SULTAN KUDARAT STATE UNIVERSITY

Lesson 2: Analysis, Interpretation, and Use of Test Data

Pre-discussion
Discussions in this lesson will build upon the concepts and examples
presented in the preceding lesson, which focused on the tabular and graphical
presentation and interpretation of test results. This time, other ways of
summarizing test data using descriptive statistics, which provides a more
precise means of describing a set of scores, will be introduced. The word
“measures” is commonly associated with numerical and quantitative data.
Hence, the prerequisite to understanding the concepts contained in this
lesson in your basic knowledge of mathematics, e.g., summation of values,
simple operation on integers, squaring and finding the square roots, etc.

What to Expect?
At the end of the lesson, the students can:
1. find the mean, median, and mode of test score distribution;
2. determine the different measures of dispersion of test scores;
3. calculate the measure of position;
4. relate standard deviation and normal distribution;
5. transform raw scores to standardized scores (z, T and stanine);
6. compute the measure of covariability using the long process and
Excel; and
7. interpret test data applying measures of central tendency, variability,
position, and covariability.

Measures of Central Tendency

The word “measure of central tendency” means the central location or

point of convergence of a set of values. Test scores have a tendency to
converge at a central value. This value is the average of the set of scores. In
other words, a measure of central tendency gives a single value that
represents a given set of scores. Three commonly-used measure of central
tendency or measures of central tendency or measures of central location are
the mean, and the mode.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 1

 SULTAN KUDARAT STATE UNIVERSITY

Mean. This is the most preferred measure of central tendency for use
with test scores, also referred to as the arithmetic mean. The computation is
very simple. When a student has added up the examination scores he/she
made in a subject during the grading period and divided it by the number of
examinations taken, then he/she computed the arithmetic mean.

That is, ̅ = where ̅ = the mean, the sum of all the scores,

and N = the number of scores in the set.

Consider again the test scores of students given in Table 1, which is
the same set of test scores used in the previous lesson.

Table 1. Scores of 100 College Students in a Final Examination

53 30 21 42 33 41 42 45 32 58
36 51 42 49 64 46 57 35 45 51
57 38 49 54 61 36 53 48 52 49
41 58 42 43 49 51 42 50 62 60
33 43 37 57 35 33 50 42 62 49
75 66 78 52 58 45 53 40 60 33
46 45 79 33 46 43 47 37 33 64
37 36 36 46 41 43 42 47 56 62
50 53 49 39 52 52 50 37 53 40
34 43 43 57 48 43 42 42 65 35

The mean is the sum of all the scores from 53 down to the last score,
which is 35, divided by the total number of cases.
That is,
̅= = (53 + 36 + 57 + … + 60 + 49 + 35)/100

You have many ways computing the mean. The traditional long and
tedious computation techniques have outlasted their relevance due to
advancement of technology and the emergence of statistical software. Using
your specific calculator, you will see the symbols ̅ and . Just follow the
simple steps indicated in the guide. There are also simple steps in Microsoft
Excel. Different versions of the statistical software SPSS offer the fastest way
of obtaining the mean, even with hundreds of scores in a set. There is no loss
of original information because you are dealing with original individual scores.
The use of statistical software will be explained later.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 2

 SULTAN KUDARAT STATE UNIVERSITY

While we organize the power of technology, there is information that is

unappreciated because of the short-hand processing of data through
mechanical computations. Look at the conventional way of presenting data in
a frequency distribution table as done in the previous lesson:

Table 2. Frequency Distribution

Class Midpoint F Xi f Cumulative Cumulative
Interval (X) Frequency (cf) Percentage
75-80 77 3 231 100 100
70-74 72 0 0 97 97
65-69 67 2 134 97 97
60-64 62 8 496 95 95
55-59 57 8 456 87 87
50-54 52 17 884 79 79
45-49 47 18 846 62 62
40-44 42 21 882 44 44
35-39 37 13 481 23 23
30-34 32 9 288 10 10
25-29 27 0 0 1 1
20-24 22 1 22 1 1
Total N = 100 if = 4720

