable of investigating. How would we know definitively that there isn’t a reality that’s beyond the reach of science to investigate? The best we can say I think is something like, “science is a method that seeks to eliminate biases in its practice to provide understandings of reality that most align with the available evidence, validated by intersubjective experience”.     

But the point about science and the scientific method is that it firstly recognises those issues and goes out of its way to mitigate against and eliminate any subjective interference in the objectivity of scientific data. So firstly, wherever possible the collection of data will be automated, removing human subjectivity.

Yes I know, but the key phrases there are “goes out of its way” and “wherever possible”. I couldn’t agree more with the sentiment, but “objective” then has to mean “subject to science’s ability to do these things” and not “absolute”.   

Secondly reproducibility - the key element of scientific that data are only valid where they are reproducible in another setting and 'another scientists hand' - this is effectively about eliminating the individual subjectivity. If many scientists, working independently are able to reproduce the data then we move toward true objectivity.

Except all scientists are people, and people tend to have the same biases. This reminds me of a TV programme about Airbus when an employee proudly explained that the software for the fly-by-wire controls had two back ups, and that each version was written by a different team of programmers. That way, even if the first system failed either system two or even system three would kick in to save the day. They then cut to a software guru at MIT (I think) who said something like, “yes, but essentially all software degrees are the same and so programmers are basically trained in the same way. This means that, in the unlikely event that a deep fault lies in system one, chances are it’ll exist in the other systems too”. I agree that reproducibility goes a long way toward eliminating bias but that’s all it can do – go a long way toward it.     

Thirdly, and linked to the second, an understanding of variability in data - and in the most appropriately designed studies, the ability to undercover the source of that variability, be it inherent in the data (and therefore objective) or an artefact of the experiment, whether due to human variability or inherent in the method.

And yet there are cases – admittedly rare, but cases nonetheless – in which wrong answers have been arrived at nonetheless. There’s a case for example in 1950s America where babies were found to have enlarged oesophagi so they zapped them with radiation to shrink them. Decades later thousands of the patients died needlessly of throat cancer – turns out the baby oesophagi weren’t enlarged at all. The data for the typical size went back to Victorian studies, when the bodies the anatomists used were regularly provided by grave robbers. Wealthy cemeteries were properly protected, but the poor were buried in mass graves that were easier to plunder. Poor people were often malnourished, one indicator of which is – an enlarged oesophagus! In short, the data the 1950s doctors relied on was corrupted, and led to a disastrous outcome despite following all the provisions you’ve set out, but necessarily only to the best of their ability.           

So although you may argue that the collection of scientific data is not truly objective in practice,…

Not quite – it may by chance be “truly" objective, but cannot knowably be so.

… it is in theory and the method itself is designed to eliminate as much subjectivity as possible to drive it as near as possible to true objectivity. As such I think it is perfectly valid to describe science as objective. Not to do so simply lumps it with other approaches that never aim at objectivity nor classify subjectivity as a flaw.

I don’t think it does. “Objective” here implies an absolute but it’s more of a spectrum I think. I’m content to say that science produces results that are more objective than, say, the claims of religion but we can’t know what’s further along the same axis.

NS

And you are quite simply wrong. As bhs has already pointed out we are by definition subjective beings, and what science does is use our ability to communicate to get an agreed intersubjective position. But since we can never be sure that any of our own experiences are real  including that anyone else exists then all of those experiences are questionable.

Objectivity is an absolute. Given we are subjective, we cannot achieve it.

In day to day discussion, it doesn't really matter. In that sense it is rather like the free will discussion. We go about pur lives acting as if there is such a thing as free will but at base it makes no sense. So with science we go about it as if it can achieve objectivity,  but given our restrictions that makes no sense either.

That doesn't devalue the importance and value of science from our quotidian experience.

Fully agree.

PD,

You are confusing bias with error.

For scientific objectivity the method needs to eliminate bias (which is a manifestation of subjectivity if that bias is human) - it does not, and ultimately cannot eliminate error. The method will need to minimise and understand error, but if there error remains, but no bias then the method and the data meet the standards of objectivity.

A well designed double blinded clinical trial will be free from bias - it can be riddled with error which would mean it would be poor science, but provided there is no bias it remains objective - rubbish, but objective none the less.

“A well designed double blinded clinical trial will be free from bias…” how so when, for example, the selection of the data to look at in the first place is done by subjective beings? Would the oesophagus case I mentioned above have passed a double blind clinical trial given that the basic data on which each trial would have relied was equally corrupted?   

