Lecture Notes: Experimental Design, Study Types, and Critical Evaluation
Experimental Design Basics
Two groups needed in an experiment: control and treatment.
Control group receives a placebo. The placebo is something that looks identical to the treatment (pill, shot, etc.) but has no active ingredient.
Placebo effect can be powerful because of psychological or other non-specific effects.
Population (number of subjects): more subjects provide more power to detect a true effect; larger sample size leads to more reliable conclusions.
Example mentioned: a large COVID vaccine study with thousands in both the treatment and control groups to provide enough power.
If you have only one subject, the study generally has no meaningful power.
Therefore, in studies, aim for a large population to increase the power of the study to validate the theory.
The treatment group is the group receiving the actual treatment.
The last step in the described sequence is the method (procedural aspect) of the study.
Placebo and Study Design Implications
Placebo can be used to control for psychological effects; the notion that people can lose weight in placebo groups is highlighted.
A hypothetical company could sell placebo (sugar pills) as a weight loss aid, illustrating the strength of placebo effects in weight loss studies.
In summary, placebo control is a key component of many experimental designs to isolate the effect of the actual treatment.
Data and Measurement
Data must be numeric and quantifiable so that statistical analysis can be performed.
Measurements need to be in numbers to enable numerical analysis and inference.
Population size affects the statistical power of the study.
Clarification question: "Which group receives the placebo?" Answer: the control group.
Placing the placebo in the control group ensures the comparison isolates the treatment effect.
Visualizing placebo: the placebo should look the same as the treatment to maintain blinding.
Study Designs: Blindness and Randomization
Single blind study: the human subjects do not know which treatment they receive, but the researchers do.
Double blind study: neither the subjects nor the investigators know which treatment is being given; a third party is responsible for randomization.
Double blind design is more unbiased, though more expensive.
In pharmaceutical research, double blind studies are typically required to ensure objectivity.
After data collection, results are published and subjected to expert peer review.
Publication and Source Evaluation
Large studies (e.g., COVID vaccination studies) are published after rigorous review.
There are different types of studies covered during the semester; the instructor will discuss published papers.
A good strategy for finding credible information is PubMed (funded by NIH) – free access to scientific papers.
When evaluating information found online (TikTok, Facebook, blogs): check if the source is legitimate and based on scientific studies; PubMed can help verify.
Types of Studies and Research Pathways
Epidemiology studies: study of populations, looking for patterns and associations; often rely on self-report (questionnaires, interviews) and data encoding.
Key characteristics: population-based, not necessarily in a controlled setting, reflects associations, not causation.
Human intervention studies (clinical trials): the gold standard for testing interventions in humans; demonstrate causation; essential for regulatory approvals (e.g., FDA).
Expensive due to costs of human subjects, housing, care, and controlled environments.
Outcomes can reflect causation if well designed.
Laboratory studies: conducted in university or research facilities; may involve animals or early-stage drug development; not directly generalizable to humans.
Drug Development and Why Drugs are Expensive
Drug development involves multiple stages:
Animal studies (preclinical) to assess safety and mechanism.
Phase I trials (small group) to assess safety and dosing in humans.
Phase II trials (larger group) to assess efficacy and side effects.
Phase III trials (large scale) to confirm efficacy and monitor adverse reactions.
Regulatory review and approval by agencies (e.g., FDA).
Timeline: development typically takes about 10 years, often ranging from 8 to 15 years.
Cost rationale: high upfront costs for preclinical and clinical testing, long development timelines, and high rate of failure.