In the traditional way, it cannot be argued that you can see at a glance
how the scores are distributed among the range of values in a condensed
manner. You can even estimate the average of the scores by looking at the
frequency in each class interval. In the absence of statistical program, the
mean can be computed with the following formula.

where Xi = midpoint of the class interval

f = frequency of each class interval
N = total number of scores
Thus, the mean of the test scores is computed as follows:

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 3

 SULTAN KUDARAT STATE UNIVERSITY

Median. It is the value that divides the ranked scores into halves, or the
middle value of the ranked scores. If the number of scores is odd, then there
is only one middle value that gives the median. However, if the number of
scores in the set is even number, then there are 2 middle values. But if there
are more than 50 scores, arranging the scores and finding the middle value
will take time. This formula will help you determine the median:

where Lm = exact lower limit of the median class

i = size of the class interval
N = total number of scores
fm = frequency of the median class
cfm = cumulative frequency below the median class

Applying the formula:

1. You need a column for cumulative frequency. This is now shown on the 5 th
column for data in Table 2.
2. Determine N/2 which is one-half of number of scores of examinees (that is,
100/2 = 50).
3. Find the class interval of the 50th score. In this case where there are 100
scores, the 50th score is in the class interval of 45-49. This class interval of
45-49 becomes the median class. We marked lines in the table to indicate
where the median class is located for easy reference when computing the
median value. (Please see illustration.)

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 4

 SULTAN KUDARAT STATE UNIVERSITY

4. Find the exact limits of the median class. In this case, class 44.5-49.5. the
lower limit then is 44.5
Summing up these steps and substituting these values to the formula,
we have:

Mode. It is the easiest measure of central tendency to obtain. The mode is the
score or value with the highest frequency in the set of scores. If the scores are
arranged in a frequency distribution, the mode is estimated as the midpoint of
the class interval which has the highest frequency. This class interval with the
highest frequency is also called the modal class. In a graphical representation
the frequency distribution, the mode is the value in the horizontal axis at which
the curve is at its highest point (peak). If there are two highest points, then,
there are two modes, as discussed in previous lesson. When all the scores in
a group have the same frequency, the group of scores has no mode.
Considering the test data in Table 2, it can be seen that highest
frequency of 21 occurred in the class interval 45 - 49. The rough estimate of
the mode is 42, which is the midpoint of the class interval.
As manual computations of the mean, median and mode are so
lengthy and tedious processes; technology makes it simpler through the use
of Microsoft Excel application. Here’s the simple guide to observe.

Average Function in Excel (Mean)

One of the most applied statistical functions in Excel is the Average.

Calculating the Average or Mean in Excel is much simpler than it was in the
school. Simply use the Average function and select the range which needs to
be averaged. In the example below, we want to know the mean of the marks
obtained by the students so we use =AVERAGE(B2:B12). Pressing ENTER
key would give us the mean of the set of scores found from cell B2 to cell

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 5

 SULTAN KUDARAT STATE UNIVERSITY

B12. Calculating the mean of scores located in several columns and rows can
also be possible provided they are all selected or defined.

Median in Excel

Median is a function which is used to find the middlemost number in a

given range of numbers. When you are finding median manually, you need to
sort the data in an ascending order but in Excel, you can simply use the
Median function and select the range and you will find your median. We take
the same example as above to find the median of marks obtained by
students. So we use =MEDIAN(B2:B12). This is also possible when treating
scores across some rows and columns.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 6

 SULTAN KUDARAT STATE UNIVERSITY

Mode in Excel

Mode helps you to find out the value that occurs most number of times.
When you are working on a large amount of data, this function can be a lot of
help. To find the most occurring value in Excel, use the MODE function and
select the range you want to find the mode of. In our example below, we use
=MODE(B2:B12) and since 2 students have scored 55 we get the answer as
55.

