YAHOO NEWS — Oct 13 — Helen Fisher, an anthropologist and research professor at Rutgers University's Center for Human Evolutionary Studies, specializes in love, marriage, and gender differences. She's the author of four books, including her most recent, titled Why We Love: The Nature and Chemistry of Romantic Love. She believes that the type of person we are attracted to is hardwired into our neurons, etched by a combination of hormones, brain chemicals, and childhood experiences. As an adviser to new spinoff, Chemistry.com (backed by Match.com), Fisher is trying to quantify that certain something we're all looking for in a mate. "Most people fall in love because they have shared values, but they stay in love because their personalities mesh. Childhood also plays an enormous role in shaping likes and dislikes. I want to know not only what your brain chemistry is, but what was successful for you in the past. What really astonishes me is that I came up with four basic personality types in my research, and these same four types have been described by Plato, Aristotle, Carl Jung, Myers-Briggs." One of the questions on Chemistry.com asks how long your index finger is compared to your ring finger. A person with an index finger shorter than the ring finger will have been exposed to more testosterone while in the womb, and a person with an index finger longer than the ring finger will have had more estrogen. In women, the two fingers are usually equal in length, as measured from the crease nearest the palm to the fingertip. In men, the ring finger tends to be much longer than the index finger.
Mark Brooks: My interview with Jim Safka, Match.com CEO, goes live on OPW on Tuesday.
At the United States and Europe there are many ONLINE DATING sites that are more scientific than others.
United States: The most important are
– Yahoo! Personals works only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Mark Thompson and Dr. Glenn Hutchinson (Psychologists) Adapted Big7 model. One person is attracted by 1 in 30 (Typical) 12 Personality Types and 12 Love Styles.
– True works only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Ilona Jerabek (Psychologist) Adapted Rasch and Big5 model / 99 relationship factors explored. All members screened about criminal records.
– eHarmony works only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Neil Clark Warren (Psychologist). A person needs to be matched (with other) on 29 dimensions of compatibility.
– PerfectMatch works only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Pepper Schwartz (Sociologist). Adapted MBTI model “DUET method”
– Lovehappens works only in English for the United States. Uses different proprietary tests AND proprietary comparison methods between prospective mates (matching algorithm) invented by their own team.
– Matchwise works only in English for the U.S. Citizens. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Kevin Leman (Psychologist) Includes the “birth order” item.
– Chemistry seems to work only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Helen Fisher (Anthropologist). Adapted MBTI / adapted DISC model??? Also includes the “birth order” item.
Europe, UK
– Cybersuitors works only in English for the UK. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Glenn Wilson (Psychologist). CQ is an integer number that indicates the compatibility quotient between persons.
Europe, France
– http://www.ulteem.fr works only in French for France. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by their own team.
Europe, Germany
– http://www.ulteem.de works only in German for Germany. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by their own team.
“Scientific” means they include topics about THEORIES OF ROMANTIC RELATIONSHIPS DEVELOPMENT and, also Psychologists, Sociologists or Anthropologists who are members of their team and had published BOOKS, scientific papers, etc. Unfortunately, NOT ALL of the Psychologists, Sociologists and Anthropologists agree between themselves!!!
There is a big weak point / a common weakness PATTERN FOR ALL these last ONLINE DATING sites: Many users / subscribers to these sites that use proprietary tests and models complaint about an actual big problem in “scientific dating and matchmaking”: lack of precision / low precision / low successful matching rates.
As far as I could analyze, it seems that proprietary tests or models have great precision in measuring different psychological variables but the matching algorithm has low precision when comparing one profile to others.
