do not assign subjects to treatment or control groups. Some subjects have condition whose effects are being studied: this is the treatment group. the other subjects are the controls
what adjustments were made for confounding? were the sensible?
Association
Observational studies can establish association (one thing is linked to another)
association may point out causation (if one group is exposed to chemicals they should be sicker than the other)
association does not prove cause: (confounding)
Counfounding
Observational studies can be misleading about cause and effect because of confounding
Confounding is a third variable associated with exposure and disease.
In observational studies a confounding factor can sometimes be for by comparing other groups which are relatively homogeneous with respect to the factor.
Control Groups Vs Observational
Questions to ask
Ask questions
Were there any control groups at all?
if so were historical control are contempary used?
how were the assigned to groups under the control of an investigator (controlled experiment) or out the control of an investigator (observational experiment)?
If controlled were they assigned via (randomized controlled) or at the discretion of the investigator?
Weakness of Observational Studies
The great weakness is confounding: randomized experiments minimized this problem