In situations where there are two or more modes in your data set, the
Excel MODE function will return the lowest mode.

When are mean, median, and mode appropriately used?

To appreciate comparison of the three measures of central tendency, a
brief background of level of measurement is important. We make observations
and perform assessments on many variables related to learning - vocabulary,
spelling ability, self-concept, birth order, socio-economic status, etc. The level
of measurement helps you decide how to interpret the data as measures of
these attributes, and this serves as a guide in determining in part the kind of
descriptive statistics to apply in analyzing the test data.

Scale of Measurement
There are four levels of measurement that apply to the treatment of test
data: nominal, ordinal, interval, and ratio. In nominal measurement, the
number is used for labeling or identification purposes only. An example is the
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 7
 SULTAN KUDARAT STATE UNIVERSITY

student’s identification number or section number. In data processing, instead

of labeling gender as female, a code “1” is used to denote Female and “2” to
denote Male. While “2” is numerically greater than “1,” in this case the
difference of 1 has no meaning; it does not indicate that Male is better than
Female. The purpose is simply to differentiate or categorize the subjects by
gender.
The ordinal level of measurement is used when the values can be
ranked in some order of characteristics. The numeric values used to indicate
the difference in traits under consideration. Academic awards are made on
the basis of an order of performance: first honors, second honors, third
honors, and so on. Some assessment tools require students to rank their
interest or hobbies, or even career choices. Percentile ranks in national
assessment test or entrance examination are examples of measurement in an
ordinal scale. Percentile score becomes more useful and meaningful than
simple raw scores in university entrance or division-wide examinations.
The interval level of measurement, which has the properties of both the
nominal and ordinal scales, is attained when the values can describe the
magnitude of the differences between groups or when the intervals between
the numbers are equal. “Equal Interval” means that the distance between the
things represented by 3 and 4 is the same distance represented by 4 and 5.
The difference between the temperature 30 o and 40o is the same as that
between 90o and 100o. However, there is no true zero point. The zero degree
in the Celsius thermometer does not mean zero or absence of heat, 0 o is an
arbitrary value, a convenient starting point. With arbitrary zero point, there is a
restriction with interval data. We cannot say that an 80 o object is twice as hot
as 40o object. In the educational setting, a student who gets a score of 120 in
a reading ability test is not twice the better reader than one who got a score of
60 in the same test.
The highest level of measurement is the ratio scale. As such, it carries
the properties of the nominal, ordinal, and interval scales. Its additional
advantage is the presence of a true zero point, where zero indicates the total
absence of the trait being measured. A 0 cm as a measure of width means no
width, 0 km as a measure of distance means no distance traveled, and 0
words spelled means no words was spelled at all. Test scores as measure of
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 8
 SULTAN KUDARAT STATE UNIVERSITY

achievement in many school subjects are often treated as interval scale.

However, if achievement in a performance test in Physical Education is
measured by the number of “push-ups” one can do in a minute or distance run
in an hour; or in a Typing Class where, we count the words typed in a minute
or words spelled correctly, then these are all on ratio scale.
Now, the most likely questions that cross your mind are:
 Which measure of central tendency should I use?
 Do I have to use all the three since the statistical program can
automatically give the three measures the easiest way?
Generally, the mean is the most used measure of central tendency
because it is appropriate for interval and ratio variables, which are higher
levels of measurement. Its value is affected by the change of a single score
such that it is regarded as the most accurate measure to represent a set of
scores. In research, this is most utilized specifically when you want to make
an inference about the population characteristics on the basis of the observed
sample value.
For median, in some cases, we could have one very high score (or
very few high scores) and many low scores. This is especially true when the
test is difficult, or when students are not well prepared for the test. This will
result to many low scores and a few high scores that will lead to a positively-
skewed distribution. In the same way, when the test is too easy for students,
there will be many high scores, which lead to a negatively-skewed distribution.
In both cases, the mean can give an erroneous impression of central
tendency because its value is pulled by the extreme values that reduce its role
as the representative value of the set of scores. Hence, the median is a better
measure. It is the value that occupies the middle position among the ranked
values; thus, it is less likely to be drawn toward the direction of extreme
scores. It is an ordinal statistic but can also be used for interval or ratio data
distribution.
The mode is determined by the highest frequency of observation that
makes it a nominal statistic.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 9

 SULTAN KUDARAT STATE UNIVERSITY

How do measures of central tendency determine skewness?