The great improvement would come when each dating site specifies
its ENSEMBLE (the whole set of different valid possibilities)
and
a figure per each comparison between persons, like:
Client#01 to Client#02 == 74.79865772%
Client#01 to Client#03 == 54.09395973%
Client#01 to Client#04 == 92.55033557%
Client#01 to Client#05 == 57.71812081%
Client#01 to Client#06 == 59.73154362%
Client#01 to Client#07 == 68.99328859%
“Client#01, please contact Client#04, the probability of being compatible is as high as 92.55033557%”
and not a bar-graphic, a written report or a series of 5 empty / half / full-filled hearts-icon.
Kindest Regards,
Fernando Ardenghi.
Buenos Aires.
Argentina.
ardenghifer@gmail.com
————————————————————–
“True works only in English for the United States. Uses a proprietary test AND proprietary comparison method between prospective mates (matching algorithm) invented by Dr. Ilona Jerabek (Psychologist) Adapted Rasch and Big5 model / 99 relationship factors explored. All members screened about criminal records.”
This comment is not accurate. The True Compatibility Test was the brainchild and product of a group of psychologists — not simply Dr. Ilona Jerabek. Furthermore, the test encompasses more than merely the Big Five model of personality. Interested readers can learn more at:
http://www.goertzel.org/dynapsyc/2005/True.htm
However, it is the only test that I currently know of that has as its foundation the gold standard of test and measurmement mathematics — Item Response Theory (IRT). IRT also serves as the basis for True’s new Sexploration, sexual compatibility, test. This test is the first of its kind to use adaptive testing technology — or learning algorithms.
Sexploration was also developed by a group of psychologists (I lead the team on its creation), although the software that runs the adaptive testing on the back end was the amazing brainchild of Dr. Rense Lange of Integrated Knowledge Systems, Inc. (www.iknowsys.org). Check his site out!
Dr. Lange now offers this fantastic software to other firms, and it can run many different applications (as opposed to just applications related to online dating).
Thanks,
James Houran, Ph.D.
Dear Dr. Houran:
Thanks for your clarification and corrections. It is always wonderful hearing from you, I am continually learning from your comments posted at different blogs.
I had been reading the paper you suggested: “Psychometric Description of the True Compatibility Test TM” at http://www.goertzel.org/dynapsyc/2005/True.htm
QUITE INTERESTING!!!
That paper also names the UK online dating site Cybersuitors.com, psychometric supported by Dr. Wilson, who invented the C.Q. Compatibility Quotient.
Please do not think I am rude or not polite, but that site never worked as it was expected. You can check its Alexa rank at
http://www.alexa.com/data/details/?url=www.cybersuitors.com
At the references can be seen:
– Wilson, G. D., & Cousins, J. M. (2003b). “Partner similarity and relationship satisfaction: development of a compatibility quotient”. Sexual and Relationship Therapy, 18, 161-170.
On pages 162&163 can be read “Each of the 25 questions in the Compatibility Indicator was scored on a 0,1,2,3,4 scale, according to the degree of difference between the couple’s responses (ignoring the direction of the difference). These scores were summed across items to produce total discrepancy scores (out of 100). These raw scores were then converted to “compatibility quotients” (CQs), aiming towards a mean of 100, and an standard deviation (s.d.) of 15. The idea was that such a transformation would benefit from widespread familiarity with the Intelligence Quotient I.Q. distribution. As with the IQ, scores above 100 represent higher than average compatibility and those below 100 are lower than average, with a working range of around 70 to 145.
The CQ scoring procedure was developed as follows: A random sample of 2,159 adults (18 years plus; 1,234 men and 925 women), representing all UK individuals who had completed the Compatibility Indicator on-line as entry to membership of a dating site, was scored for all male-female pairings. This provided a sample of 1.14 million discrepancy scores, averaging 23.58 (s.d. 5.79). Raw “discrepancy” (D) scores were then transformed using the empirically derived formula
of CQ = 172 – (3 x D) to yield a mean of 101.32 (s.d. 17.05). Given the limits on the representativeness of our sample, this was considered sufficiently close to our target distribution not to choose any more complex formula, although we later decided to set a “floor” to the CQ at a score of 28 (equivalent to the natural ceiling of 172, the level at which responses of the two members of a pair are identical to all 25 questions).”
