Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
53 Cards in this Set
 Front
 Back
Hypothesis

Testable statement about the nature of social reality; reasons for relationships.


Measurement

Assigning a unit of analysis to an attribute on a variable.


Unit of analysis

Person or thing from which data is collected.


Variable

Set of logical attributes that are of interest to the researcher.


Conceptualization

Process of formulating and clarifying concepts.


Operationalization

Describes the research operations that will specify value/category of variable on each case.


Indicator

Observable measure. Imperfect representation of concepts.


Ratio

Implied relation to 1.


Proportions and Percentages

fa/N
fa/Nx100 

Rates

Make values comparable to each other.
fa/D x 100,000 

Inferential statistics

Moving from description to explanation.


Population

Entire amount of subjects: large group actually interested in.


Sample

Selection from population  don't have access to entire population.
Infer from samples to population. Should be representative. 

Hypothesis testing

The extent to which samples reflect true numbers of population.


Dependent variable

What we are trying to explain. Variable that is measured. Depends on independent variable.


Independent variable

What is manipulated; what is causing dependent variable.


Nominal

Categories of this variable are mutually exclusive. No mathematical properties.


Ordinal

These variables can be logically ranked, but have no true mathematical properties.


Interval

These variable have true mathematical properties. No true zero point.


Ratio

These variables have mathematical properties and have a true zero point.


Sampling distribution

A hypothetical distribution of all possible sample outcomes for a statistic.
Bridge between sample & population. 

Central Limits Theorem

Many statistics have sampling distribution that is approximately normal.
On average, the sample mean is the same as the population mean. 

Areas under the normal curve

68% within 1 standard deviation.
95% within 2 standard deviations. 

Statistical significance

Unlikely it happened just by chance. Difference big enough/rare enough.


Parameter

Variable related to population.


Two tasks of classical inference

1. Estimate magnitude of parameter.
2. Test specific claims about magnitude of parameter. 

Confidence Interval

Estimate from standard deviation of statistic.
1.96 

Point & interval estimation

Using statistic as estimate of the parameter is risky. Unknown variability. Instead, create interval estimate.


Alternatives to chisquare

Phi: 2x2 table only
Cramer's V Lambda: PRE measure 

Proportional Reduction of Error

PRE measures compute prediction errors in two different situation:
a) when only raw totals are used for prediction b) when an independent variable is used for prediction 

Concordant and Discordant pairs

most PRE measures for ordinal variables based on assessment of pairs of cases.
Con: same directionality Dis: opposite directionality 

Goodman & Kruskal's Gamma

only uses cases with concordant and discordant pairs


Ttests

All types of ttests are designed to compare sample means.
Comparison of 2 "groups". Based on t distribution. 

Independent samples ttest

What does "independent" mean? Score on test variable for members of 1st group are not dependent on scores of 2nd group.
Standard form. 

Independent samples t test: assumptions

1. test variable is normally distributed in each of the 2 populations
2. the variances of the normally distributed test variable are equal 3. cases have been randomly samples from population 4. scores are independent 

Levene's test for equality of variances

Tests assumption of equality of variances.
Null=equal variances assumed. 

Paired sample t test

What we use when assumption of independent samples is violated.
Same person measured twice (per/post), or when pairs of subjects are matched in some way. 

Paired sample t test: assumptions

1. Difference scores are normally distributed.
2. cases have been randomly samples from population. 3. the difference scores are independent from each other (among sample) 

One sample t test

Used when comparison mean is:
a) unknown b) arbitrarily chosen 

Test Value

Key consideration of one sample t test.
a) midpoint of test variable b) average based on past research c) chance level of performance 

One sample t test: assumptions

1. Test variable normally distributed
2. cases have been randomly sampled 3. scores are independent of each other 

MannWhitney U test

Nonparametric substitute for equal variance t test.
Used when assumption of normality not valid. 

Significance testing

Test specific claims about magnitude of parameter. Interested in parameters that indicate relationship. Idea of null hypothesis: reject or fail to reject.


0.05 alpha level

Willing to risk 5% chance of wrong answer. If probability of observed relationship happening by chance is <5%, reject null.


Type I Error

Rejecting a null hypothesis that is true (saying there is a relationship when there is actually none).


Type II Error

Failing to reject a null hypothesis that is not true (saying there is not a relationship when there actually is).


Onetailed test

If we have directionality.
Critical value is 1.64 

Basic ideas of chisquare

Are two variables related to one another?
Null hypothesis: the two variables are independent. 

Logic of chisquare

Observed and expected counts.
Reject null if observed counts are sufficiently different from expected counts. 

How to conduct chisquare

1. Calculate marginals
2. Calculate expected counts 3. (observeexpected) ²/expected 4. then, sum across all cells 5. calculate degrees of freedom 6. compare observed w/ expected, determine if can reject null 

Calculation of expected counts

row total x column total / grand total
what we would expect to see if the two variables are independent of each other 

Degrees of freedom (chisquare)

(row1)x(column1)
How many pieces of information would I need in order to fill in the remainder of the cells? 

Limitations of chisquare

Expected cell counts must be greater than 5.
Often can't tell us relative strength of relationship. 