Finding a Cheap Date

NEWSWEEK -- May 31 -- The wave of the future, perhaps: free internet dating. In April, Craigslist saw 2.6 million personal ads posted, up from only 970k two years ago. PlentyofFish.com, a free dating site that was launched in 2003, now brings in 200k U.S users a day - and $5 million to $10 million in advertising a year according to Markus Frind, who runs the site by himself. "I think all the paid sites are going to go away, he says" Even Match.com is offering discounts to subscribers: six months free, if you don't find Prince (or Princess) Charming in the first six months. While the online dating industry has been enormously successful so far.

Comments

[Fernando Ardenghi]

At http://www.plentyoffish.com/personality_faq.aspx
can be read
"Plenty of Fish Compatibility Predictor
.....
Specifically, Item Response Theory and Rasch Scaling analyses were used to first, identify which items provided “good” indicators of certain personality characteristics and second, which items were biased for men or women. This multi-step process resulted in a 46-item relationship test that assesses individuals’ standing on five broad dimensions of personality. This is the only known relationship test where the test scoring and matching algorithm are based in the statistical gold standard of /b/modern test theoryb/ – in other words it is psychometrically the most sophisticated compatibility test available today.
....."

---- Who is/are the brain/s behind that compatibility test & algorithm?

"6. How does the CP matching algorithm work?
.......
PlentyofFish is proud to lead the way in setting quality standards for compatibility testing in the industry.
...."
---- Perhaps they should specify three main things for POF matching algorithm: ensemble, precision and how results are displayed.

"6. How does the CP matching algorithm work?
...
The preceding thus illustrates that depending on which type of "similarity" is being used, quite different results are obtained. In the top portion of the graph the two people are similar because the two people are both "average" in Confidence, Openness, Easy Goingness, Self-Control and Family Orientation. The pair in the bottom section is "similar" because both people score highest on Family Orientation and lowest on Openness – i.e. the two sets of scores are ranked about the same.
...
Also, we found that using both types of similarity simultaneously better captures the people's individual characteristics and the dynamics of a couple. Therefore, our similarity algorithm combines the two types of indices in a proprietary fashion. "

---- In the top portion of the graph the two profiles look similar because the two people *score* both *average* in those 5 factors but they *trend to score* in opposite directions in Openness, Easy Goingness and Self-Control. Perhaps POFCP should take into account not only the *score* but also the *trend to score*. i.e. the *position* and the *direction*. The pair in the bottom section is very dissimilar because one person *scores high* in all variables and the other *scores low*, and also both people trend to score in opposite directions!!!

I think PlentyofFish Compatibility Predictor is given only by a linear regression equation used to assess similarity because:
- I had taken the test (several times with several fake profiles) and received only a written report with some "compatibility matches" no one better than I could had found searching by my own, including zip code as main filter!
and
- the Plenty of Fish Compatibility Predictor (POFCP) is only a 2 integer number, like 96%.

I calculate similarity in PERSONALITY PATTERNS like these examples

Client #Man1 ------ 16PF5 Profile A:6.B:7.C:5.E:8.F:8.G:6.H:7.I:7.L:8.M:7.N:4.O:5.Q1:8.Q2:7.Q3:3.Q4:4
abv.
Client #Man1 --------- 16PF5 Profile 6.7.5.8.8.6.7.7.8.7.4.5.8.7.3.4
Client #Woman1 ---- 16PF5 Profile 5.6.4.7.7.5.6.6.7.6.3.4.7.6.2.3
/#Man1|C|#Woman1/ == 88,38926174%

Client #Man1 --------- 16PF5 Profile 6.7.5.8.8.6.7.7.8.7.4.5.8.7.3.4
Client #Woman2 ---- 16PF5 Profile 7.8.6.9.9.7.8.8.9.8.5.6.9.8.4.5
/#Man1|C|#Woman2/ == 87,58389262%

/#Man1|C|#Woman1/ is not the same as /#Man1|CQ|#Woman2/ because the algorithm I had invented takes into account the *score* and the *trend to score*

Client #Man1 --------- 16PF5 Profile 6.7.5.8.8.6.7.7.8.7.4.5.8.7.3.4
Client #Woman3 ---- 16PF5 Profile 8.9.7.10.10.8.9.9.10.9.6.7.10.9.5.6
/#Man1|C|#Woman3/ == 60,23489933%

Here /#Man1| is the reverse of |#Woman4/
Client #Man1 --------- 16PF5 Profile 6.7.5.8.8.6.7.7.8.7.4.5.8.7.3.4
Client #Woman4 ---- 16PF5 Profile 4.3.7.8.5.4.7.8.7.7.6.8.8.5.7.6
/#Man1|C|#Woman4/ == 68,15436242%

Client #Man2 --------- 16PF5 Profile 6.7.6.7.6.7.6.7.6.7.6.7.6.7.6.7
Client #Woman5 ---- 16PF5 Profile 4.5.4.5.4.5.4.5.4.5.4.5.4.5.4.5
/#Man2|C|#Woman5/ == 63,75838900%

