Fundraising 2010/Report/Draft Test 6

From Meta, a Wikimedia project coordination wiki

[edit]

For additional documentation on the testing methodology please see the following pages:


Banners[edit]

1 Message Fader. This is a fader banner and features one fading text string which fades over even intervals. The string is: "If everyone reading this donated $5 our fundraiser would be over today. Please donate to keep Wikipedia free."


2 Message Fader. This is a fader banner and features two text strings which toggle by fading into each other for equal intervals. These strings are: "If everyone reading this donated $5 our fundraiser would be over today. Please donate to keep Wikipedia free." and "Only 3 days left to make a tax-deductible contribution to keep Wikipedia free. Please help Wikipedia pay its bills in 2011."


Result[edit]

Test Time: 2010-12-29 16:00:00 UTC - 2010-12-29 17:40:00 UTC
Sampling Interval = 2 minutes
Testing Interval = 100 minutes
Total Number of Samples per Class = 50

Donations / Impression*: "2 Message Fader" banner won by 7.92%.
Amount50 / Impression**: "2 Message Fader" banner won by 19.64%.

WINNER: "2 Message Fader"


(*) The rate of donations per banner impression over a fixed time interval.

(**) The rate of amount50 per banner impression over a fixed time interval. Amount50 is the dollar amount raised from donations initiated under a given banner where all donations of more than $50 are recorded as $50 donations. This counters the skewing effect of outlier donations.

Data Analysis[edit]

This section analyzes and interprets the results of the tests. It should be noted that the data is analyzed over a period at least as large as the full testing period and that the testing period was chosen based on the period of time where significant hits and donations were observed.

Data Consistency and Cleaning[edit]

The plots below display the counts of the data sources over the testing period as verification of the consistency of the donation pipeline data used in testing. It should be noted that a certain amount of natural variance is expected in donation counts and is no cause for alarm. The bottom two figures depict the total views and donations with those coming from each banner. This illustrates when the campaign becomes active and is also a useful tool for determining where anomalous behaviour may be visible in the data.

Finally, in regard to the bottom two plots at bottom, one other banner was running along with the two being compared and so the totals in those plots can be ignored (especially the "Views" plot).

Impressions broken out in two minute intervals over the test period.


Donations broken out in two minute intervals over the test period.


Donations/Impression broken out in two minute intervals over the test period.


Amount50/Impression broken out in two minute intervals over the test period.


The views corresponding to each banner and the total views over the campaign for two minute intervals.


The donations corresponding to each banner and the total donations over the campaign for two minute intervals.

Analyzing the above plots the donation and impression data appear to be quite regular over the interval 2010-12-29 16:00:00 UTC - 2010-12-29 17:40:00 UTC. Therefore, two minute intervals will be used for sampling over this period as a source for the paired t-test to assess confidence in the winner.

Modelling and Hypothesis Testing[edit]

"2 Message Fader" won in each case for donations/impression and amount50/impression with increases of 7.92% and 19.64% respectively. The student's t-test was used to assess confidence over each metric and the confidence in the winner for donations/impression and amount50/impression is at least 99.5% and 99.0% respectively. This is a significant win for the "2 Message Fader" and particularly in the amounts of money donated.

Mean and standard deviation over test intervals over Donations / Impression.


Mean and standard deviation over test intervals over Amount50 / Impression

TOTAL DONATIONS "2 Message Fader": 567
TOTAL DONATIONS "1 Message Fader": 540

TOTAL AMOUNT50* RAISED "2 Message Fader": $9218.98
TOTAL AMOUNT50* RAISED "1 Message Fader": $7993.00

* AMOUNT50 indicates the total amount raised where all donations greater than $50 are taken to be a donation of $50.


DONATIONS PER IMPRESSION:

Between 99.5% and 99.95% confident about the winner.

Animated Banners -- 2010-12-29 16:00:00 - 2010-12-29 17:40:00

item 1  = 1 message fading
item 2  = 2 messages fading

The winner 2 messages fading had a 7.92% increase.

interval	mean1		mean2		stddev1		stddev2

0		0.00023		0.00028		0.00005		0.00004
1		0.00024		0.00031		0.00003		0.00009
2		0.00027		0.00036		0.00011		0.00025
3		0.00031		0.00026		0.00001		0.00016
4		0.00033		0.00028		0.00017		0.00002


Overall Parameters:

mean1		mean2		stddev1		stddev2
0.00028		0.00030		0.00009		0.00014


AMOUNT50 PER IMPRESSION:

Between 99.0% and 99.5% confident about the winner.

Animated Banners -- 2010-12-29 16:00:00 - 2010-12-29 17:40:00

item 1  = 1 message fading
item 2  = 2 messages fading

The winner 2 messages fading had a 19.64% increase.

interval	mean1		mean2		stddev1		stddev2

0		0.00305		0.00501		0.00003		0.00067
1		0.00396		0.00495		0.00097		0.00003
2		0.00442		0.00576		0.00265		0.00365
3		0.00471		0.00417		0.00228		0.00439
4		0.00421		0.00445		0.00085		0.00003


Overall Parameters:

mean1		mean2		stddev1		stddev2
0.00407		0.00487		0.00167		0.00257

Endnotes[edit]

  1. Campaign = "20101229JA076"
  2. "1 message fading" utm_source = " 20101229_JAFS001fader_US"
  3. "2 messages fading" utm_source = "20101228_JAFader_US"