PD,

But that isn't a relevant concept with regard to science, where objectivity is about eliminating personal bias in the conducting of scientific studies that may ultimately bias the results of that study:

'Objectivity in science is an attempt to uncover truths about the natural world by eliminating personal biases, emotions, and false beliefs. It is often linked to observation as part of the scientific method. It is thus intimately related to the aim of testability and reproducibility.'

“Attempt”, yes – but that’s all it can be: an attempt. You cannot therefore say “will be free from bias” when what you actually mean is “will attempt to be free from bias”.

But in reality well constructed scientific studies meet the scientific objectivity test,…

Yes.

… but arguably also meet the philosophical definition…

How so?

- again I think you are confusing subjective bias (which emanates from 'perception, emotions, or imagination') and error, which is not purely restricted to human observers - there is error from automated systems just as much as human operators and observers. But error isn't subjectivity unless there is a bias rather than an error.

What if the “error” is for example in a diagnostic algorithm on which an automated system relies?

If I want to measure 100 microlitres of a reagent as part of a scientific experiment I will want to use accurate measuring equipment - that might be operated by a human or entirely automated. Both will have error associated with them, but it only impacts on subjectivity if the operator biases the measurement in a particular manner.

How do you now that one in a bajillion times the software running the measuring equipment won’t produce a rogue result?

[ProfessorDavey]

How do you now that one in a bajillion times the software running the measuring equipment won’t produce a rogue result?

You don't.

But if the automated system does that this is an indication of unreliability, inaccuracy or error. None of those are indications of bias or subjectivity in that measurement.

And humans likewise are at times unreliable, inaccurate or error prone - again even if these erroneous results are produced by a human that isn't an indication of subjectivity or bias.

Science can be conducted in an entirely objective manner even if there are sources of error or inaccuracy in the measurement. Sure you want to iron out those errors as you will improve the reliability of the data, but not its objectivity.

[ProfessorDavey]

“A well designed double blinded clinical trial will be free from bias…” how so when, for example, the selection of the data to look at in the first place is done by subjective beings? Would the oesophagus case I mentioned above have passed a double blind clinical trial given that the basic data on which each trial would have relied was equally corrupted?

No - you are misunderstanding the situation.

A double-blind clinical trial doesn't lose objectivity if it involves humans (subjective beings) - no, it loses objectivity if those humans subjectively apply subjective bias towards the obtaining of data. And that's why double blinding is important - it means that neither the patients nor the people involving in data acquisition know which arm of the clinical trial each research subject belongs to. This avoids a situation where either the patient reports better outcome because they know they are on the 'new treatment' or the scientist consciously or unconsciously slants the data in favour of one outcome. They cannot do that as they have no idea whether sample X38485.2 comes from a patient taking the new drug or taking the old drug.

They may be crap at their job and therefore make all sorts of errors in their measurement of sample but that isn't subjective bias - just a useless operator.

[ProfessorDavey]

And yet there are cases – admittedly rare, but cases nonetheless – in which wrong answers have been arrived at nonetheless. There’s a case for example in 1950s America where babies were found to have enlarged oesophagi so they zapped them with radiation to shrink them. Decades later thousands of the patients died needlessly of throat cancer – turns out the baby oesophagi weren’t enlarged at all. The data for the typical size went back to Victorian studies, when the bodies the anatomists used were regularly provided by grave robbers. Wealthy cemeteries were properly protected, but the poor were buried in mass graves that were easier to plunder. Poor people were often malnourished, one indicator of which is – an enlarged oesophagus! In short, the data the 1950s doctors relied on was corrupted, and led to a disastrous outcome despite following all the provisions you’ve set out, but necessarily only to the best of their ability.

I don't know the details of this case - but will look it up, but a few comments on the basis of what was written.

Firstly it would appear that the radiation treatment wasn't scientific research at all but treatment, and the failure here is not really 'science' but ethical and on safety grounds. Effectively to allow experimental treatment on the basis of insufficient scientific research.

Secondly on the earlier victorian study (which does look to be research, albeit performed well before the embedding of clears professionalism and objective standardisation in research) - it may indeed be the case that incorrect conclusions were drawn from the data, but that doesn't mean that there was necessarily subjective bias in data collection (the scientific method itself). For there to have been subjective bias the data collectors would have needed (consciously or unconsciously) to have skewed their data to obtain a desired result. Again from what I can see - this is a case of poorly controlled science (not recognising that the subset of samples available for analysis were not representative of the overall population) rather than non objective science. It would be non objective if victorian scientists measured the oesophageal tissue of (for example) children of different races and had a racial bias - and in doing so measured the samples while knowing which race they came from and introduced biases in their data collection to support a view of differences between, say black and white children.