The compatibility result between prospective mates is only a 3 figure integer number. The matching equation is as simple as C.Q. == 172 – (3 x D)!!!
Online Dating Sites manage huge databases (from 100,000 persons’ profiles up to 20 million persons’ profiles; sooner perhaps more!!!) so they will need more power calculation / more precision in matching algorithms than they are actually using, with advanced math & statistical equations. Actual multiple/simple linear regression equations used to find prospective mates are not enough. As for Rasch model (the Item Response Theory star logistic regression method) seems to be an excellent tool to measure human variables= physique, personality, intelligence, social background, attitudes, habits and leisure preferences but not quite useful comparing one profile’s result to others.
e.g.: PATTERN#Xprofile::4.9.5.4.1.3.4.9.7.8.7.5.6.7.9.10
PATTERN#Yprofile::8.6.3.5.2.9.6.9.3.6.7.5.5.7.7.4
the probability that #X is similar to #Y == 62.18120805%
At References it is also cited the milestone paper
“Change Assortative Mating and Marital Quality in Newlyweds: A Couple-Centered Approach”, February 2005 Journal of Personality and Social Psychology
http://www.apa.org/journals/releases/psp882304.pdf
Page #19 of PDF whitepaper
Page #322 of magazine
“…………………………..
Couple Similarity and Marital Quality
….. The observed similarity–satisfaction correlations SUGGEST that similarity on personality related domains was strongly associated with satisfaction…
…………………..”
At the United States and Europe there are many Psychologists (PhD) researching on THEORIES OF ROMANTIC RELATIONSHIPS DEVELOPMENT. I had contacted most of them, and some of them replied the emails I had sent. “It seems that what is important in attracting people to one another may not be important in making couples happy.”
As many actual online dating sites CANNOT change nor adjust their matching algorithms, the whitepaper written by Drs. Thompson, M., Zimbardo, P. & Hutchinson, G. (2005). “Consumers are having second thoughts about online dating: are the real benefits getting lost in over promises?” [Industry Report]. Dallas, TX: weAttract.com; also cited at references, will be metamorphosing to “Consumers are GETTING TIRED OF ACTUAL ONLINE DATING SITES, with low precision/ low reliability matches”
I think it would be interesting to launch a new online dating site whose main matching core promotes “High probably, ALMOST complete similarity on MOST personality variables is necessarily the best way to go for prospective NORMAL persons / mates building happy couples” (I am heavily moving forward in that way) and to compare results with other online dating sites.
Kindest Regards,
Fernando Ardenghi.
Buenos Aires.
Argentina.
ardenghifer@gmail.com
Thanks for the comments, Fernando!
My push in this industry is for professional testing standards when it comes to the tests online dating sites offer as “scientific.” It comes down to a simple question: “How can anyone have faith in research findings, or for that matter, their implications, if it is not clear whether the measures used are even psychometrically valid?”
Unfortunately, the traditional way in which researchers construct and validate their assessment instruments relies on classical test theory — and this goes for virtually all compatibility tests and the theoretical and applied research on which they are based. The usual approach within classical test theory is to develop a test consisting of a number of items, and to assume that the sum of the scores received on the test items defines the latent trait (e.g., cognitive impairment or improvement). Such techniques essentially treat all items as equivalent and ignore the possibility that some items may be more difficult (or, diagnostic of individuals exceptionally high on the particular construct) than other items.
Another major flaw of this approach is that summed scores do not provide linear (i.e., interval-level) measures of the underlying trait. In addition, the standard raw score approach does not recognize that some items may be biased such that subjects with identical trait levels receive systematically different scores. This might be the case for instance when women (or younger respondents) endorse some questions more (less) often then do men (or older respondents) with equal trait levels.