The mean, median, and mode may further be compared if they have
been calculated from the same frequency distribution. In a perfectly
symmetrical distribution, the mean, median, and mode have the same value,
and the value of the median is between the mean and the mode. This shape
is illustrated in Figure 1.

Figure 1. Mean, Median, and Mode in a Symmetrical Distribution

When the distribution becomes positively-skewed as shown in Figure

2, there are variations in their values. The mode stays at the peak of the curve
and its value will be smallest. The mean will be pulled out from the peak of the
distribution toward the direction of the few high scores. Thus, the mean gets
the largest value. The median is between the mode and the mean.

Figure 2. Mean, Median, and Mode in a Positively-Skewed Distribution

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 10

 SULTAN KUDARAT STATE UNIVERSITY

On the other hand, in a negatively-skewed distribution, as shown in the

graph below, the mode remains at the peak of the curve, but it will have the
largest value. The mean will have the smallest value as influenced by the
extremely low scores, and the median still lies between the mode and the
mean.

Figure 3. Mean, Median, and Mode in a Negatively-Skewed Distribution

What are measures of dispersion?

One important descriptive statistic in the area of assessment is the
measures of dispersion, which indicates “variability,” “spread,” or “scatter.”
(Please see Figure 4).

Figure 4. Measures of Variability of Sets of Test Scores

You can see that different distributions may be symmetrical, may have
same average values (mean, median, mode), but how the scores in each
distribution are spread out around these measures are different. In A, as
shown in Figure 4, scores range between 40 and 60; in B, between 30 and
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 11
 SULTAN KUDARAT STATE UNIVERSITY

70; and in C between about 20 and 80. Measures of variability give us the
estimate to determine how the scores are compressed, which contributes to
the “flatness” or “peakedness” of the distribution.
There are several indices of variability, and the most commonly used in
the area of assessment are the following.

Range. It is the difference between the highest (XH) and the lowest (XL)
scores in a distribution. It is the simplest measure of variability but also
considered as the least accurate measure of dispersion because its value is
determined by just two scores in the group. It does not take into consideration
the spread of all scores; its value simply depends on the highest and lowest
scores. Its value could be drastically changed by a single value. Consider the
following examples:
 Determine the range for the following scores: 9, 9, 9, 12, 12, 13, 15, 15,
17, 17, 18, 18, 20, 20, 20.
Range = XH - XL
= 20 – 9
Range = 11

Now, replace a high score in one of the scores, say, the last score and
make it 50. The range becomes:

Range = XH - XL
= 50 – 9
= 41
We observed that with just a single score, the range increased so high.
This can be interpreted as large dispersion of test scores; however, when you
look at the individual scores, it is not.

Variance and Standard Deviation. Standard deviation is the most widely

used measure of variability and is considered as the most accurate to
represent the deviations of individual scores from the mean values in the
distribution.
Let us examine the following test score distribution:
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 12
 SULTAN KUDARAT STATE UNIVERSITY

Class A Class B Class C

22 16 12
18 15 12
16 15 12
14 14 12
12 12 12
11 11 12
9 11 12
7 9 12
6 9 12
5 8 12

