88 Cards in this Set
Front | Back |
---|---|
Descriptive Statistics
|
Summarize tenancies and characteristics of data
|
Inferential Statistics
|
Use sample statistics to make inferences about the population from which the samples were drawn
|
What are Inferential Statistics used for?
|
Hypothesis testing
|
distribution of sample means
|
collection of sample means for all the possible random samples of a particular size (n) that can be obtained from a population (aka sampling distribution of means/M) (ex. sampling distribution)
|
When can the Distribution of Sample Means be normal?
|
1. If the original population is normal
2. If the sample size n>30
|
Standard Error of the Mean
|
is the SD of the DSM (SD/SqrRt(n))
|
Central Limit Theorem definition
|
For any population with a mean and SD, the distribution of sample means for sample size n will approach a normal distribution with a mean and standard error as n approaches infinity
|
central limit theorem meaning
|
established statistical rule that tells us that if we were to take an infinite number of sample size n from a population of N members, the means of these samples would be normally distributed
|
Hypothesis test
|
is a test of the null hypothesis (Ho)
|
Alternative Hypothesis
|
Is the hypothesis the researcher believes in.
|
Correct Decision
|
Ho is true and retained it
Ho is false and you rejected it
|
Type II Error
|
Null Hypothesis is false and you retained it.
|
Type I Error
|
Null Hypothesis is true and you rejected it.
|
Alpha
|
The probability of making a Type I error when the null hypothesis is true.
|
Z score
|
Standard score specifies how many standard deviations an observation is above or below the mean. (sample mean - mean of DSM)/ standard error.
|
Beta
|
Probability of making a Type II Error
Retain Ho when false
|
1-tailed test
|
H1 specifies the direction of the effect
maen of group A = Mean of group B
X and Y are correlated
look for "more, better, improve,increase, less, worse"
|
2-tailed test
|
H1 has no direction
X and Y are correlated
mean of Group A not = to mean of group B
IV has "an effect"
|
Hypothesis Testing
|
State null and alternative hypothesis
Set you decision criteria (alpha)
collect data
compute test statistics
reject or retain Ho
State conclusion in ordinary language
|
Null Distribution
|
The population distribution from which the sample is drawn if Ho is true.
|
Alternative Distribution
|
The population distribution from which the sample is drawn if Ho is false
|
Z-Test answers...
|
How likely is the sample mean came from a population
|
P-value
|
What the probability that the sample mean happened by chance if the null hypothesis is true.
Reject Ho when p < Alpha
|
Normality Assumption
|
DSM is normal if:
If n< 30;Population must be normal
If n> 30; DSM is normal
|
Independence Assumption
|
each member of the sample is selected independently
Random selection
|
Statistically Significant
|
That is a result in unlikely to occur merely by chance
rejected Ho
different than meaningful
|
Z-test
|
How likely is a sample w/ a mean of __ cam from a population w/ a mean of __ and a SD of __
|
Critical Z
|
Where the critical region(s) begin
depends on Alpha
1-tailed critical region = bottom 5%
reject Ho when Observed Z < Critical Z
|
Observed Z
|
The actual score
(Xbar - sample mean)/ standard error
|
Statistical Power
|
Probablility of correctly rejecting the null hypothesis when false; ONLY has power when Ho is false
|
Power
|
1-B
Lower power the higher the chance of retaining Ho
|
The alpha lvl
|
larger Alpha more power
lower alpha lower power
However, if its too big its error prone; Too much power= too much data
|
Sample size (n) and power
|
larger n = smaller the standard error= more power
however, larger n takes more time and money
|
Effect size and power
|
Larger effect = more power (distance from 1 mean to the next, makes them skinnier)
Give a bigger dose to increase, less overlap
|
1-tailed v 2tailed
|
1-tailed is better
More power if in correct direction
|
SD and power
|
Larger SD =Larger standard error = LESS POWER
the smaller the SD the more power
|
Point Estimate
|
A single number that is used to estimate a population parameter.
