[mynamewastaken]

[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
[ QUOTE ]
This suggests to me that being in the bottom 3 (or equivalent) has at least a small rallying effect, in that there is a strong correlation between being in the bottom 3 and being safe the next week.

[/ QUOTE ]

This statement is probably inaccurate, since there are more people not in the bottom 3 than in the bottom 3. I haven't done the math, but 3 out of 9 could still mean a positive correlation between being bottom 3 and being eliminated.

[/ QUOTE ]

I'm not sure I understand your math. To me, the people who DON'T wind up in the bottom 3 cancel out to a zero-sum.

I was trying to answer the question: Is being in the bottom 3 helpful? To figure this out I discount all the people who are NOT in the bottom 3 (or equivalent) and then look to see if those people who ARE in the bottom 3 are hurt or helped by it.

From the limited data set posted (and my recollection as well) it seems to me that those people who find themselves in the bottom 3 one week are likely to be safe the next, and I attribute this to the rallying effect discussed herein. Certainly, 3 of 9 last year went home after being the bottom 3, but the fact that 67% survived the next week suggests to me that the concept of mobilizing one's fanbase is very real.

As to whether this rallying effect can sustain a contestant through the competition, the answer is much more difficult but I would guess no, as those contestants who are perennial bottom-3ers do not fare very well overall.

You'll notice that those contestants in the bottom 3 on top12 night (Phil, Sanjaya) were not even in the bottom this week. Although there are many possible factors for this, it is perfectly in line with the trend of contestants in the bottom 3 being safe the next week.

Anyway, wasn't trying to argue but I'm not sure I see your point.

[/ QUOTE ]

It's called regression to the mean. Look into it.

[/ QUOTE ]

There's no need to be condescending; I was genuinely looking for a response and I appreciate it.

I hardly think that given the vote totals are hidden and the widespread fear of vote tampering, statistical ties, etc. my analysis can be legitimately construed as a regression fallacy. If and only if we knew that the bottom 3 who are safe were significantly lower in vote totals than those contestants NOT in the bottom 3 could we deduce a regression towards the mean. As it stands, for all we know the difference between Richardson and whoever was 3rd last on BI night was 1 vote or a million votes... no way to tell where Richardson fell on the bell curve.

Even if your alternative theory is legitimate, which it very well could be, it is just further evidence that being in the bottom 3 is helpful as it increases the likelihood of a future positive regression, leading to future safety from elimination.

Is your point that a contestant's performance the week before has no bearing on his ranking the next week? Or is it that a person in the bottom 3 is pretty much on the cusp of elimination, and has no chance to recover? Or is it that the answer is situation dependent?