General fitness, health and nutrition · Public discussion

low-carb kicks butt in studies - again

Started by Tcomeau · · Last activity · 22 posts · 899 views

Thread navigation

Jump through the discussion

Go to the original post, the replies on this page, or the latest preserved contribution.

Thread details

What we know about this thread

Original section
General fitness, health and nutrition
Published
18 May 2004
Last activity
23 May 2004
Original author
Tcomeau
Posts
22
Discussion status
Public discussion
Total views
899
Views / 30 days
0

The navigation and discussion metadata provide context. Posts remain in their original chronological order.

Showing posts 21–22 of 22
Posts remain in their original chronological order.

Text size
  1. On Tue, 18 May 2004 11:30:49 -0400, Jackie Patti <[email hidden]>

    Quoted message said:
    Quoted message said:
    Quoted message said:

    Results: By 1 year, mean (±SD) weight change for
    persons on the low-carbohydrate diet was –5.1 ± 8.7
    kg compared with –3.1 ± 8.4 kg for persons on the
    conventional diet. Differences between groups were not
    significant (–1.9 kg [95% CI, –4.9 to 1.0 kg]; P =
    0.20).

    Quoted message said:

    p = 0.20 is a description of the measurable significance,
    and this is a very low measure, comparable to just one
    widget's weight being off in our example above. It means
    they consider this to possibly be just luck and can't know
    if it's real for the population without more study.

    Quoted message said:

    Generally, a scientist wants p > 0.90 or > 0.95 or > 0.99.
    Which cut-off is acceptable depends on the specific type
    of study.

    But p = 0.2 *definetly* is not good enough to say the
    differences between those groups is true (for that
    particular measurement, other measurments in the study are
    significant).

    Again, the difference between the scientific definition and
    how the man-in-the-street reacts. If your p=.2 means that
    there is a 20% chance that the result was arrived at by
    chance, OK, the man-in-the-street may be willing to agree
    that the matter has not been PROVEN - but he'll probably be
    willing to bet with the 80% chance that it WASN'T by chance.

    Quoted message said:

    So in summary, they are saying they measured a difference,
    but the difference isn't meaningful enough to signify
    anything beyond what happened to their sample.

    They're saying that they haven't PROVED that the difference
    in results was caused by the difference in diet under study.
    Nonetheless, they are also saying that the odds are 4-1 that
    it WAS, in fact, the lowcarb diet that made the difference.
    Many people are willing to invest some resources (time,
    effort, money, etc.) if the odds are 4-1 in their favor.

  2. Eric Bohlman <[email hidden]> wrote in message news:<[email hidden]>...

    Quoted message said:

    Bob in CT <[email hidden]> wrote in
    :"]news:[email hidden]:

    Quoted message said:
    Quoted message said:

    Results: By 1 year, mean (±SD) weight change for
    persons on the low-carbohydrate diet was â??5.1 ± 8.7
    kg compared with â??3.1 ±
    8.4 kg for persons on the conventional diet.
    Differences between groups were not significant
    (â??1.9 kg [95% CI, â??4.9 to 1.0 kg]; P
    = .20).

    I always wonder how they call 2 kg, about 4 pounds, not
    significant. It's pretty darn significant in my life!

    In statistics, "significant" doesn't mean "personally
    meaningful." In this case, it means that there would be a
    20% chance of finding a difference of at least 1.9 kg
    purely as a result of "luck of the draw" in choosing these
    particular samples, even if there were no difference in
    the underlying populations. In other words, you don't have
    a whole lot of confidence that you'd be able to replicate
    that particular result.

    Let's not forget that Gaussian statistics are typically
    assumed to be valid. I think this is not really a justified
    assumption since the body types are not a random thing and
    there is a rather distinct division along the line of
    insulin resistance. When you don't have Gaussian statistics
    then the significance test is useless.

Active in the last 60 minutes

Active in this thread

0 users · 0 guests ·0 bots ·0 total

No signed-in users are active right now.

No known search crawlers active right now.