WEEK 9) Comparing research strategies Flashcards
Note these notes are very brief because I want to write an essay comparing the different research strategies.
Descriptive, correlational, quasi-experimental/non-experimental, experimental research strategies are all appropriate depending upon the questions asked
What is a monotonic relationship?
What would it look like?
monotonic: relationship is in the same
direction all throughout the distribution.
Linear on scatterplot or increases/decreases faster etc.
What does a non-monotonic relationship look like?
U shaped, relationship changes throughout scatterplot.
What is the p and r values in pearsons correlation?
p value: the significance/ probability that are results could happen if no relationship (null hypothesis is true.
r value: is the correlation, tells us the strength and direction ( -1 to +1)
What should you consider in this….
• A weak correlation (eg. r = 0.2) may be statistically significant (p < 0.05, so something systematic happening) but…..
but it may account for very little variance
• r squared = percent of variation accounted for by the relation between X & Y
e.g. accounts for 4% here.
what is r squared
percent of variation accounted for by the relation between X & Y
Whats the general cut off rules for reporting correlations.
e.g. whats a small, medium or large amt of variance explained?
General rule for reporting correlations:
Small: r = 0.1
Medium: r = 0.3
Large: r = 0.5
(CHECK WHAT THIS MEANS)
What is vulnerable to time related threats to validity?
Btn subjects or within?
within.
What is vulnerable to individual differences threats to validity, btn or within subjects?
btn
Differentiate btn: Extraneous and confounding variables
- Extraneous variables: all variables in a study other than IV & DV e.g. age, gender etc anything thats not iv or dv.
- Confounding variable: an extraneous variable that changes systematically along with the IV & has the potential to influence the DV
What are some ways one could control for confounding variables?
Holding a variable constant
• eg. time of day & performance; testing only females; testing psychology 1
students to keep education level constant
• Restricting the range
Matching values
• Balancing levels of the variable across treatment conditions
• eg. Balancing males/females across conditions by having the same number of them in each; counterbalancing test order
Randomisation
• Use of a random process to avoid a systematic relationship b/w variables
• Does not guarantee that there will not be differences between groups, just that they are not systematic
Random assignment
• Use of a random process to assign participants to treatment conditions
• Strategy to disrupt systematic relationships b/w variables
what is a circular explanation?
why is it bad?
when an effect is used to explain itself e.g. ‘My factory makes more goods than yours because it is more productive’.
DONT do this. Cause MUST precede the effect if you want to establish causality. “determinism”
Why might you use a mixed methods design?
- Complementarity: Develop deeper understanding of a research problem
- Development: Results from one study help develop or inform the other method e.g. interview someone with particular experience e.g. refugee experience, to get idea of kind of experiences, before you write your large scale survey.
• Initiation: Clarifying contradictions in findings
e.g. may have found something you didnt expect in surveyn and can follow up some strange answering with interviews.
• Expansion: To extend the breadth and range of a study might want to do quantitative follow up to get more data and expand more on top of your 10 interviewees say