precise but almost always wrong
|
Interval Estimate
|
an interval of numbers around the point estimate, within which the parameter value is believed to fall
"95% chance Meu is btw 2 and 8"
|
Confidence Interval
|
z-test rearranged
an interval calculated using sample statistics to contain the population parameter within a certain degree of confidence
|
1.96
|
Z-score of Alpha in a 1-tailed hypothesis
|
Slope
|
(x)(slope)+(intercept)
change in Y/ change in X
|
Intercept
|
The value of y when x=0
|
Y-hat
|
The predicted value of Y
|
Residual error
|
vertical distance between Y and line
e=Y - Yhat
|
Variance
|
correlation coefficient squared r^2
|
3 Measures of Error
|
Variance not explained by X=1-r^2
Sum of the squared residuals
Standard error of the estimate
|
r^2
|
Percent variance in Y accounted for by X
Regression
|
Standard error of the estimate
|
average distance of Y from the predicted value of Y
SD of errors= Standard error
|
confidence interval
|
an interval estimate calculated using sample statistics to contain the population parameter, within a certain degree of confidence
|
Margin of error
|
The distance from the point estimate to the upper/lower bound of a confidance interval
|
Cohen's d
|
The distance between 2 means in SD units
|
b1
|
slope
|
b0
|
intercept
|
Sum of the Squared Errors
|
sum of the squared differences between each score on the criterion variable and its predicted score
|
In the Conditional Distribution
|
Mean= Yhat
SD= Standard error of the estimate
What Y would look like if people scored X
|
One-tailed & Ho>H1
|
Reject Ho when Observed Z < Critical Z
|
One-tailed & H1>Ho
|
Reject Ho when Observed z > Critical z
|
Nominal
|
A set of categories without numerical values
eg. colors
|
Ordinal
|
A set of category with numerical order but unequal distances
|
Interval
|
A set of categories with numerical order and equal distances.
|
Ratio
|
A set of categories with numerical value, equal intervals and a TRUE ZERO.
|
Skew
|
Asymetric distribution of a certain kind
Tail points in (+/-) direction.
|
Mode
|
Most frequently occuring thing/score in a data set
|
Parameters
|
descriptive statistics used to define apopulation
|
Statistics
|
describe samples
|
Central Tendancy
|
A representative # from a distribution
|
If positivly skewed...
|
Mode < Median <Mean
|
If negativly skewed
|
Mean <Meadian <Mode
|
Variability
|
Degree of spread (scatter in a distribution).
|
4 Roles of variability
|
clarify precision/ representativness of central tendancies
Identify outliers
compute other stats
theoretically interesting on its own
|
Range
|
(high score)-(low score)
only two most extreme scores measured
very suseptable to outlires.
|
Variance
|
SD^2
|
Z Score
|
How many SDs the raw score is above or below the mean.
SD=1
M=0
maintains shape of original distribution
|
Restricted Range Problem
|
True nature of correlation is hidden due to restricted range.
|
Effect Size
|
Standadized measure of the magnitude of a variable's effect on another (Cohen's D)
|
Cohen's d
|
Distance of 2 means in terms of SD
0.2 small
0.5 medium
0.8 large
|
Degrees of freedom
|
the number of scores that are free to vary when estimating a population parameter from a sample
df=N-1
Higher # df bigger the peak
|
t-distribution
|
Shape depends on df
Peaked at top (critical regions)
fatter in tails
as n...> infinity t-dist becomes normal
|
1-Sample t-Test
|
Use if population is unknown but SD is known
critical t> critical z
|
Within-Person Pairs
|
Longitudinal (same variable different times)
Within-Person Contrast (different variables same person)
|
Dyads
|
Same variable different people
must be comparable or have defined relationship
must be distiguishable
|
2 Types of Dyads
|
Distinguishable (particular role; parent/child)
Indistinguishable (don't have differences; couples, coworkers)
|
In Expected Mean Difference Scores...
|
If Ho is true then m1 = m2
Therefore Mean Difference is zero
|
Paired Sample t-Test
|
is a 1-sample t-test on difference scores
|
Assumptions of t-Tests
|
Normality Assumption
Independent assumption
Homogeneity of Variance Assumption (Independant t-Test)
|
Normality assumption
|
That the DSM is normally distributed
|
Independanct Assumption
|
No scores are related to other scores
|
Homogeneity of Variance Assumption
|
Whether two populations have equal variances (levenes test)
|