This question probably won't help.

What's the p-value for these post-hoc pairwise comparisons:

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

You don't have much of a grasp of statistics, do you?

Ad hominem.

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

5th place in Hawaii isn't FOP?

---

Swimming Workout of the Day:

Favourite Swim Sets:

2020 National Masters Champion - M50-54 - 50m Butterfly

Constitutio fact.

Was the null hypothesis rejected at a p-value of 0.01?

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

As I said, your grasp of statistics seems to be lacking. Here is a link that might help you understand why the alpha level you've chosen is ridiculously low - scenario D in particular is highly relevant:

http://www.graphpad.com/...r_which_values_o.htm

ad hominen, again. dkennison commented about a great link...so i read it and posed a question. if you can't answer my question to him, oh well. <shrug>

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

Constitutio fact, again...as the link I provided so clearly demonstrates.

haha! Still waking up on East Coast time I see!

---

---

Heath Dotson
HD Coaching:Website |Twitter: 140 Characters or Less|Facebook:Follow us on Facebook

Since you don't seem to want to answer that question, how about this one (I'd ask rchung, but he's looking into p-values for other comparisons):

In the link dkennison mentions, what is the p-value for the best to worst comparison (bicyle+equipment) outdoors?

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

It's a naive question.

Paraphrasing Dr. Motulsky:

"The sample size required to show statistically-significant differences using conventional values for alpha and 1-beta are too high to be feasible. You simply can't run that many replicates. After thinking about it, you realize that the consequences of making a Type II error (falsely concluding that one frame is faster than another) are much less than making a Type I error (falsely concluding that two frames are equal). A false hit will have little impact on your performance. On the other hand, falsely calling two bikes to be equal means that you'll miss out on a real competitive advantage. Therefore you choose a low value of alpha and also a low power."

Seems like you didn't actually read the article dkennison gave kudos to if you ask me. A simple p-value calc by you, as requested, could end this.

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

The difference I found between my wife's P2T and P3C track bikes is significant at alpha = 0.01 with 1-beta = 0.27.

Conversely, if in this context* you foolishly insist on 1-beta = 0.80, then alpha = 0.16.

(*Personally, the statistic I've always relied upon is $ per second per km saved, with something around $500/(s/km) being my cut-off.)

Sorry, screwed up my math - standard has been ~$1500/(s/km).

This has been an interesting thread.

• It's posted on a forum contributed to by the top percentages of triathlon athletes and technology mavens. The "Alphas", the top 1%'ers.
• Part of that is my error- part isn't.
• I asked how much credibility you assign to Bike Brand wind tunnel testing. Some people answered that question. Most didn't.
• The behavior of the thread then seemed to draw frequent and learned contributors out. They discussed topics only peripherally related to original inquiry.
• I theorize, but cannot prove, that the bottom 80% or so of triathletes would respond differently to this inquiry.
• Since the bottom 80% is the group that buys most of the full-price (key words: "full-price") triathlon equipment and bikes, that is the response group I'm most interested in. That's where the business is. The sport has filled from the bottom. Look at any big 70.3 or 140.6 race results page.
• The "Alpha" contributors in this thread have been contributing here for years and likely have not paid full retail price for a triathlon or road bike in a long time.

---

Tom Demerly
The Tri Shop.com

shorter, less mathy, less University of Marylandy (circa 2001 - click the linky-link), more empathetic andy:


"I'm sorry, Kraig, for not actually reading dkennison's link and subsequently calling you names".

---

=================
Kraig Willett
http://www.biketechreview.com - check out our reduced report pricing
=================

Perhaps you'd be better off posting elsewhere? Maybe like Beginner Triathlete. Everyone who has been on here for several years knows you have an axe to grind re: Aerodynamics, which what led to what you feel is a devolution of the OP.

---

---

Heath Dotson
HD Coaching:Website |Twitter: 140 Characters or Less|Facebook:Follow us on Facebook

I buy in the mid-price range, of just about everything in tri. (top of the line wetsuits excepted).

In the mid range (and probably higher than that) for frames and wheels, I think there is actually very little difference between brands and models. they all claim to be the best and if you dig into the data, it will be one data point they're talking about, and overall, 5/8 of sweet FA difference.

The only (aero) data I trust, even a little bit, is independent testing, and only when all brands are tested in the same tunnel at the same time etc etc.

At the end of the day, even when I trust independent aero data, it's only one factor I consider, when making a purchase. Many times, I will buy say a wheel, and aero is important but so are other factors weight,braking surface, bearings and their serviceability and then of course price.

If I was at the pointy end of the race, I'd probably be searching for greater gains from my equipment, with cost being less of a consideration, but as MOP living in a variation of the real world, lotsa things impact on my purchases and aero is just one of those factors.

---

TriDork

"Happiness is a myth. All you can hope for is to get laid once in a while, drunk once in a while and to eat chocolate every day"

I'm not sure why people are so reluctant to provide p-values, it seems like a reasonable request to me, if someone is trying to convince people of the value of their testing.

For example, a paired t-test gives a p-value of 0.0004 for my CdA change from 2011 to 2012, and 0.0003 for my CdA change from 2012 to 2013, pairing the CdA values from the same course:



Of course, it's easier to get a low p-value if the change is larger, so the standard deviation and number of reps are arguably more informative. In my example above, the low p-value is mainly a result of the changes being large rather than the testing being high quality (however it did cost me nothing to collect the data, and very little time to analyse it).