Reply To: Why do the system seems a falicious loop?

Index Forums General Psychology Why do the system seems a falicious loop? Reply To: Why do the system seems a falicious loop?

Auburn
Keymaster
  • Type: TiNe
  • Development: l--l
  • Attitude: Adaptive

Heya Star,

I share your same concerns with the data collection and I am aware of these problems. There are even more problems than the ones you listed, and fixing all of them would indeed be hard. Not just hard, but it would require an entirely different approach from the very beginning of the data collection pipeline. Potential issues I can think of, just off the top of my head include:

  • Bias in the selection of videos
    • Example: The samples CT has are largely two groups: celebrities, and people interested in personality psychology. These two groups don’t represent everyone, or an even cross-section of the population.
    • Example: Some people don’t like being on camera, so they won’t make videos, and then you only get videos of people who are willing to go on camera.
    • Solution? A fully randomized selection process, to get an equal cross-section of the population.
  • Unknown level of subject variation between videos
    • Example: A subject may look very different at different times of day, at different weeks, or in different moods.
    • Example: A  subject may change vultology when speaking different languages.
    • Solution? Do a controlled study on a few samples where you record them over the course of weeks in different moods, different languages, etc — to see how similar or different their vultology is, and whether a reliable baseline vultology can be extracted from this, despite the variation. This is similar to testing hypothesis 1.
  • Video setting and quality differences
    • Example: Sometimes there is an interviewer, sometimes there isn’t.
    • Example: Lighting and resolution may vary, as well as camera orientation.
    • Example: Setting (outside/inside), crowds, one-on-one, may factor into different performance.
    • Solution? Video settings would need to be the same/identical for all subjects in the study. One way to do this is to bring samples into the same room for the interview, using the same recording equipment and camera angles.
  • Demographic Differences
    • Example: We don’t know if different cultures have different tonal or articulation habits, which change expression.
    • Example: We don’t know if different racial groups have facial anatomy that obscures or changes the appearance of signals.
    • Solution? Control for demographic and racial variables by doing specific studies on specific demographics first, then compare them to each other for similarities and differences.

As you can see, the ideal solution would involve a randomized study group that comes in for an interview in the same setting with the same camera equipment, and for all the interviews to be analyzed by factoring in demographic data. Additionally, each sample should be recorded in each language that they know. They should also come in for an interview a week later, for several weeks, to compare the similarities or differences in different times.

My wish-list is very long when it comes to this, and I would love to be able to do all of these things. The problem is, as you must know, this sort of process requires a research grant and can easily pass $50,000-$80,000 USD in cost. If I had a grant like that, the “garbage data” problem would surely be solved!

Anyhow, there is a reason why the wiki says in the main page, under “Scientific Research“:

The CT hypotheses and theory tenants have not yet been tested under a formal, scientific setting. Currently the cognitive typology community is preparing a proof of concept involving a database of 1,000 public figures, to demonstrate the efficacy of the hypotheses. The community aims to use this proof of concept to prompt investigation and collaboration from the scientific community.

The current data collection process is filled with inadequacies, I wholly agree. But that’s not the current point, because this data isn’t meant to be absolute. The point of this endeavor is to put together 1,000 videos into a proof-of-concept pilot study in order to demonstrate that there is reason to justify that $50,000+ grant money. And it is also to help be a pathfinder for how we might go about creating controlled studies around vultology and psychology.

We have already learned a lot in CT, and a lot of the mistakes. For example, too much attention to anatomical features has lead to misreadings. Ethnicity does contribute to confusion in signals. So far we haven’t done the most study on native vs learned languages, but that is another candidate for where we could look more into things.

Lastly, I think it’s crucial to remember that CT is not funded by any university or institution, and is an independent research project. Naturally, when compared to academic work, it’s not up to par, but that’s a mismatched comparison. On the other hand, when compared to the surrounding typology community, it is doing far more than anyone else I’ve seen to aim for experimental control, repeatability and transparency.

Thank you for your video though! I do hope these and all possible problems are fixed in the end. I’m also a perfectionist and am bugged by these things too, but we gotta go one step at a time. 🙂

  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.
  • This reply was modified 1 month ago by Auburn.

© Copyright 2012-2020 J.E. Sandoval
SEE HERE FOR MORE INFORMATION

DISCLAIMER

The content on this site is not
intended for medical advice, diagnosis,
or treatment. Always seek the advice
of your physician or other qualified
health provider with questions you
may have regarding a medical condition.
For more information visit this link.

SHARE: FACEBOOK, SUPPORT: PATREON