How Much Did It Cost You To Become/Retain EXP status for 2010?
Nov 7, 2010, 1:14 pm
#76
Join Date: Jan 2001
Location: ORD
Programs: AA EXP, HH Diamond
Posts: 376
I spent 15.5KUSD up to the 100K mile point. Another 1K after that with an additional ~0.8K later this month.
Nov 8, 2010, 5:14 am
#77
Join Date: May 2001
Location: New York, New York
Posts: 1,602
By my understanding of the question, 32 usable replies so far.
Minimum = $2,600
Maximum = $36,500
Mean = $12,498 (average spend)
Median = $8,350 (16 spent more, 16 spent less)
Standard Deviation (entire population basis) = $9,357
Methodology:
Excluded last year info, excluded incomplete info (e.g., fare without taxes)
When reported cost resulted in >100K EQM/EQP, prorated cost back to exactly 100
Counted vouchers at reported value
Note considerable variance possibly due to DEQM.
Minimum = $2,600
Maximum = $36,500
Mean = $12,498 (average spend)
Median = $8,350 (16 spent more, 16 spent less)
Standard Deviation (entire population basis) = $9,357
Methodology:
Excluded last year info, excluded incomplete info (e.g., fare without taxes)
When reported cost resulted in >100K EQM/EQP, prorated cost back to exactly 100
Counted vouchers at reported value
Note considerable variance possibly due to DEQM.
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
Nov 8, 2010, 6:17 am
#78
Join Date: Oct 2009
Location: Austin, TX
Programs: AA LT Plat, UA 1k/1mm+, National EE, IC Plat, Bonvoy Gold
Posts: 2,605
This is an example of how NOT to use statistics, as the OP has violated some of the basic rules. To use statistics the way the OP has requires that the distribution be Gaussian (bell shaped/normal), and clearly the data are not! Such conclusions from this application would be that about 17% of the flyers were paid by AA sums up to $9364 to qualify. (Those that were between 1 and 2 Standard Deviations below the mean).
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
Nov 8, 2010, 6:54 am
#79
FlyerTalk Evangelist
Join Date: May 2004
Location: DFW/DAL
Programs: AA Lifetime PLT, AS MVPG, HH Diamond, NCL Platinum Plus, MSC Diamond
Posts: 21,422
Nov 8, 2010, 7:14 am
#80
Join Date: Mar 2009
Location: PHX/NYC/LA
Programs: AA Plt, Marriott Gold, National EE, Hertz PC, El Mambero De Mucci, PWP Aide to Generalissimo Godot
Posts: 4,893
If I do make it, It seems I will be in the neighborhood of $2750-$3750 spent towards EXP.
Nov 8, 2010, 8:04 am
#81
FlyerTalk Evangelist
Join Date: Mar 2004
Location: SJC
Programs: AA EXP, BA Silver, Hyatt Globalist, Hilton diamond, Marriott Platinum
Posts: 33,533
This is an example of how NOT to use statistics, as the OP has violated some of the basic rules. To use statistics the way the OP has requires that the distribution be Gaussian (bell shaped/normal), and clearly the data are not! Such conclusions from this application would be that about 17% of the flyers were paid by AA sums up to $9364 to qualify. (Those that were between 1 and 2 Standard Deviations below the mean).
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
Without looking at the actual distribution of the numbers, I'm guessing it's bimodally-distributed. It it actually is, this is consistent with the comment above that the modes of the distribution are based on who is paying
Cheers.
Nov 8, 2010, 8:10 am
#82
Join Date: Feb 2008
Location: San Francisco Bay Area
Programs: AA Plat
Posts: 757
Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_1 like Mac OS X; en-us) AppleWebKit/532.9 (KHTML, like Gecko) Version/4.0.5 Mobile/8B117 Safari/6531.22.7)
Im guessing poisson distribution. Jk
Also did those who had overnight mileage runs add hotel and visa costs?
Im guessing poisson distribution. Jk
Also did those who had overnight mileage runs add hotel and visa costs?
Nov 8, 2010, 9:02 am
#83
Join Date: Nov 2009
Location: ORD/PSP
Programs: AA LT Plat, 6.0 MM, Marriott LT Platinum
Posts: 1,002
This is an example of how NOT to use statistics, as the OP has violated some of the basic rules. To use statistics the way the OP has requires that the distribution be Gaussian (bell shaped/normal), and clearly the data are not! Such conclusions from this application would be that about 17% of the flyers were paid by AA sums up to $9364 to qualify. (Those that were between 1 and 2 Standard Deviations below the mean).
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
With such small numbers, the OP should have taken the log of how much each flyer spent and then tested the distribution to see if it was Gaussian. Another and far more practical approach would have been to rank the costs of each flyer, and report what those at percentiles such as 5, 25, 50, 75, and 95 had spent.
We all would have been in line if AA were paying us to qualify for EXP!
