Lesson 01 - Research Methods
To believe any results of any study we need to ensure
- Good sample size
- Representative sample
- Sound Methodology
Constructs
- We have constructs i.e. things that are difficult to measure. e.g. Intelligence, Effort, Hunger.
- To properly measure them we need a operational definition i.e. a definition that allows us to properly measure them
Lurking variables
- Whenever analyzing data we should think about externeous factors that we may not have thought about. These are called lurking variables.
- If we hold more things constant then our confidence in results becomes more as we are bringing down lurking variables
Population and Samples
- All possible people taken together is a population
- Metrics related to population -> population parameter
- $\mu$ is the average of population
- A subset of people from population is a sample
- Metrics related to sample -> sample statistics
- $\bar{x}$ is average of sample
- sampling error = $\bar{x}$ - $\mu$
- Although $\bar{x}$ is not the same as $\mu$ it can give us an interval in which $\mu$ lies as long as we have large enough random unbiased sample. It is our best estimate of $\mu$
Correlation does not mean causation
- McDonald's example of world peace
- Observational studies are good enough for showing relationships
- For showing causation we need to do controlled experiments
Surveys
- easy to conduct, inexpensive
- important to word it properly, cannot rely on it
Controlled Experiments
- People receive actual/placebo treatment
- People are not told which of these groups they are in (blind)
- Researchers measuring the effects are not told either (double blind)
Why Sleeping May Be More Important Than Studying
Summary
We discuss that for doing any research there are certain points that must be taken care of. Then we go and discuss some terminology related to research methods including constructs, lurking variables and statistical definitions. We conclude by going over few research methods - surveys and controlled experiments.