[bluehillside Retd.]

Hi Prof,

You don't.

But if the automated system does that this is an indication of unreliability, inaccuracy or error. None of those are indications of bias or subjectivity in that measurement.

No, but it could be an example of any of these things in the people who designed or made the measuring machine. That’s the point: the machine will only measure “objectively” – ie, accurately - insofar as it’s capable of doing that. You can use the machine while observing the most meticulous protocols to avoid bias etc, but still you cannot thereby guarantee that your answer will be objectively correct.   

And humans likewise are at times unreliable, inaccurate or error prone - again even if these erroneous results are produced by a human that isn't an indication of subjectivity or bias.

…or incompetence etc but in any case, why isn’t it?

Science can be conducted in an entirely objective manner even if there are sources of error or inaccuracy in the measurement. Sure you want to iron out those errors as you will improve the reliability of the data, but not its objectivity.

I just don’t see how you get to “entirely objective”. I understand perfectly well that you can incorporate into its methods every conceivable protocol to avoid bias, mistake etc as best you can but still that can give you only a good enough rather than certainty. Another example: in Los Angeles county IQ tests were anonymised so the markers couldn’t draw conclusions from the identities of the candidates (ethnic names, impoverished locales etc). That is, the test was “entirely objective” as you put it, yet still it produced statistically significant variations according to ethnicity in particular. How comes? Were the black kids especially just less able? Nope – turned out that questions asked were culturally skewed: when asked “what have water and salt got in common?” for example the white kids had been taught that they were both chemical compounds (2 points) whereas the black kids’ educational experience meant they were more likely to come up with answers like “they’re both found in the ocean” (0 points). Both answers are correct, but only the first one aligned with the cultural expectations of the question setters and of one portion alone of the candidate group. That is the IQ tests were actually tests of acquired knowledge, not of reasoning ability.   

The point here is that it’s still possible to have tests designed specifically to eliminate bias, but you can never be certain that you actually have eliminated bias for bigger picture reasons.

No - you are misunderstanding the situation.

A double-blind clinical trial doesn't lose objectivity if it involves humans (subjective beings) - no, it loses objectivity if those humans subjectively apply subjective bias towards the obtaining of data. And that's why double blinding is important - it means that neither the patients nor the people involving in data acquisition know which arm of the clinical trial each research subject belongs to. This avoids a situation where either the patient reports better outcome because they know they are on the 'new treatment' or the scientist consciously or unconsciously slants the data in favour of one outcome. They cannot do that as they have no idea whether sample X38485.2 comes from a patient taking the new drug or taking the old drug.

They may be crap at their job and therefore make all sorts of errors in their measurement of sample but that isn't subjective bias - just a useless operator.

Yes I know, but in the “oesophagus” case (it was actually the thymus gland – sorry, I’d misremembered) for example a double blind test would have given half the babies the radiation and half a placebo, then compared the results. And sure enough the result would have been that the radiation treatment alone had the desired effect because the thymuses (thymi?) shrank. The point though is that the test would have been asking entirely the wrong question – which should have been, “is the data we’re using for this test accurate and reliable in the first place”? It’s a double whammy too – armed with the results for the test the researchers would have had even more confidence: “look, we know radiation is the thing to do because we’ve done a double blind test that proves it works!” In other words the very test intended to achieve an objective plan of action would actually have compounded the certainty about completely the wrong plan of action.   

I don't know the details of this case - but will look it up, but a few comments on the basis of what was written.

It’s described in a very good episode of Radiolab (part 3 – “How to Cure What Ails You” – 11.05):

https://www.wnycstudios.org/podcasts/radiolab/episodes/91662-diagnosis

Firstly it would appear that the radiation treatment wasn't scientific research at all but treatment, and the failure here is not really 'science' but ethical and on safety grounds. Effectively to allow experimental treatment on the basis of insufficient scientific research.

Secondly on the earlier victorian study (which does look to be research, albeit performed well before the embedding of clears professionalism and objective standardisation in research) - it may indeed be the case that incorrect conclusions were drawn from the data, but that doesn't mean that there was necessarily subjective bias in data collection (the scientific method itself). For there to have been subjective bias the data collectors would have needed (consciously or unconsciously) to have skewed their data to obtain a desired result. Again from what I can see - this is a case of oppression science (not recognising that the subset of samples available for analysis were not representative of the overall population) rather than non objective science. It would be non objective if victorian scientists measured the oesophageal tissue of (for example) children of different races and had a racial bias - and in doing so measured the samples while knowing which race they came from and introduced biases in their data collection to support a view of differences between, say black and white children.