Obviously the distribution is not normal and I did not suggest it was, I'm not the one who calculated that "17% of the flyers were paid by AA sums up to $9364 to qualify", you did that and then accused me of treating the distribution as normal.
Nov 8, 2010, 9:09 am
#84
FlyerTalk Evangelist
Join Date: Mar 2004
Location: SJC
Programs: AA EXP, BA Silver, Hyatt Globalist, Hilton diamond, Marriott Platinum
Posts: 33,533
I still think it was a nice treatment, even if only the median has significance
Cheers.
Nov 8, 2010, 9:13 am
#85
Join Date: Nov 2009
Location: ORD/PSP
Programs: AA LT Plat, 6.0 MM, Marriott LT Platinum
Posts: 1,002
I think the point is that, while mean and standard deviation are just formulae and can be calculated for any distribution just by plugging in, the meaning and significance of these values, especially the standard deviation, are held to have little predictive significance or value in non-Gaussian distributions. In other words, it's a number that can be computed, but may not tell us much if the distribution is not normal.
I still think it was a nice treatment, even if only the median has significance
Cheers.
I still think it was a nice treatment, even if only the median has significance
Cheers.
Last edited by siankaan1; Nov 8, 2010 at 10:35 am
Nov 8, 2010, 3:53 pm
#86
Join Date: Jan 2010
Location: BOS
Programs: AA EXP, 1MM. HH Diamond
Posts: 240
Trips 2010 = 26
Spend to date = $16,031
Average flight cost = $615
Median = $503
Cheapest flight = $229 (BOS<->SJC)
Priciest flight = $1645 (BOS<->DFW, in Y no less)
Spend to date = $16,031
Average flight cost = $615
Median = $503
Cheapest flight = $229 (BOS<->SJC)
Priciest flight = $1645 (BOS<->DFW, in Y no less)
Last edited by gamehendge2000; Nov 8, 2010 at 4:00 pm
Nov 8, 2010, 4:08 pm
#87
Join Date: Oct 2006
Location: AUS, LAX
Programs: AA EXP - 2.2 MM, Admirals Club, Hilton Diamond, Marriott Titanium, Avis Presidents, National Exec
Posts: 1,581
$0 for me, but for my company, probably $12,000-$15,000.
Mainly domestic, average ticket price is usually just under $400 with ~4 legs. Qualified in June so I just factored what it cost to get me back to EXP. I'll probably fly about 150% what was required to get to EXP this year. I've been deferring as much Q4 travel as possible to Q1 since it's diminishing returns now
Mainly domestic, average ticket price is usually just under $400 with ~4 legs. Qualified in June so I just factored what it cost to get me back to EXP. I'll probably fly about 150% what was required to get to EXP this year. I've been deferring as much Q4 travel as possible to Q1 since it's diminishing returns now
Nov 8, 2010, 7:34 pm
#88
Join Date: May 2001
Location: New York, New York
Posts: 1,602
None of what I wrote posits a distribution. The mean and standard deviation of the sample are what they are.
Obviously the distribution is not normal and I did not suggest it was, I'm not the one who calculated that "17% of the flyers were paid by AA sums up to $9364 to qualify", you did that and then accused me of treating the distribution as normal.
Obviously the distribution is not normal and I did not suggest it was, I'm not the one who calculated that "17% of the flyers were paid by AA sums up to $9364 to qualify", you did that and then accused me of treating the distribution as normal.
Just use cutoff values, (i.e. what someone at the 25th and 75th, and/or 5th and 95th percentile paid) and it will make sense to all who read your post.
Great idea. I look forward to seeing this (your) data next year!
Nov 9, 2010, 11:27 am
#89
Join Date: Nov 2009
Location: ORD/PSP
Programs: AA LT Plat, 6.0 MM, Marriott LT Platinum
Posts: 1,002
The basic point is that some of the data you provided was very valuable, and I suspect we all endorse what you did, but it was in how you did it that I had a minor squabble. If you provide the statistics that you did, then the assumption is that the distribution is Gaussian. It may be what BRP suggests of being bimodal, but it clearly does not support the statistical treatment you provided. Hence the conclusions do not make sense. If you provide the median and the SD, then the data that corresponds to about 17% of the flyers had negative values (i.e. were rewarded for flying rather than paid).
Just use cutoff values, (i.e. what someone at the 25th and 75th, and/or 5th and 95th percentile paid) and it will make sense to all who read your post.
Great idea. I look forward to seeing this (your) data next year!
Just use cutoff values, (i.e. what someone at the 25th and 75th, and/or 5th and 95th percentile paid) and it will make sense to all who read your post.
Great idea. I look forward to seeing this (your) data next year!
Nov 9, 2010, 1:59 pm
#90
Used to be 'Travelergcp'
Join Date: Jul 2003
Location: New Orleans
Programs: AA Plat, Marriott Gold, Hyatt Globalist
Posts: 2,826
probably 13k total. 2-3k on MR/personal travel to get me to EXP.