Transcript Dominion Training Survey

Dominion Training
Survey
BADM 621
October 19,2005
Shawn Miller
Kylee Fink
Soumya Prasad
Thomas O’Neill
Casey Brown
Area Explanation
Descriptive Statistics - Test Scores
Region
Statistic
Southern
Western
Northeastern
Central
Count
30
17
24
14
Mean
8.70
8.76
8.04
9.14
9
9
8
9
Standard Deviation
1.09
0.97
1.16
0.77
Variance
1.18
0.94
1.34
0.59
Median
Geographic Location & Test Scores
• Ho – There is no difference in test scores
when comparing the four geographic
locations.
• P-value = 0.01338
• F = 3.79354, F crit = 2.71734
• Conclusion – There is a difference (Ha).
Location/Test Score Detail
• Central performed significantly higher in
comparison to the sample average.
– Mean = 9.142857, P-value = 0.015935
• Northeastern performed significantly lower
in comparison to the sample average.
– Mean = 8.041667, P-value = 0.015724
Practice makes perfect?
• Hypothesis: Talks given in the Northeastern
and Central regions were better because the
instructors had given the talk at least twice
already, at the Southern and Western
regions.
Cont…
• 2 columns: Early sessions (Southern +
Western) vs. later ones (Northeastern +
Central).
• 2 categories (early vs. late) compared on a
continuous variable (test score, quality
score): Use t-test.
Test score:
• F-test gives a p-value of 0.26 => Equal
variances;
• T-test: p-value: 0.25
• Statistically speaking…Test score did not
show that the instructors “practice” had
helped.
Quality score:
• F-test gives a p-value of 0.44=> Equal
variances;
• T-test: p-value: 0.163
• Statistically speaking…The quality score
did not show that the instructors “practice”
had helped.
Quality Score (cont.)
• In theory, practice should improve scores.
• Perhaps:
(a) The time period between the early and
late sessions (1 week) was not adequate.
(b) Inherent difference in quality of
speakers.
(c) Inherent difference in quality of
audiences.
Why did some people fare better?
• Hypothesis: Perception of having enough
time resulted in a better test score.
• Relationship between the 2 continuous
variables of time and test score: Correlation
function
• Correlation = 0.19; Yes, there is a positive
linear correlation between perception of
having enough time and final test score.
Cont…
• Did the perception of having time predict a
better test score? Use simple linear
regression.
• Dependent variable = Test score,
independent variable = perception of time.
• Regression: P-value: 0.08
• Perception of more time did not predict a
better test score.
Supporting Evidence
• Hypothesis: Perception of higher
knowledge before training resulted in a
better test score.
• Correlation = 0.22; Yes, there is a positive
linear correlation between perception of
knowledge base and final test score.
Helpful talks?
• Was there a relationship between perception
of base knowledge and gain in knowledge
from the talks?
• Correlation: r=0.69, positive linear
correlation exists.
Cont..
• Does the perceived higher base knowledge
predict a higher post-talk knowledge?
• Regression: Dependent variable =
knowledge after, independent = base
knowledge.
• P-value= 1.26E-13, almost 0
• Knowledge after = knowledge before (0.43) + 5.53
• R square = 0.48
• 48% of the variance in the data can be
explained by this regression; “Goodness of
fit” is high.
Self Assessment Scores
• Is there a relationship between self
assessment scores and actual test scores?
• At first glance, individuals in the
Northeastern Area indicated they learned
more than others during the training – an
average increase of 2.95 on a 10 point scale.
• Northeastern had the poorest test scores.
Self Assessment Scores (cont.)
• Is this significant?
• F-test yields p value of 0.34
• T-test (for equal variances) yields
P value of 0.99 – not significantly different from
the mean across all areas.
Questions?