Tags

  • Published on

    Research and statistics in the collective unconscious

    Cognitive Dissonance Research Research Engineer Statistics

    Research Engineer will always be free

    Fighting the good fight for statistics and research

    It behooves me to make Research Engineer free to the world.  Statistics and research often cause a "knee-jerk" reaction of anxiety and cognitive dissonance.  I am reminded of this on a daily basis, because I live it on a daily basis.  

    It seems to me that statistics and research, in the most basic applied sense, have acquired a negative connotation within the collective unconscious.  That is scary because researchers communicate the findings of their research using statistics! Other researchers and clinicians are supposed to be able to interpret and generalize the research methods and statistical findings of the studies, denoting possession of the necessary competencies and experience to do so.  

    Research Engineer will always be free, because research and statistics belong to everyone.  Names like Aristotle, al-Haytham, Bacon, Descartes, Bayes, Gauss, Pearson, Fisher, Spearman, Bonferroni, Tukey, Cox, and Cronbach have come before me and have invented the empirical and statistical methods presented in this website.  All I did was take their methods and build THE FIRST decision engine designed to avert the aforementioned anxiety and cognitive dissonance. The brilliant men and women that came before me deserve to be recognized and their methods, which we use in everyday life to make important decisions, should be held in a higher esteem within the collective unconscious.

    I tell my clients that when they feel cognitive dissonance associated with research or statistics, I will be standing there in their unconscious mind, ready to help.  It always gets a reaction, but they always come back and they always receive my best efforts.

    Now, when anyone around the world feels that "knee-jerk" reaction to research and statistics and they want to retreat into their unconscious minds, Research Engineer will be waiting, always free and always able to help.  

    Click on the button below to continue.
  • Published on

    Establishing causal effects

    Blinding Causal Effects Experimental Designs Intention-to-treat Random Assignment Random Selection Randomized Controlled Trial Systematic Review

    Necessary conditions for establishing causal effects

    Experiments require random selection and random assignemnt

    Everyone conducting research is looking for associations between predictor, confounding, and outcome variables. People will often talk about wanting to know the correlation, relationship, or effects between variables.  These types of research questions are answered using observational research designs.

    However, when researchers are attempting to explain causal effects, there are certain criteria that have to be met.  

    1. An experimental design is the only method by which a causal effect can be inferred. This means that study participants are selected at random and randomly assigned to treatment groups. Any differences at baseline are thought to occur by chance with random selection and random assignment (whereas with observational designs they are due to selection and observation biases).  

    2. Analyses have to be conducted in an "intention-to-treat" fashion. This means that all study participants are analyzed in the original groups that they were randomly assigned to at the beginning of the study.

    3. Blinding can bolster the validity of causal effects, but is not necessary for establishing causal effects. Blinding can reduce observational biases that can occur in experiments where human beings are studied.    

    4. As much as possible, all pertinent demographic, clinical, and confounding variables related to the association between the primary predictor and primary outcome variable should be accounted for in experiments. True causal effects (in reality and practice) are always multivariate in nature. Clinicians do not make treatment decisions based on one form of evidence, they want as many forms of evidence as possible. The statistics that establish the efficacy of treatments should reflect the multivariate nature of treatment decisions.

    5. There has to be a sufficient amount of time or follow-up for an outcome. True causal effects are found when the temporal aspects (etiological, prognostic, time) of developing an outcome are accounted for methodologically and statistically.

    6. One clinical trial that shows valid evidence of a causal effect and meets all of the aforementioned criteria is STILL NOT ENOUGH! The results of the most methodologically sound randomized controlled trials or true experiments need to be aggregated in systematic reviews, syntheses, and synopses of syntheses. Pooled effects account for more variance in the general population and strengthen the "causal" understanding of causal effects.  
  • Published on

    Number Needed to Treat is an important epidemiological calculation

    Absolute Risk Reduction Control Event Rate Epidemiology Experimental Event Rate Number Needed To Treat

    Efficacious treatment effects

    The magnitude of a positive treatment effect

    Number Needed to Treat (NNT) is an important epidemiological calculation that shows how many people have to be treated to preventing a future bad outcome. In a perfect situation, the NNT would be 1, meaning that every person that you treat will not get the bad outcome in the future.

    NNT is calculated using other epidemiological calculations:

    Control Event Rate (CER) - Proportion of the control group with the outcome
    Experimental Event Rate (EER) - Proportion of the treatment group with the outcome
    |CER - EER| is the Absolute Risk Reduction (ARR) or the effect of the treatment in comparison to a control treatment.
    Then, in order to calculate NNT, just use (1/ARR).  

    With smaller values of ARR, or small treatment effects, the NNT will increase meaning that more people have to be treated to prevent a future bad outcome. Higher values of ARR show a strong treatment effect, fewer people will have to be treated to prevent a bad outcome because it is so efficacious.
  • Published on

    Case series are the lowest level of clinical evidence

    Basic Science Case Series Effect Size Observational Research Pilot Study

    Case series are used to study rare outcomes and generate hypotheses

    Yield measures of effect size and test methodologies

    Case series designs yield the lowest form of observational evidence. Researchers choose a series of cases in the population that share some sort of similar characteristic and then they analyze pertinent predictor, demographic, and clinical factors associated with the outcome in the group of cases.  Case series designs are at times also called pilot studies.  

    Case series designs are often employed in basic science and pre-clinical research.  They are useful for generating hypotheses, effect sizes for future power analyses, and studying extremely rare outcomes.  

    Case series designs are an excellent choice for novice researchers looking to get their feet "wet" in empirical pursuits. They also prove their worth when evidence-based measures of effect do not exist in the literature.  Running a small pilot study or case series can yield important measures of effect. 
  • Published on

    Follow Research Engineer on Facebook, Twitter, Google, YouTube, Tumblr, Pinterest, Instagram, LinkedIn, Flickr, and Vimeo

    Facebook Isomorphism Research Engineer Twitter YouTube

    Social media integration

    The social journey begins for Research Engineer

    Hello everyone!

    Please follow Research Engineer on Facebook and Twitter!  New content is on the way!

    I am currently revising the manuscript for my dissertation on isomorphism.  After six years of an uphill battle for isomorphism, it is finally accepted with revisions!  Fingers crossed!

    I'm so excited about integrating social media into Research Engineer.  My SEO friends and colleagues say it is time to get a YouTube channel and put this "mug" to work.  Hahaha, get ready world!

    Thanks for the support, everyone!

    Dr. H

    Click on the buttons below to follow Research Engineer!
  • Published on

    Between-subjects one-sample median test

    One-sample Median Tests

    One-sample median tests

    Simple and effective with an a priori hypothesis

    When it comes to running statistics on one sample, an a priori hypothesis is a necessity for making proper inferences. Many survey instruments lack fundamental psychometric evidence to back up the constructs and/or items it is intended to measure.  When a Likert-scale type response set or an ordinal outcome is to be assessed statistically in one sample, the one-sample median test is used.

    In order to run the test, researchers must specify where along the ordinal continuum that they hypothesize that the population mean exists.  The one-sample median test then compares the observed median to the hypothesized median and the p-value is interpreted.

    Due to limited precision, accuracy, and variability in ordinal outcomes, it behooves researchers to use either 5-point, 7-point, or higher level Likert scales.  With more options, more unique variance can be accounted for the in the analysis and statistical power is increased.  One-sampled tests possess more statistical power than other between-subjects statistics because there is only one group being analyzed, no other independent groups are included.