critical analysis of a scientific paper...

[-Savage-]

Has anyone done this before? I am a bit stuck as to what I need to do here.

I am going through it but don't really know what I am supposed to be criticising... it all seems fine lol.

Could anyone give me pointers?

For those interested paper is titled - Delayed diagnosis of females with respiratory presentation of cystic fibrosis did not segregate with poorer clinical outcome. - and is attached
____________________
My Bikes - - - Royal Enfield 500 with sidecar, VFR800, CBR600FX,
Peugeot Speedfight 100, Kymco Hipster 125, GSXR750K5, Jawa TS 350

[Whosthedaddy]

Find other articles relating to similar findings, other research into the same area / topic, audit data and then argue to which one is more accurate in your opinion.

Bit hard to explain in short but quite 'simple' in reality.
____________________
Current : MSX 125 Past : CBR 900RR Monkeybike : c50 LAC : ZXR750 H2 : FZR600 : ZX7R P3 : YW100 : TRX850: Trophy 900 T309 : GSXR 600 L0: Monkeybike : XJ6S Whosthedaddy

[krytenboog]

The book "bad science " by Ben Goldacre is a great intro into looking at scientific evidence from a critical perspective and would give you a starting point in terms of how to start critically examining a piece of research.

If you struggle getting a copy you can have mine for postage costs, just pm me if so

[james1988]

[hellkat]

I should say that book sounds like it would be interesting applied to any scientific/geeky research field.

Tempted to even buy a copy myself.
____________________
Not nearly as interesting in real life.

[Frost]

Despite what it claims in the title, there is little actual science in computer science, it's mostly just discrete maths. Real science uses whatever concept of theory best matches what can be observed, comp sci regularly uses something inferior or just plain stupid as it is what has caught on and is most popular, this is a key difference.

The main thing to criticise a paper on is the direction that they took with the research. Have they left any unanswered questions? made any assumptions without supporting them? have they provided sufficient references for their sources? Is the structure okay with regards to abstract, content and summary of findings? Have they covered something that has already been covered? Have they used a sufficient number of diverse references? Have they used appropriate analytical techniques, or just rehashed ones that they know over and over?

Just remember that no paper ever has been or ever will be perfect. You can always criticise something about anything, just pick the criticisms most relevant. If you struggle to find any relevant criticisms, it's probably a good paper.

[GhostRider]

Usually you can get anal about their statistics and the way they have calculated them, but then you need to have a firm grasp on statistics yourself in order to do so, and that's a bloody headache waiting to happen, I still haVe nightmares....

I believe this is a good text on the subject, haven't read it myself but I've seen the medical doctors I work with thumbing through it and im pretty sure it's recommended in the 'Bad Science' book aswell:https://books.google.co.uk/books?id=_5a0UyIOx_MC&dq=how+to+read+a+paper&printsec=frontcover&source=bn&hl=en

GhostRider
____________________
I have all the characteristics of a human being: blood, flesh, skin, hair; but not a single, clear, identifiable emotion, except for greed and disgust. Something horrible is happening inside of me and I don't know why. My nightly bloodlust has overflown into my days. I feel lethal, on the verge of frenzy. I think my mask of sanity is about to slip.

[DrDonnyBrago]

Every paper has certain weaknesses and strengths, it's just a matter of pointing out the issues in the paper you are looking at.

For example, they could have a small smaple size, use methods not universally accepted or widely used, misinterpret results or produce data that conflicts with other sources.

[hellkat]

Most papers I've ever had a hand in (usually from an editing point of view, rather than actual input/critique) - what I've noticed is that they pick a specific thing, like study size, target population, particular disease or subject matter, and then go on to suggest that a larger study/more centres/closer targetting needs to be studied.

So basically the conclusion always says "There have been some answers in this field, but there need to be more" ... ergo, we need extra money to fund research so please give us some, this study is terribly worthwhile.

That's your bottom line: more research needs to be done (funded) in this field.

Or you can extrapolate it out into whatever field you are interested in, thus:

"Well, we think these results are good, and the study is well made, good numbers, excellent results, but it doesn't study [precisely what we are interested in being funded for]."
____________________
Not nearly as interesting in real life.