Note that while the distributions contain different scores, they have the
same mean. If we ask how each mean represents the score in their
respective distribution, there will be no doubt with the mean of distribution C
because each score in the distribution is 12. How about in distributions A and
B? For these two distributions, the mean of 12 is a better estimate of the
scores in distribution B than in distribution A. We can see that no score in B is
more than 4 points away from mean of 12. However, in distribution A, half of
the 12 scores is 4 points or more away from the mean. We can also say that
there is less variability of scores in B than in A. However, we cannot just
determine which distribution is dispersed or not by merely looking at the
numbers, especially when there are many scores. We need a reliable index of
variability, such as variance or standard deviation that takes into consideration
all the scores.
Recall that ∑(X- ̅ ) is the sum of the deviation scores from the mean,
which is equal to zero. As such, we square each deviation score, then sum up
all the squared deviation scores, and divide it by the number of cases, this
yields the variance. Extracting its square root gives us the standard
deviation.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 13

 SULTAN KUDARAT STATE UNIVERSITY

The measure is generally defined by the formula:

( ̅)

where = population variance

̅ = population mean
X = score in the distribution
Finding the square gives us this formula for the standard deviation.
That is,

( ̅)
√

where = population standard deviation

̅ = population mean
X = score in the distribution
If we are dealing with the sample data and we wish to calculate an
estimate of , the following formula is used for such statistic:

̅)
√ (

where: s = standard deviation

X = raw score
̅ = mean score
N = number of scores in the distribution
This formula is what statisticians termed as an “unbiased” estimate,
and more often preferred considering that in both research and assessment
studies, we deal on sample data rather than actual population data.
With the standard deviation, you can also see the differences between
two or more distributions. Using the scores in Class A and Class B in the
above data set, we can apply the formula:

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 14

 SULTAN KUDARAT STATE UNIVERSITY

Class A Class B
X (X – ̅ ) (X – ̅ )2
X (X – ̅ ) (X – ̅ )2
22 22-12 100 16 16-12 16
18 18-12 36 15 15-12 9
16 16-12 16 15 15-12 9
14 14-12 4 14 14-12 4
12 12-12 0 12 12-12 0
11 11-12 1 11 11-12 1
9 9-12 9 11 11-12 1
7 7-12 25 9 9-12 9
6 6-12 36 9 9-12 9
5 5-12 49 8 8-12 16
̅ = 12 ̅
∑(X– )2 = 276 ̅ = 12 ̅ 2
∑(X- ) = 74

The values 276 and 74 are the sum of the squared deviations of scores
in Class A and Class B, respectively. If these are divided by number of scores
in each class, this gives the variance (S2):

The values above are both in squared units, while our original units of
scores are not in squared units. When we find their square roots, we obtained
values that are on the same scale of units as the original set of scores. These
too give the respective standard deviation (S) of each class and computed
as follows:

Computing the population or sample standard deviation using the

manual or traditional means is so complicated mainly if one is careless. Again,
the calculation of these measures becomes easier and accurate with the use
of Excel. Consider the scores in Classes A and B:

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 15

 SULTAN KUDARAT STATE UNIVERSITY

When you are finding the variance using Excel, you can simply use the
VAR function and select the range and you will find the desired variance. We
take the data in Class A to find the variance of the scores obtained by
students. So we use =VAR(A2:A11). In the case of Class B, the same function
is used but the data in the cells from B2 to B 11 are considered. Thus, we use
=VAR(B2:B11).

To compute the standard deviation in Excel, similar process will be

observed. Let us consider the same data to solve for SA and SB. The STDEV
function is used to calculate the standard deviation; hence, =STDEV(A2:A11)
and =STDEV(B2:B11) are applied to Class A and B scores, respectively, as
shown. (Note: Using the lowercase “var” and “stdev” formats has no effect on
the results.)