But the point here is that scientific procedures like double blind tests don’t exist in isolation. They happen in a context, and contextual factors influence them. Thus you might be reinforced in you convictions about irradiating infantile thymuses precisely because you’d conducted a double blind trial. And taken in isolation, sure enough the trial confirms that irradiating is the right way to go!

In short, I understand very well what science tries to do and how it does it, but science happens in a cultural, societal, educational etc context and you cannot just wrench it out of that and assume that it thereby gives you “real” objectivity. Sure it give you a lot more of it than anything else we know of, but you need to see the bigger picture too if your claims for its objectivity aren't to overreach.         

[ProfessorDavey]

I just don’t see how you get to “entirely objective”. I understand perfectly well that you can incorporate into its methods every conceivable protocol to avoid bias, mistake etc as best you can but still that can give you only a good enough rather than certainty. Another example: in Los Angeles county IQ tests were anonymised so the markers couldn’t draw conclusions from the identities of the candidates (ethnic names, impoverished locales etc). That is, the test was “entirely objective” as you put it, yet still it produced statistically significant variations according to ethnicity in particular. How comes? Were the black kids especially just less able? Nope – turned out that questions asked were culturally skewed: when asked “what have water and salt got in common?” for example the white kids had been taught that they were both chemical compounds (2 points) whereas the black kids’ educational experience meant they were more likely to come up with answers like “they’re both found in the ocean” (0 points). Both answers are correct, but only the first one aligned with the cultural expectations of the question setters and of one portion alone of the candidate group. That is the IQ tests were actually tests of acquired knowledge, not of reasoning ability.

Beyond the notion that these types of sociology-type studies necessary become highly qualitative and therefore subjective, I think you are missing the point.

What the science says here is very simply - that there is an association between the results in a particular test and ethnicity in the population sample tested. Nothing more, nothing less. And that can be completely objective - the interpretation of the science and its use are another matter. That may be subjective but that doesn't mean the science isn't objective.

And taken in isolation, sure enough the trial confirms that irradiating is the right way to go!

No it doesn't - in isolation all the science say is that irradiation results in a reduction in the size of the thymus - it tells us nothing about whether this is the right way to go. That is a clinical decision/judgment, which is subjective but the science (provided there is no subjective bias in conducting that study) will tell us, with objectivity, whether irradiation (at a particular level in the study) results in a change in the size of the thymus in the sample of people tested.

I think you are confusing the science itself (which is purely about answering a tightly defined scientific question) and the interpretation and application of those results, for example by clinicians, engineers - or in your first example perhaps racists!

[bluehillside Retd.]

Hi Prof,

Beyond the notion that these types of sociology-type studies necessary become highly qualitative and therefore subjective, I think you are missing the point.

What the science says here is very simply - that there is an association between the results in a particular test and ethnicity in the population sample tested. Nothing more, nothing less. And that can be completely objective - the interpretation of the science and its use are another matter. That may be subjective but that doesn't mean the science isn't objective.

With respect, I think you are. What the “science” (ie, anonymised testing) actually said in real life was that there was a clear correlation between ethnicity and outcomes. The tests were anonymised, so they satisfied your test of objectivity right?

The problem though was that these tests, while conducted correctly according to the rules to eliminate bias, actually contained significant bias because of the context in which they were constructed rather than the way they were marked – that is, the question setters shared cultural biases with one sector of the candidates, but not with the other. No matter how anonymised and bias free the marking, the tests themselves were a rigged game. Science happens within a context, and the context can affect its results: that’s the point.       

No it doesn't - in isolation all the science say is that irradiation results in a reduction in the size of the thymus - it tells us nothing about whether this is the right way to go.

Yes it did! The tests results were informed by fallible, bias-prone people and the results were adopted by fallible, bias-prone people. And because the tests asked the wrong question, they drew the wrong conclusion from the results. It’s no good saying, “but within the confines of a perfect isolated test there was nothing wrong with the results”. I know that, but you’re a bit like the surgeon who said, “The operation went perfectly. The patient died, but the operation went perfectly…”     

That is a clinical decision/judgment, which is subjective but the science (provided there is no subjective bias in conducting that study) will tell us, with objectivity, whether irradiation (at a particular level in the study) results in a change in the size of the thymus in the sample of people tested.

