Inference Analysis Flashcards
3 types of Statistical Inferences
Parameter Estimation: estimating through Confidence Interval
Hypothesis Testing:
Comparing sample statistics with hypothesized population values
Tests of significant differences:
Comparing 2 or more groups
Parameter estimation
Estimating the range of a population value
Sample statistic: Mean or percentage
Standard error of the mean
SEM = {sample σ} / sqrt (n)
Standard error of the Percentage:
SEP = sqrt{p*∂/n}.
∂ = 1-p
p & Sp in DECIMAL!!!!!!!!!!!
Confidence Intervals:
Use z for Upper & Lower bounds - z*SEM (or SEP)
p+ that or p- that –> Bounds!
pop mean: x +/- z*SEM
pop %: p +/- z*SEP
pop percentage:
TAHT IS WHY WE NEED SEP OR SEM IN DECIMAL!!!!!
Hypothesis Testing fundamentals
Hypothesis Ho: Expected population values
A statistical procedure to reject or fail to reject Ho
Hypothesis Testing procedure
1) Test of the Hypothesized population parameter Value
Mean : z = {x~-µ} / SEM
As SEM = σ/ sqrt(n)
Percentage:
z = {p-µ} / SEP
2) Check z to determine
3) Directional Hypothesis (one tailed test)
Used when we have strong convictions
Two differences for directional hypothesis:
Null Hypothesis
Smaller critical z-value: 1.64 (90% Conference interval)
Testing for significant differences between two means
1) Differences between means with two groups (Ho; No difference between the two groups)
2) Differences between two means within the same sample
- Paired sample t-test
3) Small sample sizes; use the t-test
Differences between means with two groups
The test of difference between two sample means :
Ho = no difference between two sample means
The test of equal variances; levene’s test) - Ho: Same variance
If P-value > .05 (equal σ assumed) FAIL TO REJECT
If P-value < .05 (equal σ not assumed) REJECT
Small Sample Sizes
Use T-test for n < 30
Shape of t-distribution determined by df
df = Sample size - Number of population parameters estimated
When n > 30 t = Z, so computer always shows t
