Definitions Flashcards
One categorical variable use what type of graph?
pie chart, bar chart
Two categorical variables = what type of graph?
contingency table
Categorical Data
words and categories
Quantitative Data
numbers
Graphs for 1 Quantitative Variable
histograms, dot plots, stem and leaf plots, box plot
Graph for 2 Quantitative Variables
Scatterplot
Graph for 1 Quantitative Variable w/ multiple categories
Boxplots
Standard deviation is the
square root of the variance
Find Mean & Standard Deviation in JMP by:
> Analyze
>Distribution
Sample Statistic for one proportion
phat
Sample Statistic for one mean
ybar
Population Parameter for one proportion
p
Population Parameter for one mean
mu
Sample Statistic for 2 independent means
ybar1 - ybar2
Sample Statistic for matched pairs
ybard
Sample Statistic for 2 independent proportions
phat1 - phat2
Population Parameter for 2 independent means
mu1 - mu2
Population Parameter for matched pairs
mud
Matched Pairs example
twins, husbands & wives, pretest/posttest, sales vs no sales from SAME store
If you see average then…..
MEAN
If you see proportion then…..
PROPORTION
Standard error =
the standard deviation of the SAMPLE
For Means to be NORMAL:
- large enough sample size
- random
- independent
- not larger than 10% of population
For Proportions to be NORMAL:
- need at least 10 successes and 10 failures
- random
- independent
- not larger than 10% of population
Mean/Proportion problem, finding probability: first find standard error (std. deviation of the sample mean) then JMP =
> Distribution Calculator
Width of a CI =
2(ME) OR (Upper CI - Lower CI)
As confidence level increases, the width of the confidence level will
increase
If sample size increases, width of the confidence interval will
decrease
IF null hypothesis is contained within the confidence interval and when null is zero and contained you would
FAIL TO REJECT HO
IF null hypothesis is NOT contained within the confidence interval or when null is zero and not contained you would
REJECT HO
P-value tells us the probability of
obtaining the sample assuming the null hypothesis is true
P-value lower than a (significance level)?
REJECT THAT HO
Results are statistically significant
There is sufficient evidence for HA
P-val less than 0.01
VERY STRONG
P-val less than 0.05
STRONG
P-val greater than 0.10
NONE
To determine where equal variances or not =
larger std dev / smaller std dev < 2 then ITS EQUAL/POOLED
Ho and Ha are always with the
PARAMETERS
Means use __ score
T
Proportions use __ score
Z
To find the CI of 2 proportions/means…
Use Hypothesis Test!!!
Previous / expected in a study is the
p for the Ho and Ha
You get a test statistic then what conclusion?
JMP: Dist calculator, input values to calculate probability, put test statistic as value, 2 sided, find p-val and make conclusion
Type 1 & 2 Errors!!!
1 - Ho True - Reject - Alpha
2 - Ho False - FTR - Beta
Increase alpha = increase probability of Type 1 means Type 2 will
DECREASE
Increasing sample size will decrease probability of a ___ error
Type 2
Power is
ability of a hypothesis test to detect the Ho is false
The bigger the effect side, the easier to
reject Ho
Chi Sq GOODNESS OF FIT
determine if data fits a given distribution
one categorical variable
whole what night people go out more problem
Chi Sq INDEPENDENCE
determine if two categorical variables are independent
if 3X3 then it must be independence
Chi Sq HOMOGENEITY
ONE categorical but over TWO populations
first semester, last semester problem testing crammed stuff
Linear regression is always ___
EXPECTED
residual - e =
actual (y) - predicted (yhat)