Observational Methods Flashcards
Observational Methods
Useful in Psychology
Cannot determine the causal relationship between two factors.
Cannot know the direction of causality.
What are the different types of Observational Methods?
Case Studies
Naturalistic Observations
Correlational Studies
What is a Case Study?
Indepth study of one or a few individuals.
What are the Pros/Cons for Case Studies?
Pros:
Huge amount of information about the person(s)
Cons:
Genarilzability/External Validity
What are Naturalistic Observations?
Observing people in their common, everyday environment.
What are the Pros of Naturalistic Observations?
Pros:
Can be used with non verbal participants (infants)
Only method that looks at the “real world”
What are the Cons of Naturalistic Observation?
Cons:
Cannot determine cause
Infrequently occurring events
Reactance (Hawthorne) Effect
What is the Reactance (Hawthorne) Effect?
Biasing of the participants responses because they know they are being observed.
Ex: Yelling at your children
What is the variation of the Naturalistic Observations?
Participant Observation
What is Participant Observation?
Primarily observes participants but also interacts with people being studied.
What are the 2 variations of Participant Observation?
Observer as Participant
Participant as observer
What is Observer as Participant?
Researcher becomes part of the group being studied. Researcher is not really part of the group being studied.
What is Participant as Observer?
Living as part of the group. Emerged in the group being studied. Researcher becomes part of the culture by working and interacting with people being studied.
What are the Pros of Participant Observation?
Lots of info
What are the Cons of Participant Observation?
Very expensive
Very time consuming
Not necessarily accepted
Loss of objectivity (once you become part of a group you loose objectivity)
How are cultures like pyramids?
The top is fully accepted, they make the rules.
The middle is fully accepted, but they just follow the rules.
The bottom is hanging on, not fully accepted.
What is Time Sampling?
Observations of a behavior at different time periods.
Ex. how aggressive preschoolers are in AM and PM
Brothers, Lovers, Others game.
What is Situation Sampling?
Observations of the same type of behavior in different situations.
What are the two ways in which you might record behaviors you study?
- Qualitative
2. Quantitative
What is qualitative?
Tell me a story of what happened.
Narrative description of what happens.
What is qualitative?
Give me the counts, #s.
Ex: how many red lights were run.
What is inter-observer reliability?
The extent to which observers agree.
Must be 85% agreeability between observers.
Must have at least two inter-observers/interrators, then calc how many times the 2 agree divided by how many times they could have agreed.
What is a Correlational Study?
Investigates whether two or more variables of interest are meaningfully related.
Ex: # of drinks had and rated attractiveness of fellow bar goers.
Scatter Plot
Dots are actual data points, mathematically fit on a line.
Line of Prediction
The mathematically score we would have mathematically predicted for this individual, on the Scatter Plot.
How do we interpret Correlational Data?
•Use the Correlational Coefficient (r)
Correlational Coefficient (r):
A numerical estimate of the strength and direction of relationship.
Correlational Coefficient (r): Strength
Strength of the relationship indicated by the number (range 0 to 1).
•Weak Relationship = smaller #
(Dots are scattered more away from the line of prediction)
•Strong Relationship = larger #
(Cluster more together around line of prediction)
Correlational Coefficient (r): Direction:
Direction of the relationship indicated by the +/-.
Correlational Coefficient (r): The Perfect Relationship
[If r is 1.00 you will always be right but it never happens.]
r: tells us how good we would be at predicting an individual is standing on one variable by knowing how they are standing on the other variable.
Ex:
Var 1: Drinking & Accidents
r=.45 *Weaker correlation
Var 2: Phone Use & Accidents
r=.68 *Stronger correlation
The Direction of the Relationship:
+ Positive + = r < 1
+ Positive + = r < 1
Variables go in the same direction:
Both Up or Both Down
Ex: Less drinks = Less accidents
The Direction of Relationship:
- Negative - = r > 1
- Negative - = r > 1
Variables go in opposite directions:
One goes Up, One goes Down
Ex: More heat = Less starbucks purchases
The Direction of the Relationship:
ø No Relationship ø
No pattern or cluster around line of prediction.
The Direction of the Relationship: Curvilinear
As one Variable goes Up the other Variable goes Up & Down at the same time.
U or Horseshoe Shape Line of Prediction.
•r represents a linear relationship, dots make a line.
If you have a curvilinear line the r is going to underestimate the relationship between two variables and assumes normality.
Drawbacks of Correlational Design
1. CANNOT DETERMINE CAUSALITY: •Third variable confound •Reciprocal causation •Temporal ordering 2. Statistical Issues: •Linearity of relationship •Restriction of range •Size of the sample
CANNOT DETERMINE CAUSALITY:
Third Variable Confound
•Evaluating the strength by absolute
#. The +/- ONLY tell you what side.
•Reduces the size of r.
CANNOT DETERMINE CAUSALITY:
Reciprocal Causation
•Both variables cause each other.
•Reduces the size of r
•If restrict the range of values for either or both variable r will inaccurately estimate the true relationship of two variables.
Ex: You are STRESSed so you DRINK more, DRINKing causes STRESS.
CANNOT DETERMINE CAUSALITY:
Size of the Sample
- Size of the sample won’t change r size.
* The smaller sample size the less likely we are to find r statistically significant.
Critical Thinking
Ex: Ice-Cream sales are strongly correlated with crime rates.
Does ice-cream cause crime?
No. There must be a third variable.
The Third Variable
Just because two things occur together doesn’t mean they are casually related.
Another variable could explain the relationship.
How to Interpret a Correlational Design
Write the Hypothesis: either:
•Non-Directional (two-tailed) hypothesis:
•Directional (one-tailed) hypothesis
How to Interpret a Correlational Design:
Non-Directional (two-tailed) Hypothesis
There IS a relationship between X and Y. (One effects the other)
•H0: r = 0 (Null)
•H1: r ≠ 0 (Alternative)
How to Interpret a Correlational Design:
Directional (One-tailed) Hypothesis:
Positive Relationship
- Positive Relationship: Higher levels of X are associated with higher levels of Y. One goes Up, One goes Down.
- H0: r ≤ 0
- H1: r > 0
How to Interpret a Correlational Design:
Directional (One-tailed) Hypothesis:
Negative Relationship
Negative Relationship: Higher levels of X are associated with lower levels of Y. One goes Down, One goes Up
•H0: r ≥ 0
•H1: r < 0
How to determine Statistical Significance of Correlational Studies
•DF: Degrees of Freedom: n -2: (n = number of pairs), to find p value on SPSS or Table. •If p is < .05, significant than Hypotheses: •One-tailed (directional) •Two-tailed (non-directional)
If you have significance then compute
Effect size measures
•r2 = Square of r.obtained
•Coefficient of Determination:
What % of the variance in X is explained by Y: 100X… = %
Special Considerations
•Pearson's product moment correlation: *Only use for Interval or Ratio data •Spearman's: *Only use for Ordinal (rank) data •Point Bi-Serial: *Only use for One Dichotomous= Yes/No or On/Off •The Phi-Coefficient: *Only use if Both Dichotomous= 2 Options Only
