PSY 1004 Study Guide - Final Guide: Scatter Plot, Dependent And Independent Variables, Literature Review
Topic 7
Correlational Research
●Intended to demonstrate the existence of a relationship between two or more variables
●Does your age influence how much you sleep at night?
●Using variables that already exist in the world
●Relationships can be described- not explained
●There is no attempt to manipulate, control, or interfere with the variables
●Correlation does not equal causation
Experimental Research
●Demonstrates a cause-and effect relationship between two variables
How could one investigate a cause and effect relationship (i.e. using an experimental design)
1. Obesity and television watching are positively associated
a. Independent variable: TV ; 1hr a day, 3hrs, 6hrs
i. After a year measure their weight
b. Dependent variable:
2. The more money people have is negatively associated with free time
a. Independent variable: time 1hr a day 3hrs a day
b. Dependent variable: $$$
Evaluating Data
●Scores can be shown in a table
○Scores in each pair are identified as X and Y
○Data can be presented in a list showing the two scores for each individual
●Scores can be shown in a scattergram graph (scatterplot)
○Individual score is a single dot with a horizontal coordinate (X) and a vertical
coordinate (Y)
●Correlation coefficient ranges from -1 to +1
○Tells you the direction and strength (or consistency) of relationship
○Direction of relationship
■Positive - both increase or decrease together
■Negative- both increase or decrease opposite of each other
○-1 strong 0 weak 1 strong
●Goodness of fit
○If r
is close to 1 (or -1), the model is considered a “good fit”
○If r
is close to 0, the model is “not a good fit”
○If r = ±1, the model is a "perfect fit" with all data points lying on the line
○ If r = 0, there is no linear relationship between the two variables
●Technically, the correlational strategy can be done with more than 1 group
○Differential Research (we know it as non-experimental, a Category 3 strategy)
■Sometimes, 1 variable is numerical, the other is nominal
■Pearson correlation- point biserial correlation
■When both variables are non-numerical (nominal) - display data in a
matrix the chi-square test
●Strengths and weaknesses of the Correlational Design
○(+)
■describes relationships between variables
■nonintrusive - natural behaviors
■High external validity
○(-)
■Cannot access causality
■Directionality problem - can’t determine cause & effect
■Third-variable problem - C is causing A and B
■Low internal validity
○Not good internal validity but good external validity
Applications
●Regression: using correlational strategy to predict 1 variable from another
●Predictor variable: the 1st variable
●Criterion variable: the second variable (being explained)
●Goal: find equation that produces the most accurate predictions of Y (criterion) for each
value of X (predictor)
●Multiple regression: a statistical procedure for studying multivariate relationships
○Useful when you think there is a 3rd (4th, 5th etc) variable problem
■Controlling influence of potentially confounding variables
●Correlational designs are useful to determine reliability and validity of measurements
○Example: test, retest reliability
○Construct validity (convergent and divergent)
Thinking Critically about Correlation
●Spurious correlations
○Note* not a scatter plot
○Why do you think these spurious correlations exist?
○Spurious (adj): outwardly similar or corresponding to something without having
its genuine qualities
Descriptive Research
●Goal: Measure a variable or set of variables as they exist naturally
●inductive/bottom
○“Ad hoc”- requires flexible process of theory development
●Observational Research: systematically recording behavior
○Concerns:
■Demand characteristics from researcher
■Reactivity from subjects
■Reliability
○Advantages:
■Real world observation of behaviors, good internal validity
■Easy to conduct and cheap
■Provide evidence for rare behavior
■Large amounts of data
○Disadvantages:
■Certain behaviors are hard to observe
■Observer’s bias
○Quantifying behavior - frequency, duration, interval method
○Direct observation- watching people’s actual behavior
○Indirect observation (physical trace)- observation of not the actual behavior but
something that infers the behaviors, the evidence of the behavior
○Naturalistic- careful record keeping, privacy, doing observation in the most least
intrusive way
○Participant Observer Research (ethnography)- research interacts with
participants
■Ethical issues: privacy, objectivity, interrupting normal behaviors of group
○Contrived observation (laboratory)- do not have to wait for behaviors to occur,
less natural
○Put in strengths and weaknesses of each from powerpoint.
○Archival research- measures behaviors or events that occured in the past
■Concerns:
●No real possibility of informed consent
●Ruling out alternative hypotheses may be difficult
●Surveys/Questionnaire
○Obtains a description of a particular group of individuals
○Qualtrics, survey monkey, amazon mturk
○Considerations
■Response rate (% complete)
■Method of sampling
Document Summary
Intended to demonstrate the existence of a relationship between two or more variables. Using variables that already exist in the world. There is no attempt to manipulate, control, or interfere with the variables. Demonstrates a cause-and effect relationship between two variables. How could one investigate a cause and effect relationship (i. e. using an experimental design: obesity and television watching are positively associated a. Independent variable: tv ; 1hr a day, 3hrs, 6hrs i. After a year measure their weight: dependent variable, the more money people have is negatively associated with free time. Independent variable: time 1hr a day 3hrs a day a: dependent variable: 5605$ Scores can be shown in a table. Scores in each pair are identified as x and y. Data can be presented in a list showing the two scores for each individual. Scores can be shown in a scattergram graph (scatterplot)