What were the pool sessions like in your OW course?

[runsongas]

4 half days over 2 weekends. i think the deep end was 12 ft or so. i was reasonably comfortable in the water but my buoyancy was not great. the pool was way warmer than the checkout dives.

 

[Wookie]

You touched on something (maybe meaning to, maybe not) that I was discussing at BTS with some folks, and that was ratios. We were trying to figure out what the drop rate was for students through their first 10 dives. I believe it to be high, but how high I don't know. How many don't feel safe, and what would it take for everyone to feel safe. My response was that ratios need to be lowered. If you are going to keep a student "safe", that student needs to be your sole focus. Then, run them through an advanced class, but for goodness sake, don't let them consider them to be advanced in any way.

I fully understand that it's hard to offer a $99 class one on one, but I wonder what the retention would be if we stopped letting our divers get scared and drop out..... I wonder what a retained diver spends versus one that stops diving on OW dive 3 or 4....

 

[rhwestfall]

would have been interesting to see:

1. what geographic location the respondents were from
2. do they dive regularly "at home"

 

[RJP]

would have been interesting to see:

1. what geographic location the respondents were from
2. do they dive regularly "at home"

Thought of that too... but was trying to keep it to a reasonable length.

Plus, there'd be so many confounding factors in between "pool experience" and "where people dive" that it'd be impossible to draw any conclusions from that information... without asking even more questions.

As a market research professor told me once about study design: "don't ask a question if you can't do anything with the answer."

 

[gypsyjim]

I did my best to answer you poll, but my original class was a Y class in 70, so, being an aging diver I couldn't truly, accurately every one of the questions, so I did my best to give answers based on somewhat cloudy recall.

 

[rx7diver]

As a market research professor told me once about study design: "don't ask a question if you can't do anything with the answer."

With respect, this is only one of the major differences between how a competent statistician approaches this type of thing, versus how a market researcher approaches this type of thing. Statisticians do not, in general, ignore relevant information just because this information is potentially confounding. With statisticians, margin-of-error (or "sampling error") is paramount. Therefore, knowing whether or not there is underlying confounding is extremely important to a competent statistician.

Save Diving

rx7diver

 

[RJP]

With respect, this is only one of the major differences between how a competent statistician approaches this type of thing, versus how a market researcher approaches this type of thing. Statisticians do not, in general, ignore relevant information just because this information is potentially confounding. With statisticians, margin-of-error (or "sampling error") is paramount. Therefore, knowing whether or not there is underlying confounding is extremely important to a competent statistician.

Save Diving

rx7diver

It would be confounding because sample size would be inadequate, and not sufficiently representative of the overall population, to draw any meaningful conclusions of subanalyses by geography or location of diving. Beyond confounding, that information is also irrelevant to the point of the survey.

With respect, a competent statistician would understand that. Of course that sort of difference between "statistics" and "information" seems to confound even the most competent statistician.

You can go back to trying to understand how it is possible for the average family to have 2.5 children* (p<.001) now.

 

[rx7diver]

It would be confounding because sample size would be inadequate, and not sufficiently representative of the overall population, to draw any meaningful conclusions of subanalyses by geography or location of diving. Beyond confounding, that information is also irrelevant to the point of the survey.

With respect, a competent statistician would understand that. Of course that sort of difference between "statistics" and "information" seems to confound even the most competent statistician.

You can go back to trying to understand how it is possible for the average family to have 2.5 children* (p<.001) now.

RJP,

Just what kind of "meaningful conclusion" can be drawn if easily discoverable confounding issues are ignored? There are well-understood approaches one might try, to ensure a representative sample. And the uncertainty resulting from a small sample size is reflected by the margin-of-error. A too-large margin-of-error suggests one should not have confidence in his "meaningful conclusion." I'm sure you know all of this.

Safe Diving,

rx7diver

 

[RJP]

RJP,

Just what kind of "meaningful conclusion" can be drawn if easily discoverable confounding issues are ignored? There are well-understood approaches one might try, to ensure a representative sample. And the uncertainty resulting from a small sample size is reflected by the margin-of-error. A too-large margin-of-error suggests one should not have confidence in his "meaningful conclusion." I'm sure you know all of this.

Safe Diving,

rx7diver

Again, you're worried about statistics. This survey is looking for information. That said, I would love to know what "the well-understood approaches that one might try to get a representative sample" on a platform like ScubaBoard? It's the ultimate example of a innately-biased convenience sample. When I use Harris Interactive for "a real project" I get access to >1,000,000 households and all of Harris' statisticians. Of course I also get a 5-figure invoice at the end of the project...

It's an online straw-poll... don't lose any more sleep over it.

:cool2:

 

[rx7diver]

Again, you're worried about statistics.

Yes.

It's an online straw-poll... don't lose any more sleep over it.

Okay.