Yep, but see above.

I think you are confusing the science itself (which is purely about answering a tightly defined scientific question) and the interpretation and application of those results, for example by clinicians, engineers - or in your first example perhaps racists!

I’m not confusing it at all – I’m saying that’s what actually happens! Yes if in some way you could conduct a science experiment in a hermetically sealed bubble in which you could absolutely guarantee the accuracy and completeness of its inputs, and you could absolutely guarantee the use and application of its results then perhaps you could argue no bias. How though could such a thing be?   

[ProfessorDavey]

Science happens within a context, and the context can affect its results: that’s the point.

No the context may inform the type of research that is conducted, but it doesn't affect the results, which are independent of context. For the context to affect the results you'd have to argue that were the test to be conducted identically today (exactly the same test, exact same cohort of subjects) that you'd get a different result. You wouldn't - you'd get the same result even though we are (hopefully) far more culturally sensitive to ethnic diversity than was the case back then. But the results would be the same.     

Yes it did! The tests results were informed by fallible, bias-prone people and the results were adopted by fallible, bias-prone people. And because the tests asked the wrong question, they drew the wrong conclusion from the results. It’s no good saying, “but within the confines of a perfect isolated test there was nothing wrong with the results”. I know that, but you’re a bit like the surgeon who said, “The operation went perfectly. The patient died, but the operation went perfectly…”     

Sorry, once again you are confused - this time confusing research and treatment - a common mistake in clinical research where it is often called therapeutic misconception.

The purpose of research is to answer a question and in the case of clinical research that research is conducted on people. Clinical treatment is to provide the best clinical outcome for a patient - research cannot guarantee that as, by definition, you are answering a question the answer to which isn't currently known. Sure this is an extreme example, but in any given clinical trial there will be one arm of that trial who do not, again by definition, get the best treatment.

How we deal with this is via ethical review - that's what determines whether the trial is ethically acceptable - in other words whether the question is valid enough, the potential results important enough to allow the trail members to be subject to potential risk. Sure previous science informs that ethical debate, but the science in itself is purely and objectively answering the question. Whether we should ask that question and what we do with the results are different societal matters.

[bluehillside Retd.]

Prof,

No the context may inform the type of research that is conducted, but it doesn't affect the results, which are independent of context.

What if the context is that bad data informs the research – wouldn’t that context affect the quality of the result?

For the context to affect the results you'd have to argue that were the test to be conducted identically today (exactly the same test, exact same cohort of subjects) that you'd get a different result. You wouldn't - you'd get the same result even though we are (hopefully) far more culturally sensitive to ethnic diversity than was the case back then. But the results would be the same.

I wouldn’t have to do that at all. Conceptually at least the same data in would produce the same marks out yes, but the point is that the test scores didn’t map to objective reality at all even though the marking was perfectly conducted. That is, they eliminated bias in the marking but failed to eliminate it in the tests themselves. That’s the point – the marking wasn’t independent of the context in which it was conducted. How could it have been?       

Sorry, once again you are confused - this time confusing research and treatment - a common mistake in clinical research where it is often called therapeutic misconception.

No I’m not. I’m explaining that however perfect, bias free etc the conduct of a test it cannot exist in isolation – it has to connect with the world in its inputs and in its outputs. And both those things are variables the test itself cannot control.

The purpose of research is to answer a question and in the case of clinical research that research is conducted on people. Clinical treatment is to provide the best clinical outcome for a patient - research cannot guarantee that as, by definition, you are answering a question the answer to which isn't currently known. Sure this is an extreme example, but in any given clinical trial there will be one arm of that trial who do not, again by definition, get the best treatment.

Yes I know! But the research itself cannot decide whether the question is the right one to ask, the data it examines is accurate or complete, the outcome will be used appropriately etc. In a universe in which all those things (and more) could be rendered perfect, then yes – with the proper protocols the science in the middle of it could be said to be an “objective” window into reality. That’s not the real world though - the real world is contextualised, and science sits within its context. Much as you may want it to be guaranteed to map to an objective reality, there are simply too many variables outwith its control to be certain that it does. That’s why truth is probabilistic – science produce results that hang together as provisional truths, but that’s as much as can be said about them.     

How we deal with this is via ethical review - that's what determines whether the trial is ethically acceptable - in other words whether the question is valid enough, the potential results important enough to allow the trail members to be subject to potential risk. Sure previous science informs that ethical debate, but the science in itself is purely and objectively answering the question. Whether we should ask that question and what we do with the results are different societal matters.

