Sei sulla pagina 1di 3

POST LABORATORY REPORT FOR EXPERIMENT 1: APPLICATION

OF STATISTICAL CONCEPTS IN THE DETERMINATION OF


WEIGHT VARIATION IN SAMPLES


L. TORRES
NATIONAL INSTITUTE OF GEOLOGICAL SCIENCES
UNIVERSITY OF THE PHILIPPINES, DILIMAN, QUEZON CITY 1101, PHILIPPINES
DATE SUBMITTED: 22 AUGUST 2014
DATE PERFORMED: 20 AUGUST 2014




INTRODUCTION

An analytic chemistry student depends mostly on the
generation and interpretation of precise chemical data
for success in experimentation, as well as accuracy in
results.
1


A typical experiment involving a significant amount of
statistical analysis requires data from several
independent experimental measurements, each of
which is subject to a random uncertainty and each of
which contributes to the net random error of the
result
2
. It is necessary to asses both the accuracy and
precision of the data to determine some important
aspects of statistical calculations like central
tendency, measure of accuracy and measure of
precision.

The experiment used 25 centavo coins as the
samples. Each was weighed on the analytical
balance, and the results were then recorded. After the
weigh ins, the students were tasked to find some
values that will assess the statistical measures of the
population, such as the mean, the standard deviation,
etc.

RESULTS AND DISCUSSIONS


1
Rorabacher, David B. Statitistical Treatment for
Rejection of Deviant Values Critical Values of Dixons
Q Parameter and Related Subrange Ratios at the
95% Confidence Level. Michigan: Department of
Chemistry
2
Skoog et. al. Fundamentals of Analytical Cehmistry
8
th
ed. Thomson-Brooks Cole, 2004
Here is the tabulated results from the experiment.
Sample No. Weigh, g
1 3.6184
2 3.6139
3 3.5854
4 3.6054
5 3.6312
6 3.7744
7 3.5671
8 3.5630
9 3.5886
10 3.5938

*Remember that although the class used a single
device (analytical balance) to measure the weights of
the samples, and used a common type of item (25
centavo coin) as sample, the recorded weights will
not be the identical with each other and with the other
groups because not all 25 centavo coins have uniform
mass.

First, each experiment involving a statistical treatment
in one way or another, needs to reject any outlying or
deviant results as not to make the computed
measures too error-prone. The Q-test by Dixon
proves handy in eliminating such outlying results.

For this experiment we computed the Q-test by, and
as always, we first determined both the lowest and
highest recorded sample data as possible outliers.

The calculations are here as follows:
Eqtn. 1
Set 1= H=3.7744

Q-tab at 95% confidence level=0.625

= 0.7577
Therefore, Q(exp)>Q(tab), meaning 3.7744 will be
rejected.

After performing all the Q-test for all the possible
outliers, we have come up with these results:
Table 2,
Data
Set
Suspect
Values
Q tab Q exp Conclusion
1 3.7744 0.625 0.7577 Rejected
3.5854 0.625 0.1058 Retained
2 3.7744 0.466 0.6774 Rejected
3.5630 0.466 0.0194 Retained

After determining the outliers and eliminating them, it
is now possible to calculate for the other measures of
the samples. The mean of the population was:
Eqtn. 3
Set 1:


=3.61086
Set 2:



=3.5963
The mean of Set 1 was= 3.6109 while Set 2 has a
mean of =3.5963

After determining the mean, we were tasked to find
the standard deviation of the population,
Eqtn. 4
s=

)))



Set 1, s=0.0170
For set 2 the standard deviation was 0.02303

After getting the standard deviation, it will now be
easy to get the relative standard deviation of the
population.
Eqtn. 5
Set 1: RSD:

X1000=3.4ppt
Set 2: RSD:

X1000=2.558ppt

For the range, we just subtract the highest value with
the lowest value in a set of measurements.
Eqtn. 5
Set 1: R=3.6312-3.5854=0.0458
Set 2: R=3.6312-3.5630=0.0682

The relative range is fairly easy to compute given R,
Eqtn 6.
Set 1: RR=

x1000=9.160
Set 2: RR=

x1000=7.578

The last part of the statistical analysis was finding the
confidence interval of the samples.
Eqtn. 7
Confidence limit=mean

, where s is the standard


deviation, n is the sample size and t depends on the
confidence level and degree of freedom.
Calculating for the confidence interval of Set 1=
3.6109
))

=3.59 to 3.63 0.02


Set 2=
3.5963
))

=3.58 to 3.61

PRINCIPLES AND CONCEPTS
*Answers to guide for discussion questions

All the calculations that we have done are important.
They will surely be helpful in upcoming experiments,
mainly because this is an analytical chemistry course
and involves a lot of calculations.

In measuring certain samples, there is a possibility
that there are some samples that are strikingly
imprecise with the other samples. Therefore, if they
are included for the calculation for the test for
centrality or accuracy, the whole population will suffer.
That is why the Q-test tries to answer this paradox by
eliminating outliers that will greatly affect the
calculations in a error prone way. Therefore, when the
outliers are rejected, the spread of value will
approximate a bell shaped formation, signifying a well
distributed result.
3


Meanwhile, measures for central tendency, more
specifically the mean, tries to determine the centre of

3
Dean, R.B. and Dixon, W.J. Simplified Statistics for
Small Numbers of Observations. Eugene: University
of Oregon
distribution. It tries to give a certain value that will
represent the scattered values of the results in a more
equal or averaged way. On the other hand, the
standard deviation approximates the amount of
variation or dispersion from the mean.
4
It also
measures the nearness of the results from one
another because it is one of the measures for
precision.

The confidence interval is the measure of a reliability
of an estimate. It just says that the experiment, when
repeated will exhibit this kind of results 95% of the
time.

The difference of the results of set 1 from set 2 is:
1. Set 1 has fewer number of samples
2. Set 1 has a higher mean, higher relative
standard deviation higher range and higher
relative range, compared to results from set
2
3. The only similarity they have is they both
have in the confidence interval.


CONCLUSION:

All scientific courses require an ample, if not
significant, amount of statistical knowledge. Analytical
chemistry is a primary example of the relationship
between math and statistics and science. All scientific
workers are fully aware that science requires
exactness of results and data. And to satisfy these
need of science, we need statistics to aid us in the
endless computations that we will encounter as we
thread along the scientific path.

4
Blant J.M., Altman D.G. Statistics Notes:
Measurement Error. BMJ

Potrebbero piacerti anche