Biostatistics #1-3 (Types of Study)

Relationship vs Causal Effects

 

<Correlational Method (상관연구법)>

 

: 둘 이상의 변인들 간의 상관(관계)에 초점을 맞추어 진행하는 연구방법.

: two (or more) variables are observed to see if there is an association 

• 한 변인의 변화에 따라 다른 변인(들)에서 일어나는 변화의 정도나 방향성을 밝히기 위해 사용됨.
• Correlational studies CANNOT demonstrate cause-and-effect relationships
• look for consistent patterns in the data to providence evidence for a relationship between the variables
• Ex) 상사의 유머 감각과 직원들의 직무 만족도

---

<Experimental method (실험법)>

: 실험 대상자들을 둘 혹은 여러 개의 집단으로 나눈 후, 인과관계의 원인이라고 추정되는 변수를 각 집단에 다르게 조작하여 그 결과가 집단들간에 다르게 나타나는지를 봄으로써 변수들간의 인과관계 규명.

: goal is to establish cause-and-effect relationships between variables 

• Requires Manipulation & Control 
  • Manipulation (조작): Researchers manipulate one variable by changing its value.
  • Control (통제): Researchers must exercise control over the research situation to ensure that other confounding variables do not influence the relationship being examined.
  • Aim: To demonstrate that changing the value of one variable (independent variable) will cause changes to occur in the second variable (dependent variable).
  • In an experiment, 독립 변수는 조작되는 변수이고, 종속 변수는 관찰되는 변수임.
• Experimental & Control Group
  • Experimental Group: receives treatment.
  • Control Group: receives no treatment (or receives a placebo).
• Confounding Variable (혼재변수)
  • uncontrolled 
  • a source of error in interpretation (해석의 오류 원인)
  • Ex) participant variables (e.g., age, sex, and intelligence) & environmental variables (e.g., lighting, time of day, weather conditions).
• Control confounding variables
  • Random assignment (무작위 할당) to both conditions, and to levels of the independent variable
  • Matching: use equivalent groups or environments
  • Holding constant of variables (e.g., age; only use 10-year olds in study).

---

<Quasi-experimental method (준실험법 or 유사실험설계)>

: 모든 관련 변인을 완전히 통제하거나 조작할 수 없는 실험설계

: no actual manipulation, but groups are defined by 'natural' variations, either between subjects or over time. 

: one that looks a bit like an experimental design but lacks same level of control.

• 엄격한 실험절차를 밟지 않더라도 실험설계와 유사한 방법을 적용하여 연구에 필요한 자료를 수집할 수 있는 자연적 상황이 많다. 
• Aspects that might not be controlled are:
  • Not having a control group with which to compare results
  • Not randomly assigning participants to control or experimental group
  • Not having control over the intervention (e.g., using an intervention that is in place in practice
• Example
  • a study looking at sex differences isn't technically an experiment because you can't assign sex.

---

 

출처: Youtube > Academic Lesson, 네이버 지식백과, 구글