They may be “different societal matters” but science cannot escape them nonetheless. You can have a many ethical committee hearings as you like but you know well that ethics change over time – and societally. Thus an ethics committee may have concluded answer A twenty years ago and answer B today – whence then the objective truth you claim about “whether the question is valid enough, the potential results important enough” etc?     

[ProfessorDavey]

Prof,

What if the context is that bad data informs the research – wouldn’t that context affect the quality of the result?

I wouldn’t have to do that at all. Conceptually at least the same data in would produce the same marks out yes, but the point is that the test scores didn’t map to objective reality at all even though the marking was perfectly conducted. That is, they eliminated bias in the marking but failed to eliminate it in the tests themselves. That’s the point – the marking wasn’t independent of the context in which it was conducted. How could it have been?       

No I’m not. I’m explaining that however perfect, bias free etc the conduct of a test it cannot exist in isolation – it has to connect with the world in its inputs and in its outputs. And both those things are variables the test itself cannot control.

Yes I know! But the research itself cannot decide whether the question is the right one to ask, the data it examines is accurate or complete, the outcome will be used appropriately etc. In a universe in which all those things (and more) could be rendered perfect, then yes – with the proper protocols the science in the middle of it could be said to be an “objective” window into reality. That’s not the real world though - the real world is contextualised, and science sits within its context. Much as you may want it to be guaranteed to map to an objective reality, there are simply too many variables outwith its control to be certain that it does. That’s why truth is probabilistic – science produce results that hang together as provisional truths, but that’s as much as can be said about them.     

They may be “different societal matters” but science cannot escape them nonetheless. You can have a many ethical committee hearings as you like but you know well that ethics change over time – and societally. Thus an ethics committee may have concluded answer A twenty years ago and answer B today – whence then the objective truth you claim about “whether the question is valid enough, the potential results important enough” etc?   

None of which changes the situation that science and the scientific method - the design of studies that can test a hypothesis through the generation of data with those studies being repeatable and therefore independent of individual experimenter bias - is inherently objective.

Sure society can ban or allow certain scientific questions to be answered, but if they are being answered by science then that process must be objective. Societies may like or not like the answers, and may interpret the data to fit their prejudices, but again that doesn't alter that fact of objectivity in that data generation, provided the science is done properly.

[bluehillside Retd.]

Prof,

None of which changes the situation that science and the scientific method - the design of studies that can test a hypothesis through the generation of data with those studies being repeatable and therefore independent of individual experimenter bias - is inherently objective.

If by “inherently objective” you mean something like, “in its methods does its best to eliminate the subjective” then I agree with you. That’s a narrow description of “objective” though, and it’s that very narrowness that means you can’t assume that science necessarily maps to an objective reality. I agree that it’s the most verifiable method we’ve come up with for doing that, but I don’t see how you can go further than that.

Sure society can ban or allow certain scientific questions to be answered, but if they are being answered by science then that process must be objective. Societies may like or not like the answers, and may interpret the data to fit their prejudices, but again that doesn't alter that fact of objectivity in that data generation, provided the science is done properly.

But you referenced ethics committees as the arbiters of “whether the question is valid enough, the potential results important enough to allow the trail members to be subject to potential risk”. What is an ethics committee if not something fundamentally rooted in its societal time and place? Would a Taliban-era ethics committee arrive at the same answers to these questions as a liberal Dutch ethics committee for example?

I understand entirely that once you’ve decided what to investigate, what data should be used, whether or not it’s accurate or complete etc then the piece in the middle called the scientific method is designed to be as bias-free as possible. So what though? 

Consider our surgeon. The ethics committee approved the operation, the anaesthetic worked like a dream (ha ha), every stitch was properly in place, all the instruments were removed and accounted for. It was objectively perfect. Yet the patient died. Maybe someone forgot to note that he was allergic to the anaesthetic, maybe his records were mixed up with someone else’s, maybe anything. The “sciency” bit was in other words absolutely, objectively spot on. But it was still the wrong thing to do because of…yup, contextual factors outside its relatively narrow remit. That’s the point: science can only “do” objective within its own confines, but its own confines are effectively meaningless because there’s no such thing as science in isolation from the world.               

[SusanDoris]

I have not been following this topic at all, but decided to take a look at the latest page now. As a result, I have had the pleasure of reading through a most interesting set of posts, for which I thank you.