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 16

 SULTAN KUDARAT STATE UNIVERSITY

In both instances, the Excel values match with the results using the
manual process. For larger number of scores in a distribution, Microsoft Excel
or other software is more appropriate and efficient in obtaining the variance
and standard deviation. This can be done in few seconds if you have already
entered and saved the data used to get the measure of central tendency. In
addition, the VAR and STDEV functions can still be used even if scores are
encoded in several columns and rows as shown in the next illustration.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 17

 SULTAN KUDARAT STATE UNIVERSITY

In additional, since the standard deviation is a measure of desperation,

it means that a large standard deviation indicates greater score variability
than a lower standard deviation. If the standard deviation is small, the scores
are closely clustered around the mean, or the graph of the distribution is
compressed even if it is symmetrical or skewed.

Figure 5. Homogenous Test Score Distribution in Different Skewness

Standard Deviation and Skewness

In the preceding discussion, we determined the skewness of a
distribution only graphically. The standard deviation provides an application
to determine the degree of skewness of test scores through a coefficient of
skewness formula:

where Sk = skewness
̅ = Mean
Mdn = median
SD = standard deviation

Let us recall our past discussion on how the relationships of measures

of central tendency are related to symmetry and skewness of the distribution.
The above formula further tells us that if the mean and the median are equal,
Sk will be equal to 0; therefore, the distribution will be symmetrical. As the

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 18

 SULTAN KUDARAT STATE UNIVERSITY

difference between mean and median moves farther from 0, the coefficient of
skewness changes to either lower or higher values.

Measures of Position
While measures of central tendency and measures of dispersion are
used often in assessment, there are other methods of describing data
distributions such as using measures of position or location. It is about the
score’s position in the distribution. What are these measures?

Quartile. In our discussion about the measure of central tendency, we

learned that median of a distribution divides the data into two equal groups.
In a similar way, the quartiles are the three values that divide the set of
scores into four equal parts, with one-fourth of the data in each part. This
means about 25% of the data falls at or below the first quartile (Q1); 50% of
the data falls at or below the 2nd quartile (Q2), and 75% falls at or below 3rd
quartile (Q3). Notice that Q2 is also the median. We can also say that Q1 is
the median of the first half of the values, and Q3 the median of the second
half of the values. Thus, the upper quartile represents on average the mark
of the top half of the class, while the lower quartile represents that the bottom
half of the class.
Quartiles are also used as a measure of the spread of data in the
interquartile range (IQR), which is simply the difference between the third
and first quartiles (Q3 – Q1). Half of this gives the semi-interquartile range
or quartile deviation (Q).

Figure 6. Quartiles

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 19

 SULTAN KUDARAT STATE UNIVERSITY

The following example illustrates these measures.

 Given the following scores, find the 1st quartile, 3rd quartile, and quartile
deviation.
90, 85, 85, 86, 100, 105, 109, 110, 88, 105, 100, 112
Steps:
1. Arrange the scores in the decreasing order.
2. From the bottom, find the points below which 25% of the score value
and 75% of the score values fall.
3. Find the average of the two scores in each of these points to determine
Q1 and Q3, respectively.

Find Q using the formula:

Note that in the above example, the upper and lower 50% contains
even center values, so the median in each half is the average of the two
center values. Consequently, applying the formula gives the quartile

deviation. That is,

Q= = 10

Decile. It divides the distribution into 10 equal parts. There are 9 deciles such
that 10% of the distribution are equal or less than decile 1, (D 1), 20% of the
scores are equal or less than decile 2 (D2); and so on. A student whose mark
is below the first decile is said to belong in decile 1. A student whose mark is

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 20

 SULTAN KUDARAT STATE UNIVERSITY

above the 9th decile belongs to decile 10. If there are a small number of data
values, decile is not appropriate to use.

