Lecture 5- Study Design 2

Question 1

What is the purpose of analytic studies?

Answer 1

To test hypotheses

Question 2

What are the two important concepts behind analytic studies? (what do you need to do?)

Answer 2

• Replication: you can’t base findings on one subject. Replication allows us to determine true effects from chance effects/ natural variability
• Control: control groups provide context/ a comparison point for looking at your area of interest. With them you can accurately determine what would have happened without manipulation and therefore see the true effect your independent variable had on your experimental group.

Question 3

What are the two types of analytic studies?

Answer 3

• Experimental : the researcher actively manipulates the conditions. All other factors must remain the same so can isolate the effect. This is the best way to study causation. Randomization ensures that the effects of unmeasured factors are equalized across groups as everyone has an equal chance of being selected for each group.
• Observational: the researcher does not intervene. Instead they just observe and collect data from a naturally occurring process (groups self allocate)

Question 4

What is the difference between random sampling and randomisation?

Answer 4

Random sampling happens at the level of choosing the sample from the population. Randomization on the other hand is when you allocate individuals into groups for different conditions after your population has already been selected.

Question 5

When would you use an observational study versus a experimental study?

Answer 5

Sometimes it is not ethical to intervene and allocate groups manually. E.g. can can’t tell a group of people to smoke so that you can measure it effects (experimental approach). Instead you find a pre-existing smoking and non-smoking group (observational approach)

Question 6

What is a consequence of observational studies in terms of group selection?

Answer 6

Because individuals have effectively chosen their own groups other factors may come into play and become confounding variables. For example, with the smoking study it could be that those who choose to smoke are less athletic/ health conscious anyway and so naturally have worse health (it is not necessarily the smoking that is the problem)

Question 7

What are the two types of observational studies and how do they differ?

Answer 7

• Cohort study: this is prospective. For this we are looking to the future.
• Case control: this is retrospective. For this we are asking about past happens e.g. in smoking find people with heart disease and people without and then ask how many of you are smokers (go the other way around)

Question 8

What is an issue with case-control/ retrospective studies?

Answer 8

Information obtained from participants may not be accurate because you are asking them to remember stuff from the past. It could be possible they have forgotten information or they may lie/ want to remember themselves in the best light.

Question 9

What is the primary experimental analytic study used? Why is it so good? Where may it fall down?

Answer 9

• Randomized control trial (RCT)
• This is the gold standard because both randomization of groups and control via a non-experimental group can be achieved. Ideally the experiment would be double blind. This way we can sure results are due solely to the intervention we implemented.
• A disadvantage is that depending on the circumstance a study like this may not be ethical or feasible

Question 10

Describe a cohort study/ What are the advantages and disadvantages of this method?

Answer 10

• Participants are selected before outcome has developed (prospective) and are followed over time to determine development of outcome. Information is collected about exposures at baseline and during follow-up. Cohort studies do have a control group, but allocation is not randomized so there is always the potential for confounding.
• Advantages are that it is the closest observational study to a randomized controlled trial. It is also good for examining common outcomes. Finally it can evaluate the effect of exposure on multiple outcomes.
• Disadvantages are the long duration needed if the outcome takes a long time to develop after exposure. In addition if the outcome is rare, the number of participants needs to be very large. It is very hard (if not impossible) to remove all the effects of confounding.