Pols 306 Exam 2 Flashcards
What is correlation?
A relationship between two variables.
What is a spurious correlation?
Relationships that are not genuine. When two measures of two or more variables are not in fact directly related.
What is a lurking variable?
A third variable that influences the correlation between two other variables. This is why we control for other factors.
What are the 5 R’s?
Refining, replicating, reducing, revealing, and reversing.
What are the seven elements of a good table?
A clear title.
Figures are in browns instead of columns.
IV is on the top and DV is on the left.
Percentages over Absolute Numbers
Write totals at the bottom of the column.
Round numbers when you can.
Footnotes for sources and important information.
In social science, how do you define significance?
It means that it is probably true.
What is the p-value?
Probability that the sample selected is false
How do you deductively determine the significance of the findings?
Testing important and substantive findings may suggest ways to modify the theory or seriously undermine the theory.
r
Correlation coefficient, the strength of the relationship between two variables
(1.0 = perfect positive relationship -1.0 perfect negative relationship)
Sensitive to outliers and extreme values
Marginals
The totals along the columns and rows
Expected Frequencies
The expected frequency is the null hypothesis. The Chi square is comparing observed observations with null hypothesis
Control Variable
The variable that has not changed
Substantive Significance
The substance of a finding takes on importance only to the extent that it contributes to information valuable for the understanding of political behavior or business or public policy.
p<.05
p<0.5 means that such an error would occur less than five times in 100 samples. *
Slope
How much your dependent value changes based off of one unit of the independent variable
Quartiles
4 parts of data
If you have 100 people 25 people would be in each quartile 2 inner quartile are closest to the mean and the 2 outer quartiles are farther from the mean because the graph flattens out
Observed Frequencies
the number of times a specific event or outcome actually occurs in an experiment or real-life situation
Revealing Effect
The original relationship changes are strengthened under certain conditions.
Outlier
Does not follow the pattern of data
M
Median
Correlation Coefficient
A measure of the degree to which two variables are linearly related
r2
the proportion of the variation in the dependent variable that is predictable from the independent variable
How close the data fits the line of best fit
Y intercept
X=0
Ecological Fallacy
When the researcher makes claims
about one lower-level unit of analysis that is based on data from some higher-level unit of analysis. Aggregate data to individualized assumption
Reversing Effect
Original relationship changes from positive or negative.
Type 1 Error
False Positive: Mistakenly claiming there is a relationship when there isn’t one (p-value of .05 indicates you are willing to accept a 5% chance of being wrong)
Panel Study
Asking the same people the same questions as they change over time
Type 2 Error
False Negative: When analysis concludes there is no relationship between the independent and dependent variables when in fact there is a relationship (particularly problematic in disease control)
Replicating Effects
Control variable has no effect on the original relationship.
Reducing Effect
Original relationship diminishes or disappears
S
Standard deviation
Cohort Study
Examines specific subpopulations, or cohorts, as they change over time. Typically, it is where demographers select a particular group (generation) and follow the changes that occur to those born at a certain year.
Chi Square
Level of significance for the relationship between variables measured at a nominal and ordinal level; compares actual observed values to expected values; compares observed frequencies with the null hypothesis
Z score
The number of standard deviations the value is away from the mean. It standardizes data so you can compare apples to oranges.
Refining Effects
The original relationship between the dependent and independent variables only holds up under certain conditions.
Spurius Relation
Relationships that are not genuine. When two measures of two or more variables are not in fact directly related.
t value
the ratio of the departure of the estimated value of a parameter from its hypothesized value to its standard error.
Lurking Variables
A third variable that influences the correlation between two other variables. This is why we control for other factors.
Reductionism
When claims about some higher-level unit of analysis are made based on data at some lower-level unit of analysis.
Small to big
