Social Science MethodsFall 2010
Soyoung JungKevin Balster
Melvin Hale
Social Science MethodsFall 2010
Bivariate analysis is the analysis of two variables; one independent and one dependent.
Ex. The effect of library instruction on library use
It is undertaken to see if there is any association between the independent and dependent variables.
Selection of a method depends on the level of measurement used for each variable : the nominal, ordinal, and interval or ratio levels
Common Methods of Bivariate Analysis
Contingency Tables (Cross- Classification)• often used when both variables are nominal or ordinal (categorical variables)• do not require assumptions about the nature of the association
Correlation Coefficient• used for linear associations between two
numerical variables • measures the strength and direction of the
linear relationship• scatter-plot graph (scattergram)
Social Science MethodsFall 2010
GenderBorrowing materials
Using Public Computers
Studying Other Total
Male 30 40 20 10100
(100%)
Female 35 30 30 5100
(100%)
Primary Reason for Library Use (n=200)
Example of a Contingency Table
Example of a Scatter plot graph
Social Science MethodsFall 2010
Bivariate hypothesis – proposing a relationship between two phenomena.
Null Hypothesis: H0 - No relationship between the two variables.
: the values of the two variables are independent of one another, the correlation
coefficient r = 0)
Alternate Hypothesis: H1 – There is a relationship between the two variables.
This hypothesis can only be accepted after the null hypothesis is rejected.
The usual goal is to reject the null hypothesis (H0), to conclude that the variables are not independent of one another.
Social Science MethodsFall 2010
Associations in Bivariate analysis
Determine the existence of an associationwhen one changes as the other changes
Describing associations:
Strength strong – moderate - weak
Direction Positive : the variables change in the same directionNegative : the variables change in the opposite direction
Nature (Pattern) – Linear or Curvilinear
Social Science MethodsFall 2010
http://mste.illinois.edu/courses/ci330ms/youtsey/scatterinfo.html
An LIS exampleAn LIS example
Health Information Ties: Preliminary Findings on the Health Information Seeking Behavior of an African-American Community
Ophelia T. Morey
The study was performed to see if there was a correlation between gender, age, or relationship strength and how people sought out health information.
Social Science MethodsFall 2010
Independent Variables: GenderAgeRelationship Strength
Dependent Variable: Source of health information
Note: While there are three different independent variables in the study, they were all analyzed independently of each other so the analyses were bivariate.
Social Science MethodsFall 2010
Hypotheses:
H0: Age/Gender/Relationship Strength have no affect on where individuals search for health information.
H1: Age/Gender/Relationship Strength have an affect on where individuals search for health information.
Social Science MethodsFall 2010
The following data were collected:
Gender (nominal)Age Group (ordinal)Strength of relationship to information source (ordinal)Information source (nominal)
All analyses were performed using a χ² test.
Social Science MethodsFall 2010
Findings:
There was no correlation found between gender and where respondents received their health information.
A correlation was found between age and where respondents searched for health information.
A correlation was found between the closeness of a relationship and where respondents got their information.
Social Science MethodsFall 2010
Bivariate Analysis Bivariate Analysis SummarySummary
The analysis of empirical relationships among Pairs of Variables, an ObjectiveExplanation.
Social Science MethodsFall 2010
Two Types of Two Types of ExplanationsExplanations
Idiographic Explanations: Multiple factors affect a specific outcome, with limited generalizability.
I missed my flight.
1) Alarm clock failed, 2) Car was on “E”, 3) Had to park in the remote lot, 4) Everybody at security refused to go through the body scanner, 5) The flight was overbooked.
Social Science MethodsFall 2010
Two Types of Two Types of ExplanationsExplanations
Nomothethic Explanations: A few causal factors impact a class of conditions or events, and can be explained with an economy of terms, usually applied as groups.
- People who are usually Late for their appointments.
- People who make Excuses.
A Bivariate analysis is a nomothethic explanation.
Social Science MethodsFall 2010
RelationshipsRelationships
Causal: Changes in one variable (independent) affects the other variable (dependent). The cause takes place before the effect. This is called a correlation.
Spurious: A statistical coincidence shown to be caused by a third variable.
Social Science MethodsFall 2010
Pros and Cons Pros and Cons
Advantages:
Quantify results – Simplify Relationships – Make predictions
Disadvantages:
Over-simplify results – Identify spurious relationships (Type 1 or Type 2 errors)
Social Science MethodsFall 2010
Best Used for:Best Used for:
Exploratory Research & Less Complex Situations
Testing the Ice
Social Science MethodsFall 2010
Thank You
Social Science MethodsFall 2010