Basic Statistics Using

Minitab

L. GOCH – FEBRUARY 2011

 AGENDA
 Comparing 1 Group to a Target / Specification OR
Comparing 2+ Groups to Each Other:
1) Stability – Run Chart (Feb 4th) or Control Chart (Mar 4th)
2) Shape – Histogram (Feb 4th) or Probability Plot
3) Spread – Test for Equal Variances
4) Centering – 1-Sample T-test, 1-Sample Sign, Paired T,
2-Sample T-test, ANOVA or Mood’s Median Test
 Comparing Proportions - 1 Proportion, 2
Proportion
 Basic Linear Regression – Fitted Line Plot
 Chi-Squared – Cross Tabulation and Chi-Square
 Trend Analysis – Under Stat > Time Series: We
won’t be covering in this class. See Tutorials for
more information.
PROBABILITY PLOT:

GRAPH > PROBABILITY PLOT

PROBABILITY PLOT: GRAPH > PROBABILITY PLOT
 Use to display overlaid probability plots of multiple
variables and/or multiple groups on the same graph.

Open worksheet FlameRTD.mtw

 PROBABILITY PLOT
Probability Plot of Fabric1
Normal
99
C oating
A
B
95 N one
90 90 Mean StDev N AD P
3.013 0.4138 15 0.321 0.497
80
2.727 0.3575 15 0.545 0.133
70 3.573 0.5700 15 0.310 0.517
60
Percent

50
40
Since all p-values are >
30 0.05, conclude each data set
20 is NORMA LLY DISTRIBUTED.

10

5
3.185

3.544

4.304

1
2.0 2.5 3.0 3.5 4.0 4.5 5.0
Fabric1
TEST FOR EQUAL
VARIANCES:

STAT > ANOVA > TEST FOR

EQUAL VARIANCES
TEST FOR EQUAL VARIANCES: STAT > ANOVA > TEST FOR
EQUAL VARIANCES
 Use to display overlaid probability plots of multiple
variables and/or multiple groups on the same graph.

Open worksheet FlameRTD.mtw

TEST FOR EQUAL VARIANCES (SESSION WINDOW & GRAPH RESULTS)

Use for Normal Data

Use for Non-Normal Data

Test for Equal Variances for Fabric1

Bartlett's Test
Test Statistic 3.19
A P-Value 0.203
Levene's Test
Test Statistic 1.42
P-Value 0.253
C oating

None

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
95% Bonferroni Confidence Intervals for StDevs
CENTERING COMPARISON ANALYSES
Response is Numeric & Factors are Text or Numeric
Analysis # of Grps Comparison Statistic
1-Sample T 1 Target Average
1-Sample Sign 1 Target Median
Paired Data
Paired T 2 (e.g. before Difference
vs after)
2-Sample T 2 Each Other Average
ANOVA 2+ Each Other Average
Mood’s Median Test 2+ Each Other Median
1-SAMPLE T-TEST:

STAT > BASIC STATISTICS >

1-SAMPLE T
1-SAMPLE T-TEST: STAT > BASIC STATISTICS > 1-SAMPLE T
 Performs a one-sample t-test or t-confidence
interval for the mean.
Open worksheet
EXH_Stat.mtw
1-SAMPLE T (SESSION WINDOW & GRAPH RESULTS)
Note: For n<30, data is assumed to be Normally Dist’d

95% of the time the True

Avg will be between
4.5989 & 4.9789

Boxplot of Values
(with Ho and 95% t-confidence interval for the mean)
The sample Avg is
significantly different
from the Target of 5.

_
X

Ho

4.4 4.5 4.6 4.7 4.8 4.9 5.0 5.1

Values
1-SAMPLE SIGN:

STAT > NONPARAMETRICS>

1-SAMPLE SIGN
1-SAMPLE SIGN: STAT > NONPARAMETRICS> 1-SAMPLE SIGN
 Performs a one-sample t-Sign or t-confidence
interval for the median.
Open worksheet
EXH_Stat.mtw
1-SAMPLE SIGN (SESSION WINDOW RESULTS)

The sample Median is NOT

significantly different from the
Target of 115.

95% of the time the True Median will

be between 108.5 & 211.7
PAIRED T-TEST:

STAT > BASIC STATISTICS >

PAIRED T
PAIRED T-TEST: STAT > BASIC STATISTICS > PAIRED T
 Tests the mean difference between paired (related)
observations.
Open worksheet
EXH_Stat.mtw
PAIRED T (SESSION WINDOW & GRAPH RESULTS)
95% of the time the True
Avg Difference will be
between -0.687& -0.133
which does NOT contain
ZERO.

Boxplot of Differences
(with Ho and 95% t-confidence interval for the mean)

The sample Avgs are significantly

different from each other.