Thus, traditional scaling approaches offer no indicators of the true internal validity of respondents’ scores. Furthermore, response biases can systematically distort research findings thereby leading to spurious correlations or factor structures (Lange et al., 2000).
Professional tests, and indeed all relationship research, would be better if they were based on Rasch scaling techniques (for reviews, see Wright & Stone, 1979; Bond & Fox, 2001). In brief, Rasch scaling allows for the quantification of the response consistency of items and persons, thus yielding important diagnostic information that goes far beyond deriving just “scores” or “measures.” Specifically, it is possible to determine the fit of each response record and to identify malingering or otherwise deviant respondents. Additionally, Rasch scaling enables researchers to identify item and response biases. Although item biases within a test are generally considered undesirable as they distort the estimates of individuals’ and groups’ (average) trait levels, such biases can actually be integrated into the into the test and used as an additional diagnostic tools. Finally, the Rasch approach provides “fit” information that enables researchers to judge the internal validity of respondents’ answers.
Accordingly, misfit is a property of the data, rather than the model. As Bond and Fox (2001) explained, “the goal is to create abstractions that transcend the raw data, just as in the physical sciences, so that inferences can be made about constructs rather than mere descriptions about raw data” (p. 3). Researchers are then in a position to formulate initial theories, validate the consequences of theories on real data, refine theories in light of empirical data, and follow up with revised experimentation in a dialectic process that forms the essence of scientific discovery.
It’s too bad that there are very few experts in the world when it comes to applied Rasch scaling. The leader in Rasch scaling as it applies to compatibility testing (long-term or casual relationships) is Dr. Rense Lange of Integrated Knowledge Systems (IKS: http://www.iknowsys.org). His mathematical foundations combined with the expertise of my team produced the True Compatibility Test, as well as the new Sexploration test (i.e., sexual compatibility).
But even those testing products are essentially outdated by my (and his) standards. You see, the best part is… technologies and research now exist to produce even more sophisticated and precise compatibility matching (and other types of tests) that companies using classical test theory methods can’t duplicate. And, these data collected can then be used for accurate psychographic profiling and targeted marketing!
Compatibility testing, indeed all online testing, is still in its infancy — but just wait and see what’s coming to the right company that’s seriously interested in the state-of-the-art of applied psychology 🙂 !!!
Thx, Dr. Jim
References
Bond, T. G., & Fox, C.M. (2001). Applying the Rasch model: Fundamental Measurement in the Human Sciences. Mahwah, NJ: Lawrence Erlbaum.
Lange, R., Irwin, H. J., & Houran, J. (2000). Top-down purification of Tobacyk’s Revised Paranormal Belief Scale. Personality and Individual Differences, 29, 131-156.
Wright, B. D., & Stone, M. H. (1979). Best Test Design. Chicago, IL: MESA Press.
More info could be seen at:
http://www.rasch.org/memos.htm
http://www.rasch.org/rmt/index.htm
Institute for Objective Measurement, Inc.
http://www.rasch.org/
TIP: Rasch model is excellent to MEASURE human variables. Unfortunately, Rasch model is not enough good to COMPARE a group of measurements with others;
E.G: Suppose you have PATTERN#Xprofile::4.9.5.4.1.3.4.9.7.8.7.5.6.7.9.10 and PATTERN#Yprofile::8.6.3.5.2.9.6.9.3.6.7.5.5.7.7.4
Which is the probability that #X is similar to #Y?
Using the method I had invented: 62.18120805%
Kindest Regards,
Fernando Ardenghi.
Buenos Aires.
Argentina.
ardenghifer@gmail.com
Thanks, Fernando
Actually, if all scores/measurements are Rasch scaled, then IRT is the valid way to equate scores across tests/subscales. In fact, only Rasch scaling/IRT can do this — this is why professional tests like the LSAT use IRT: they need to have different tests every year but they also need to have those test scores to be equated.
IRT is not just for measuring variables — it allows an objective way to model complex data!
Thanks,
Dr. Jim