Class Notes (837,284)
COMM 2002 (65)
Lecture

# COMM November 11, 2013.docx Premium

3 Pages
85 Views

School
Department
Communication Studies
Course
COMM 2002
Professor
Heather Pyman
Semester
Fall

Description
Bivariate Analysis • Ordinal Measures (Strength measures) -> find the measure that is appropriate for the lowest level of the two measures • Can use ordinal and ordinal or ordinal and interval • Limitations on ordinal level data Looking at nature of the relationship • What the direction of the relationship is -> positive or negative? What does this mean? • How strong is the relationship • Is it statistically significant • Does it support or not support our hypothesis • We look at a correlation coefficient that is interpreted in a different way • Are not just concerned with what the relationship is within the sample, but based on the relationship, is it likely to be a relationship in the population? o Look at Chi-square that assesses if there is a null relationship • Look at frequency of the observed value • Compare them to what we would expect there to be if there is not relationship between independent and dependent variable • Correlation is dependent on rows and columns in this table • Cannot tell strength of relationship from Chi-square, have to use Kramer- V co-efficient to figure it out o Expected Values • Multiple row total by column total and dividing by the sample size  Simple tables with observed values  What numbers would get into the calculation to calculate the expected values • Value you expect there to be • The more discrepancy there is between observed and expected, the more likely there is a relationship because it is an observed value o Cramer's V is non directional o Nominal variables are just categories • Not a positive or negative number, it is just a strength measure  Would actually need to go back to the table and look at the column percentages in the columns of the independent variable o 0-.25 = weak, o 2.6-5.0 = medium o 5.6-10 = strong o 1 = perfect correlation -> if it is graphed, it would be a perfect straight line Ordinal Level Data • Logic = Proportional Reduction of Error (PRE) o All of these measures tell us whether knowing the value of one variable reduces the chance in making an error in predicting the value of other variable o Calculation is a ratio of how good/accurate we would be in predicting the value of the dependent variable o How much more accurate would we be if we introduced an independent variable? • Understand what direction the scale is going -> what is at the low/high end of the scale • Have direction and strength • Measures which tell us whether knowing the value of one variable reduces chance of making an error in predicting the value of the other • Correlation value which varies between -1 and +1 • Indicates direction and strength o Example: Looking at Television viewership in Canada [In the thousands] • Channel WEST EAST Central CTV 60 10 25 25 GLOBAL 30 20 5 5
More Less

Related notes for COMM 2002
Me

OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Join to view

OR

By registering, I agree to the Terms and Privacy Policies
Just a few more details

So we can recommend you notes for your school.