The FastDay Forum

This is interesting (particularly for "Overweight" category males):

From the progress tracker forum stats...

4:3 VERSUS 5:2 Median loss Values lbs/week (Males)

males 4:3 overweight = 2.21
males 5:2 overweight = 1.45

4:3 Advantage = 0.76 lbs/week

males 4:3 obese = 1.87
males 5:2 obese = 1.56

4:3 Advantage = o.31 lbs/week

Conclusion:

There appears a significant advantage in 4:3 for males that are in the overweight category, but less so if obese. The advantage of nearly 50% further weekly loss appears directly proportional to the additional fast-day in the overweight category.

[carorees]

4:3 overweight men n=5
5:2 overweight men n=48

n = 5 is so small, I think we need a lot more 4:3 data before we could really say that it gives an advantage. Even though it ought to!

"n = 5"

None the less, interesting though! Particularly as we are all here as a result of Dr Ms n=1. I agree it will be interesting to see how it develops.

But there is an even more interesting 4:3/5:2 trend

[carorees]

I completely agree re MM's n=1!

What is the more interesting trend? Am I being dense?

"What is the more interesting trend?"

Depends whether n=22 is significant enough?

C'mon Lastchance spill the beans. I'm dying to know!

Faster=better? I question that assumption. To me long-term sustainability is more important than speed of weight loss. With 5:2 I'm gradually losing weight without seriously impacting my lifestyle. Why do I need to add another fast day?

"What is the more interesting trend?"

Depends whether n=22 is significant enough?

Since the sample wasn't randomized and sample differences are not accounted for, probably not But it might be fun!

OK,

I thought this was intersting:

Female 5:2 overweight = 0.99 n=505
Female 4:3 overweight = 0.79 n=22

4:3 'Advantage' = -0.20 lbs/week

i.e. Ladies in the overweight category may be 25% better off if doing 4:3 to change to 5:2???

So the question is am I seeing a male/female pattern here in the progress tracker data?

[carorees]

Too few data on 4:3 to know I reckon, especially the men. For the women I would think we need at least 100 (figure pulled out of thin air) to be confident that the difference is 'real'. PhilT is a stats wiz (I hope) and can probably work out the number needed to see a statistical difference. The methodology for doing that is complex and beyond me, I'm afraid.

I completely agree re MM's n=1!

What is the more interesting trend? Am I being dense?

I started to read this thread, but got completely lost by about the 2nd comment - now that's dense Caroline - something which you're definitely NOT. I've given up now - I think I'll go back to the Crafty section

"Too few data on 4:3 to know I reckon"

Afraid I have to disagree, search PubMED, plenty of papers with n=5 or n=22.
My understanding is that n=3 is acceptable in most experiments.

[carorees]

Not sure about that...usually n=3 only in phase I trials of dose finding where more subjects are added depending on the findings at the lowest dose. Or case reports where no stats are done.

The standard way to determine what the sample size should be is to look at the size of effect you want to detect and the normal variation in the population, then some clever stats will tell you you're required sample size.

As you and others have noted the day to day weight variation is large so the sample size will have to be correspondingly large.

"Too few data on 4:3 to know I reckon"

Afraid I have to disagree, search PubMED, plenty of papers with n=5 or n=22.
My understanding is that n=3 is acceptable in most experiments.

N=3 or even 5 is definitely not acceptable in "most experiments". The smaller the difference that you want to detect, and the larger the variation in the group you study, the larger the sample has to be. In addition, if your not randomizing, a lot of background variables and variations come into play that you have to control for (increasing the number of required participants). The progress tracker/this forum is not an experiment, where comparable people are included and then followed up systematically. Just consider the dropout rate of those not succeeding.
At best (and even that can be debated) it is sort of observational. And the majority of positive findings from observational studies is not confirmed in repeat study, if I remember correctly (Ioannides 2005, DOI: 10.1371/journal.pmed.0020124)

[carorees]

I think the progress tracker takes care of drop outs by not including data for weeks where someone has not posted a weigh in...dominic and Moogie spent a long time discussing this so I think its OK from that POV. All the other weaknesses apply though!

Coffecat, can you comment on my proposed design for a survey? I know there are a lot of weaknesses but do you think there is a chance of yielding some useful observations?