_
X

Ho

-1.2 -0.9 -0.6 -0.3 0.0

Differences
2-SAMPLE T-TEST:

STAT > BASIC STATISTICS >

2-SAMPLE T
2-SAMPLE T-TEST: STAT > BASIC STATISTICS > 2-SAMPLE T
 Performs a two-sample t-test or t-confidence
interval for the mean difference.
Open worksheet
Furnace.mtw
2-SAMPLE T (SESSION WINDOW & GRAPH RESULTS)

95% of the time the True

Avg Difference will be
between -1.450 & 0.980
which contains ZERO.

Boxplot of BTU.In
20

The sample Avgs are

15 NOT significantly different
from each other.
BTU.In

10

1 2
Damper
ONE-WAY ANOVA:

STAT > ANOVA > ONE-WAY

ONE-WAY ANOVA: STAT > ANOVA > ONE-WAY
 Compares the Averages for 2 or more Groups.
Open worksheet
Exh_AOV.mtw
ONE-WAY ANOVA (SESSION WINDOW & GRAPH RESULTS)
Boxplot of Durability
At least 1 Pair of Avgs are 22.5

significantly different from 20.0

each other. 17.5

15.0

Durability
12.5

10.0

7.5

5.0
1 2 3 4
Carpet
MOOD’S MEDIAN TEST:

STAT > NONPARAMETRICS>

MOOD’S MEDIAN TEST
MOOD’S MEDIAN TEST: STAT > NONPARAMETRICS> MOOD’S
MEDIAN TEST
 Compares the Medians of 2 or more Groups.

Open worksheet
Cartoon.mtw
MOOD’S MEDIAN TEST (SESSION WINDOW RESULTS)
At least one group’s
median is significantly
different from the others.

Group 2 is significantly different

Boxplot of Otis
from groups 0 & 1 since the 95%
140
CI’s do NOT overlap.
130

120

110
Otis

100

90

80

70
0 1 2
ED
1-PROPORTION TEST:

STAT > BASIC STATISTICS >

1-PROPORTION
1-PROPORTION: STAT > BASIC STATISTICS > 1-PROPORTION
 Performs a one-sample proportions test or p-
confidence interval for a proportion.
No worksheet is
needed for this test.
1-PROPORTION: (SESSION WINDOW RESULTS)
Note: Proportion Testing Should NOT be done when the
sample size n<30!!

95% of the time the True

Proportion will be between
55.74% & 62.10%

The sample Proportion

is significantly different
from the Target of 65%.
2-PROPORTION TEST:

STAT > BASIC STATISTICS >

2-PROPORTIONS
2-PROPORTIONS: STAT > BASIC STATISTICS > 2-PROPORTIONS
 Performs a two-sample proportions test or p-
confidence intervals for a proportion.
No worksheet is
needed for this test.
2-PROPORTIONS: (SESSION WINDOW RESULTS)
Note: Proportion Testing Should NOT be done when the
sample size for any one group is <30!!

95% of the time the True

Difference between the
Proportions will be
between -9.58% & 17.58%

The difference in Proportions

is NOT significantly different.
BASIC LINEAR
REGRESSION:

STAT > REGRESSION >

FITTED LINE PLOT
REGRESSION: STAT > REGRESSION > FITTED LINE PLOT
 Performs a Regression Analysis on 1 Input (X) and
1 Output (Y).
Open worksheet
Exh_REGR.mtw
REGRESSION: (GRAPH RESULTS)
Fitted Line Plot
EnergyConsumption = 1.25 + 0.3218 MachineSetting
40 S 12.1825
R-Sq 2.3%
R-Sq(adj) 0.0%

30
EnergyConsumption

20

10

10 15 20 25 30
MachineSetting

The line does NOT fit the curved data. Need a

quadratic, cubic or transformation of the data.
 REGRESSION: (SESSION WINDOW & GRAPH RESULTS)

The Regression Equation

shows that Machine
Setting explains 93.1% of
the variability in Energy
Consumption.

Fitted Line Plot

log10(EnergyConsumption) = 7.070 - 0.6986 MachineSetting
+ 0.01740 MachineSetting**2
100
S 0.167696
R-Sq 93.1%
R-Sq(adj) 91.1%
EnergyConsumption

10

The Quadratic Term is significant in

the model. NOTE: If a higher order 1

term is significant than the lower 10 15 20

MachineSetting
25 30

order term must remain in the model.

CROSS TABULATION AND
CHI-SQUARE:

STAT > TABLES > CROSS

TABULATION AND CHI-
SQUARE
CHI-SQUARE: STAT > TABLES > CROSS TABULATION AND CHI-
SQUARE
 Performs a Chi-Squared Analysis on Count Data.
Open worksheet
Exh_TABL.mtw
CHI-SQUARE ANALYSIS: (SESSION WINDOW RESULTS)

No significant
association between
Gender and Activity
 CONCLUSIONS

 Results need to be Supported by data

 Not based on conjecture or intuition

 Shown in 1) Graphical & 2) Statistical format

 Make sense from an 3) Engineering standpoint

 Use P-values to determine if Results could have

happened by Chance!!

Need Significant Differences

for Reliable Conclusions !!