Example: Ozempic cost can be around ext{
1000}permonth,illustratinghighout−of−pocketcostsfornewtherapies.</p></li><li><p>Whyearly−stagedrugsfail:resultsinanimalstudiesoftendonottranslatetohumansduetogeneticdifferences;translationtohumansisnotguaranteed.</p></li><li><p>Theprocessaimstoensurethatonlysafeandeffectivedrugsreachthemarket,butitisexpensiveandtime−consuming.</p></li></ul><h3id="27b5bdee−c982−4bdf−b067−e5d06ecbede8"data−toc−id="27b5bdee−c982−4bdf−b067−e5d06ecbede8"collapsed="false"seolevelmigrated="true">WhyAnimalStudiesAreLimited</h3><ul><li><p>Animalmodelscanshowpromisingresults,buttranslationtohumansisnotguaranteed.</p></li><li><p>Example:adrugmayreduceweightinratsbuthavenoefficacyinhumans;thisdiscrepancycanleadtolargefinanciallossesandmisdirectedresearch.</p></li><li><p>Thespeakermentionsthatanexpensivecompanyinvestedheavilybasedonanimaldataandfoundlittletonoclinicalbenefitinhumans.</p></li></ul><h3id="e60c4e3f−247b−479f−90d1−884547824011"data−toc−id="e60c4e3f−247b−479f−90d1−884547824011"collapsed="false"seolevelmigrated="true">CriticalThinking:EvaluatingClaimsinSocialMedia</h3><ul><li><p>WithwidespreadclaimsonTikTokandotherplatforms,verifyinformationbycheckingcrediblescientificsources.</p></li><li><p>PubMedisrecommendedforfindingpeer−reviewedstudiesandcredibleevidence.</p></li><li><p>Differentiatingbetweenlegitimatescientificfindingsandmarketingorsensationalclaimsisemphasized.</p></li></ul><h3id="04697cd1−50c6−4f25−9292−fdfbffda6d4a"data−toc−id="04697cd1−50c6−4f25−9292−fdfbffda6d4a"collapsed="false"seolevelmigrated="true">CreatineandNutritionTopicsMentioned</h3><ul><li><p>Creatineisapopulartopic;claimsaboutrapidmusclegainordramaticeffectsarecommononsocialmedia.</p></li><li><p>Thespeakernotesthathigherdosesofcreatinearenotmagicallymorebeneficial;thereisnuanceandaneedforevidence.</p></li><li><p>Afuturediscussiononcreatineisanticipated,butnotcoveredindepthinChapter1.</p></li></ul><h3id="7cd0e21c−efdc−4ca5−86db−b856efe84fec"data−toc−id="7cd0e21c−efdc−4ca5−86db−b856efe84fec"collapsed="false"seolevelmigrated="true">Chapter2andCourseStructure</h3><ul><li><p>OnMonday,Chapter2willbegin;thefollowingweekwillcontinuewithChapter2andcohort1.</p></li><li><p>OnTuesday,studentswillworkonquestionswithneighborsoralone;focusonin−classcompletionofthefirsthomework.</p></li><li><p>Theinstructorreassuresthatpriorchemistryknowledgemayhelp,butstudentswillnotneedtobalanceequationsforChapter2.</p></li><li><p>Thefirsttwoweeksaremeanttogiveafeelforthecourseandhelpstudentsdecidewhethertocontinueinthisclass.</p></li><li><p>Aquickrefresher:nutrientdensityisdefinedastheratioofnutrientstototalcalories;definedas<br> ext{Nutrient density} = rac{ ext{nutrients}}{ ext{total calories}}.</p></li><li><p>IftherearequestionsaboutChapter1,studentsareencouragedtoask.</p></li><li><p>Theinstructoremphasizesbuildingasolidfoundationandcomfortwithbasicconceptsbeforemovingforward.</p></li></ul><h3id="4d68ebed−839f−4906−861e−a2b3049510c0"data−toc−id="4d68ebed−839f−4906−861e−a2b3049510c0"collapsed="false"seolevelmigrated="true">NutrientDensity:DefinitionandFormula</h3><ul><li><p>Definition:Nutrientdensityistheratiobetweentheamountofnutrientsandthetotalcaloriesinafoodordiet.</p></li><li><p>Mathematicalrepresentation:<br>\text{Nutrient density} = \frac{\text{nutrients}}{\text{total calories}}.$$