Client #Man2 --------- 16PF5 Profile 6.7.6.7.6.7.6.7.6.7.6.7.6.7.6.7
Client #Woman6 ---- 16PF5 Profile 3.4.3.4.3.4.3.4.3.4.3.4.3.4.3.4
/#Man2|C|#Woman6/ == 41,61073800%

Client #Man2 --------- 16PF5 Profile 6.7.6.7.6.7.6.7.6.7.6.7.6.7.6.7
Client #Woman7 ---- 16PF5 Profile 7.8.7.8.7.8.7.8.7.8.7.8.7.8.7.8
/#Man2|C|#Woman7/ == 88,59060403%

Kindest Regards,
Fernando Ardenghi.
Buenos Aires.
Argentina.
[email protected]

[Dom]

I have had an ongoing interest in this craze for compatibility and relationship testing. Plus I am well versed in psychometrics.

I have read the Plenty of Fish FAQ on their "Compatibility Predictor," and it contradicts what Fernando Ardenghi seems to be saying above.

The POF index takes into account both the magnitude and direction of two people's profiles. The first (magnitude) by looking at the average levels of the ratings, and the second (direction) by computing a correlation coefficient between the profiles.

Actually, this seems better than what Fernando proposes; he seems to order factors and then checks how scores increase or decrease between adjacent pairs of factors. But, of course, the ordering of the factors is arbitrary and thus his "directional" component does not make sense.

POF's use of the correlation coefficient is clearly superior. Kudos to Markus Frind and POF for a nicely constructed test, and one that is free to users (very nice feature).

Dom

[Fernando Ardenghi]

Dear Dom:

"POF's use of the correlation coefficient is clearly superior"

You must be joking!
Example for the matching algorithm I had invented
- Ensemble (whole set of different valid possibilities): 1 * E16
- Precision: better than 0,00000001% with Self-Adjustment
3 most compatible persons in a 100,000 persons database,
12 most compatible persons in a 1,000,000 persons database,
48 most compatible persons in a 10,000,000 persons database,
- Results are displayed with 2 integers + 8 decimals, like 92.55033557% +/- 0,00000001%

the Plenty of Fish Compatibility Predictor (POFCP) is only a 2 integer number, like 96% !!! POFCP Precision: not better than +/- 1% ????

Please see:

the paper "METHODOLOGICAL AND DATA ANALYTIC ADVANCES IN THE STUDY OF INTERPERSONAL RELATIONSHIPS: INTRODUCTION TO THE SPECIAL ISSUE"
http://lilt.ilstu.edu/personalrelationships/contentsofissues/pdf_articles/Volume%206,%20Issue%204/Kashy.pdf

At page 413 says: "It is vital for the study of personal relationships, as for any scientific discipline, to develop methodologies that are specifically designed to address the questions posed by the discipline. The articles in this special issue represent an effort in that direction. Perhaps equally important is the need for individuals who are involved in relationships study to learn these new techniques and to apply them in their research. It is also important for investigators to challenge statisticians to create new analytic techniques when existing ones are inadequate. These tasks are left to you, the reader."

http://mb.internetdatingconference.com/scientific-papers-t395.html

http://en.wikipedia.org/wiki/Braket

http://en.wikipedia.org/wiki/Quantum_mechanics

Kindest Regards,

Fernando Ardenghi.
Buenos Aires.
Argentina.
[email protected]

[Jon Coward]

I operate a fledgling free Interracial dating site called InterracialOasis.com. While free may be the wave of the future, getting started with credibility is the hardest part. Being perceived as credible is the key to making money with a free site. For the behemoths, switching to a free model is easy. For start ups, it is very hard. People don't want free if you don't have a quality site or a low number of members. My site has a low member count, but I can guarantee that everyone on there signed up. No list of daters was purchased. Yes, it outs me at a disadvantage, but the longer I slug it out, the better off everyone will be in the beginning.

[April]

Compatability Predictor is not new and has been used by dating sites for a long time. So plentyoffish is not leading the industry.

Club Intimate (www.ClubIntimate.com) also uses this. Check it out.

[syc]

Everything that applies to the Internet, applies to online dating as well. The Internet as we know allows for unlimited possibilities in communication, and it is this feature that has proved to be at the same time the biggest boon, as well as bane for online dating.

But at the same time this possibility for unlimited communication leaves a lot of space for guile as well. The human race is endowed with a remarkable ability to use, misuse and abuse the same thing. And naturally, online dating too has been and is still being used for vile purposes.

The person who is misusing this facility may either be a practical joker or may be someone with more devious intentions who is out to get some victims. It is because of this reason that a little bit of homework is good before you actually hit the road. "5 Steps To Online Dating Success" is a great e-book that offers easy to follow guidelines to help with online dating.
(ebookstore.itgo.com)