Lecture 6 Flashcards
When deciding on participant samples, what are 23main questions we should ask?
- who will participant in my experiment?
- how can I generalize my results to?
- how large should my sample be?
What is a convenience sample?
a sample that is convenient for the researcher.
How might a convenience sample lead to different results than going out into the world to collect participants?
by going out into the community, you can get a different perspective than if you only got university students as a sample. Uni students are often more financially stable, young ,etc . The findings may not generalize. If you go out to the community, you get a more representative sample.
studies investigating norms, beliefs and preferences tend not to generalize very far because culture really shapes those things.
if what you are studying is assumed not specific to people of certain cultures ages etc, then it shoudl be fine for you to use convenience samples.
What is external validity?
whether or not the results can be generalized to people of a different group
How large should a sample be if using random assignment?
we should consider if we will have the statistical power to observe a relationship to
How does the number of participants effect risk of participant variables preventing the ability to observe a relationship?
with smaller numbers of participants the risk is higher and the risk decreases as participants increase. as your study size gets larger, participant variables will be es likely to have an impact Note that the risk never gets to zero, instead the graph depicting this relationship would asymptote and level off. AKA at some point the number of participants has diminishing returns.
What is the size of effect we observe a ratio of?
manipulation strength and noise
is noise the same as confounds?
noise is not the same as confounds because confounds have to be systematically different across conditions. We are using random assignment so this isn’t a confound.
What do we mean by noise?
unpredictable behaviour or random things that could change participant behaviour
What do we mean by manipulation strength?
the way we operationalize our variable
Are stronger manipulations easier to see?
yes
to controlled experiments reduce noise?
yes.
does control have an affect on our ability to see effects?
yes. the more we can control things like distractions, the more we can see the effects if they are there.
Should we try to investigate strong relationships first? Why?
yes.
More likely to determine if your hypothesized relationship is present
Follow up with more subtle operational definitions of IV
What is a straightforward manipulation?
Straightforward manipulations:
- Simply showing participants research stimuli
- Survey research, most computer-based research
- Advantages: Simple; cost-effective; deception not needed
