Reproducibility of Research Results

Psychological studies are a bit like sampling appetizers at parties, they must be enjoyed for what they are, at the moment..trying to reproduce these results for use at another random time is not likely to hit the same note....that doesn't mean the memory is false, just that things are constantly changing...if you tried hard enough and long enough you probably could reproduce the results, but is it really worth the effort, in most cases not...

In any experiment there is some uncertainty due to factors that are not under control. In psychology there are a lot more uncontrolled factors than in physics. To reproduce a result entails contending with uncontrolled factors both in the original and the repeated experiment.

Psychology is half science, half art, and half voodoo.

So then it is 3/2, and all this time I thought it half-fast.

To call psychology a science is a complete fallacy and you'd have to be crazy to call it that.

Perhaps the paper by John Ioannidis falls into his claim

I am from India and have a degree in Electronics + MBA from top most colleges in India+ 45+ years industrial experience. By some chance I landed recently into an opportunity to guide PhD students more due to my solid practical experience, market oriented approach to analysis and problem solving.

I always thought in USA PhD projects were market need oriented. From my experience- all research scholars are told to look up all previous publications in IEEE and such related top journals first, make an analysis. They are then supposed to come up with some novel solution. From my experience- I have shown to these scholars, in industry- there are already better solutions published from companies like TI, ST, MICRO CHIP, ONSEMI, IR, Power Integration etc etc. These scholars do not look up patent applications and industry status. they are in a world of their own. They are guided from their PhD professors who are from the same system.

Hence most publications are minor variations from already published- with little relevance to industry. While in Industry - I think I made mistake of referring to 2 IEEE articles related to my area of work, tried implementing it on shop floor, burnt my fingers and had to go back to my own old process.

While conducting experiments at lab scale- I also find limitations due to funding etc (it is natural. I cannot blame anyone). So with limited facility, results obtained are also masked by experimental set up errors / limitations. In order to finish their work in a limited time frame, results are tampered with- hence reproduce ability may be difficult for some one else. BUT LO!!! That itself creates an opportunity for next PhD scholar to declare a great finding (even he/she may be in error again!!!).

But it is true - there is a huge gap in working style of R&D in academia and in industry. Thank GOD, I never did PhD. My friend says they have Permanent Head Damage

If it was only psychology studies that weren't reproducible, it would be all giggles.

But there's plenty more, especially in biomedical science. This is old news but....

http://www.nature.com/nrd/journal/v10/n9/full/nrd3439-c1.html

My first post used psych studies as examples but I was thinking of research in general, just like the article you've posted. If you think about it, it's not unusual to find out that, say, previous medical advice changes radically with new research. In these cases it's not necessarily the case that the initial research was flawed.

Ouch, that struck a nerve.

Medical advice and the prescribed drugs that usually accompany such advice, are no longer based on unbiased, true scientific research.

Such research was, at one time, financed by taxpayer dollars and was truly unbiased.

Then drug companies, started paying lobbyists, lawyers and politicians to reduce that funding to the point that mega drug companies rushed in to "save" medical research.

One only has to look to see who has paid for the "unbiased" research to find the true reason the advice has changed and to also find who the true benefactor is. Most often, the drug company who financed the research is the real benefactor.

My wife has been a health care administrator for over 30 years, and I speak with a little bit of first hand knowledge on the subject.

One of our adopted children was just prescribed 4 drugs for a skin condition. The combined cost of those drugs is over $2,000.00 per month.

Short lunch break here... so I can't respond with much substance. But if trials with negative results are not published - all over the news about this in recent years - then the positive trials that are the ones that get published are, in essence, not reproducible. The results have already been 'not replicated' by the trials that were withheld from public view.

A law was put in place in the US a few years ago to address that situation afaik.. Here's a recent report I picked up on a quick google.

http://www.statnews.com/2015/12/13/clinical-trials-investigation/