A questionnaire on the eating frequency on which the results of this study are based signifies virtually little scientifically. There can be no meaningful outcome, and it is unethical, considering the sort of impact this may have on people's behaviors, to make a conclusion based on such an inadequate study.
It is difficult to ask people to quantify their food habits over the past few years and expect significant quantitative findings. Over the past few months, one can barely define their eating habits. There are just too many factors for which no conclusion can be controlled in such a study.
It annoys me profoundly that this is a science that is reported, used as evidence for justification of recommendations, etc.
In general (save for some sub-fields) in fact, experiments (such as this) that require important statistics are not well-trained.
The law of big numbers applies in particular (which you should know about as a physicist). Isome would argue that the results of this study, notwithstanding the noise from the report, are probably far more important than a carefully controlled randomized trial with a small cohort (~100).
Self-reporting questionnaires are undoubtedly a well-established approach for these kinds of investigations, and the weaknesses have been well known.
In particular, it is suggested that over several years, many (most) persons can provide significant numeric responses to their eating habits since their patterns seldom alter. Of course, most individuals here will be able to say about how many meals they eat meat each week, how much they consume on average, how often they drink alcohol, and so on. Only approximate averages are of relevance to us.
As for your argument that too many variables cannot be monitored, fine let you hear the name at least 10 and give a good statistical case as to why they have to be monitored in sample size by ~80 000.
In this situation, "good technique" merely indicates a common approach, not necessarily a strong scientific process.
The problem is people. We are not credible witnesses, even of our own past, and, unless we are strictly dietary, most of us mostly go randomly through a small number of food categories with our food intake (even caloric intake can be wildly inconsistent over time). It encountered enough persons who utterly minimize their intake of sweets, e.g. that in their own history records they diminish the incidence of such occurrences psychologically. Alone, many of the low-carb vegans and low-carbon meat-eaters might be in the high-carb group.
The most serious shortcoming of this technique is that there is no proof that human memory gives an accurate estimate for lengthy periods of time when it comes to food consumption. If the reliability of the most basic instrument of your technique cannot be determined, then it is practically useless by definition. The mistake in self-reporting might be so significant that the genuine 80% low-carbon categorization actually is high carb. Or the vegan group may be over 50% high-carbon, etc.
We cannot determine the margins of error without a comprehensive assessment of the precision of the instrument employed in the research (FFQ in this example). However, we have several grounds to believe that the accuracy is considerably high. In this situation, adding additional members to the group does not improve the accuracy, it just increases the loudness of the noise. Is every group's True Low Carb ratio 10% or 90%? Who knows. Who knows.
If the accuracy of the tools cannot be validated, the findings are simply a random decision amongst the various conclusions. It indicates that they did not refute the null hypothesis.
Therefore, it is encouraged to advocate some form of sequestration of smaller populations with comparable genetic composition (several groups for each diet), etc. But this is largely unworkable.
Too frequently weak science may be regarded as "soft" or "applied" scientific study, regardless of its type.
In this connection, even noisy research like these at least indicates that there is no obvious major link between, for example, a low-carb diet, and death. Even brilliant research would indicate if low-carb diet folks started to die decades younger than normal. Studies like these, however small, have some utility in this way.
First of all, because the whole of science is not dependent on genetics. We know (but don't we? At least gene-induced lactose intolerance. Is it too much to imply that reactions to other foods might likewise be genetically modified? Secondly, scientific research itself has an enormous issue. There are books and counter-research on this subject. Basically, you accept a grant, study, and then achieve something if there is nothing that should be a default in your final output, well it doesn't seem good. So, if it's cherry-picked data, you create some results. Then a decade later some statisticians take the raw data and verify that have found nothing. It's always happening. Thirdly, this is particularly a medicinal problem but may also be done in nutritional research. Surely you see the advertisements where volunteers were looking? Well, who do you believe they'll get? Nobody creates research entitled "our study among homeless and abject poverty."
Somewhat dubious, maybe, but the inmates would be a wonderful site for this research. You may accomplish this just by monitoring your consumption and not changing anything in the offer. A valid intake report should be quite realistic.
Yeah, ethics would be difficult, but you MUST regulate food consumption in order to truly perform nutrition studies. Age, weight, calories, activity control, and only varies in dietary configuration to different degrees, etc.
What kind of ethical problems do you think? It would have assumed that good quality meals would attract all the volunteers you could desire, then you would find them in a typical penitentiary. There are undoubtedly other groups where delicious food would be just as tempting as university students.
A more practical problem in jails would be the high frequency of hepatitis and drug use together with all the confounding variables (such as poverty).
In the US, federally sponsored research institutes must follow specific standards in the conduct of research in human beings. A researcher was trained and said that prisoners could not be used as subjects just because they were convenient and that the two situations where they were usually studied were cases when prisoners, criminal activity, or anything inherently specific to prisoners were studied (the example given was HIV).
Why not inmates self-imposed... like a rehabilitation center. You may leave wherever you want but you don't be paid...then you pay them for as much as $5k for 1 month or 5k/month for any study period... Then there's no jail — regulated, but "at-will"
Apart from proposals and other problems, I'm not convinced you rely on food to eat. Prisons still buy food in the commissary in most areas.
If you can acquire a control prison and take care of variations, you might compare jail A against prison B quantitatively. It would be interesting to have only a 20% greater diet between A/B.
Someone already has "done science" in other jails. Way more ethical than the normal ideas that is available on the internet for inmates.
It is contemplated actually doing nutrition study, it's basically a jail, even if it is voluntary. You should be able to lock various populations from any food source and regulate their everyday activities and so on, for at least 2 years, although it is possible that after only six months you can get some results.
And remember: anything based on modern technology will not deliver results until the 2040s or 2050s.
Previous validation studies demonstrated relatively strong relationships between FFQ evaluated energy-adjusted nutrients and various NHS(21) and HPFS food records completed throughout the previous year (20).
In view of the types of impacts that this may have on people's behaviors, it is borderline unethical to make a judgment based on such inadequate research.
This particular behavioral shift would probably not be terrible, but more of a rant for these kinds of research. Poor research in this situation has real-world repercussions (nutrition research), and we need actual knowledge into what diet works best for the person.
Even after you control all that you plan to control, you still have a great confusing element in genetic make-up. Then you need prisons with thousands of identical twins who live in every manner except diet (which is meticulously recorded). You won't just receive an experiment as you can put in physics.
The objective of the author highlighting big restrictions is to admit that this is, unfortunately, the best way to deal with it.
This seems to be a very solid concept, given the enormous amount of data relating 'plant chemicals' (polyphenols, glycosinolates, etc.) to (in the majority) significantly favorable impacts on human biochemistry and physiology.
It's worth noting that "five (fruit and vegetables) a day" is a well-known slogan in the UK and it is elsewhere. But almost anybody explains why it is recommended that the needed fat, protein, carbohydrate, vitamins, and minerals may be delivered without considerable consumption of fresh fruits and vegetables (using supplements).