2 Data and Distributions Flashcards
What are the 2 types of studies
- Observations studies
2. Experimental studies
What is an observaitonal Study?
An observationsal study has no intervention by the researcher. Examples: surveys and polls
What is an experimental study?
The researcher imposes treatements to observe changes in an experimental study. Example: plant growth trials.
What studies can establish cause and effect?
Only experiments can establish cause and effect.
What is a lurking variable?
A lurking variable is one that affects both variables.
When a lurking variable is present you cannot establish causation.
What does it mean when a lurking variable is present?
When a lurking variable exists, there can be an association, but not a causation between two variables.
What can estaiblish cause and effect?
Only a good experiment can establish cause and effect.
What is an observational study?
An observational study involves observation only.
What is an experimental study?
An experimental study imposes treatments on individuals to observe responses.
What is a control group?
A control group is the group not receiving the treatment(s)
What is a placebo?
A placebo is a fake treatment.
What is a double blind experiment?
Neither the subject nor the people administering the treatments know who is receiving which treatment.
What are the principles of good experimental design?
- Randomisation
- Control
- Replication
- Blocking
What is randomisation in an experiment?
Randomly assign subjects to treatments.
What is control in an experiment?
Control is a group not administered treatements. Control is necessary for the placebo effect another potential lurking variables.
What is replication in an experiment?
Replication is achieved by having more than one subject or experiment or by repeating the whole experiment to reduce the impact of chance variation.
What is blocking in an experiment?
Creating groups of similar cases in order to extract better information.
What is CRD?
Completely randomised design. All subjects allocated at random to the treatment groups.
What is a blok design?
IN a block design an SRS selected from each block in the population. A block is a similar group of cases.
What is a matched pairs design?
A special type of block design where two treatments only are compared.
Which 3 tupes of graphs can be used to display a quantitative variable?
- Stem and Leaf Plot
- Histogram
- Boxplot
What is an outlier?
An outlier is a value (or values) that is significantly removed from the main body of data.
Which descriptions of centre and spread are best for a skewed distribution?
Skewed distribution - use measures that are not affected by outliers.
Centre = Median
Spread = IQR (interquartile range)
Which measures are best for a reasonably symetric distribution?
Centre = mean Spread = standard deviation