Daniel Gartenberg: The Role of QS in Scientific Discovery

Today’s breakout session preview for the upcoming QS conference comes from Daniel Gartenberg, organizer of the Washington DC QS meetup group. Here is Daniel describing his session “Is QS Science? The Role of QS in Scientific Discovery:”


Do you believe in the power of using Quantified Self to solve some of science’s toughest questions, but have concerns about the validity of QS data?  There is actually a long scientific tradition of N=1 studies (i.e. studies conducted with a single participant).  Additionally, there are various advantages of N=1 studies, such as repeated, longitudinal, and naturalistic data.  These advantages of N=1 studies enable the personalization of treatments because they can take into account individual differences.

Yet N=1 studies are atypical in current scientific research.  We will be discussing why the scientific community frowns on N=1 studies, and how we can alleviate some of the scientific community’s concerns regarding QS.  This involves understanding what makes something ‘Science.’ Additionally, this will involve identifying threats to validity when conducting studies and QS research.  Threats to validity include, but are not limited to: Mortality, History, Maturation, Treatment Fidelity, Treatment Interaction, Compensatory Rivalry, Regression Towards the Mean, and Reactivity.

If done correctly, QS can be a new standard in scientific rigor, but this will require a concerted and collaborative effort by the QS community that will involve developing a system where QSers can post their data to the cloud and have the data aggregated and analyzed across individuals (e.g. curetogether).  The potentials and challenges for creating a QS database will be discussed.

Come to this breakout session if you are trying to make sense of your QS data, are interested in the scientific method, are interested in data analysis techniques, or want to create systems and tools that make QS data meaningful to the general population.

This entry was posted in Conference and tagged , , , . Bookmark the permalink.

4 Responses to Daniel Gartenberg: The Role of QS in Scientific Discovery

  1. Roger Vilardaga says:

    I’m looking forward to see your talk. People quickly forget that the laws of reinforcement were discovered using single case designs…! I just wanted to add a reference if anyone’s interested in starting to learn more about statistical tests of single case designs: Edgington, E. S., & Onghena, P. (2007). Randomization Tests. Boca Raton: Chapman & Hall.

  2. Jan Aerts says:

    This is a great idea. Unfortunately, I won’t be able to make it to the conference so will not be able to contribute to the discussion there. For future reference and to have something to refer to when trying to convince other researchers (and funding agencies!) of the value of N=1 studies it would be *very* useful to have a paper describing this issue; a paper/commentary in a peer-reviewed journal (in biological sciences, e.g. BMC).

  3. Regression to the mean is not a bad thing in large sets of QS data.

    For any particular QS variable, we’d like to be able to calculate means (and SD ranges) for every individual compared to various time frames in their past (such as last week, month, year or their lifetime avg) — so they can be alerted if/when they upload any values that fall outside their historical normal range (eg. mean +/- 1 SD).

    But mean estimates won’t be reliable if the number of available measurements is small, and so we probably should not even allow comparisons with prior history to be made until sufficient “n” for that variable have been uploaded.

    We should not have this problem when calculating means for cohorts of similar individuals who share particular characteristics (eg. if i want to compare myself with other healthy 50-60 year old men), because we’ll have so many more data points for each variable.

    This is true even if most of the individuals in the cohort only contributed one or two measurements, which would be insufficient to calculate their own means and ranges.

  4. Curtis A. Bagne says:

    Demonstration 2 at http://dataspeaks.com/resources/bagne_handout.pdf describes how many group Randomized Controlled Trials can be conducted as coordinated sets of advanced design N of 1 RCTs.

    The current gold standard for clinical trials can be described as reductionist Group Average Science (GAS). GAS uses group averages to evaluate treatments and their effects largely one by one. Some problems with this convention are that many patients take more than one drug, each drug has more than one effect, beneficial and harmful effects are not balanced against each other scientifically, patients are complex and adaptive systems, and no one patient is average.

    The alternative is more holistic Individualized Science that applies to both individuals and groups.

    Anyone interested? Please contact me at cbagne@DataSpeaks.com.

Leave a Reply to Curtis A. Bagne Cancel reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Notify me of followup comments via e-mail. You can also subscribe without commenting.