Percentile. It divides the distribution into one hundred equal parts. In the
same manner, for percentiles, there are 99 percentiles such that 1% of the
scores are less than the first percentile, 2% of the scores are less than the
second percentile, and so on. For example, if you scored 95 in a 100-item
test, and your percentile rank is 99th, then this means that 99% of those who
took the test performed lower than you. This also means that you belong to
the top 1% of those who took the test. In many cases, percentiles are wrongly
interpreted as percentage score. For example, 75% as a percentage score
means you get 75 items correct out of a hundred items, which is a mark or
grade reflecting performance level. But percentile is a measure of position
such that 75th percentile as your mark means 75% of the students who took
the test got lower score than you, or you score is located at the upper 25% of
the class who took the same test. For every large data set, percentile is
appropriate to use for accuracy. This is one reason why percentiles are
commonly used in national assessments or university entrance examinations
with large dataset or scores in thousands.

What is Coefficient of Variation as a measure of relative dispersion?

We should note that the range, semi-inter quartile range or quartile
deviation, and standard deviations are expressed in the units of original
scores. Thus, they are measures of absolute dispersion. For instance, one
distribution of test scores in mathematics may have a standard deviation of
10, and another distribution of scores in science may have a standard
deviation of 5. If we want to compare the variability of the two distributions,
can we say that the distribution with standard deviation of 10 has twice the
variability of the one with standard deviation of 5? Consider another example:
one distribution has a standard deviation of 8 meters, while another has a
standard of P15.00. Can we say that the latter distribution is more spread than
the former? Or can we compare standard deviations in meter and pesos? The
answer seems obvious. We cannot conclude anything by direct comparison of
measures of absolute dispersion because they are of different units or
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 21
 SULTAN KUDARAT STATE UNIVERSITY

different categories. In the first example, one is the distribution of mathematics

scores while the other is the distribution of science scores. To make the
comparison logical, we need a measure of relative dispersion is also known
as the coefficient of variation (CV). This is simply the ratio of the standard

deviation of a distribution and the mean of the distribution. That is, .

The above formula indicates the coefficient of variation is a percentage

value.
Thus, if the mean score of the students in mathematics is 40 with a
standard deviation of 10, then the coefficient of variation is computed as
CVmath = (100)

= 0.25
= 25%.
Suppose the mean score of students in science is 18 with standard
deviation of 5. The coefficient of variation is:
CVsci = (100)

= 0.277
= 28%.
Looking at 10 and 5 as the standard deviation of mathematics and
science scores, respectively, this may lead you to judge that the set of scores
in mathematics is twice more dispersed than scores in science. From the
computed coefficient of variations as measure of relative dispersion, we can
clearly see that the scores in mathematics are more homogenous than the
scores in science.

How is standard deviation applied in a normal distribution?

The standard deviation is viewed as the most useful measure of
variability because in many distributions of scores, not only is assessment but
in research as well, it approximates the percentage of scores that lie within
one, two, or three standard deviations from the mean. To better appreciate
this point requires an understanding of the normal.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 22

 SULTAN KUDARAT STATE UNIVERSITY

The Normal Distribution

The normal distribution is a special kind of symmetrical distribution that
is most frequently used to compare scores. It has been found that when a
frequency polygon for a large distribution of scores of a natural phenomenon
and occurring characteristics (IQ, Height, income, test scores, etc.) are drawn
as a smooth curve, one curve stands out, which is the bell-shaped curve. As
seen below. This curve has a small percentage of observations on both tails,
and bigger percentage has a small percentage of observations on both tails,
and the bigger percentage on the inner part of the curve. The shape of this
particular curve is known as the normal curve, hence the name normal
distribution.

Figure 7. The Normal Curve

It is also called the Gaussian distribution. The standard deviation

becomes more useful because it is used to determine the percentage of
scores that all within a certain number of standard deviations from the mean.
As a result of many experiments, empirical rules have been established
pertaining to the areas under the normal curve. In assessment, the area in the
curve refers to the number of scores that fall within a specific standard
deviation from the mean score. In other words, each potion under the curve
contains a fixed percentage of cases as follows:
 68% of the scores fall between one standard deviations bellow and
above the mean
 95% of the scores fall between two standards deviations below and
above the mean

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 23

 SULTAN KUDARAT STATE UNIVERSITY

 99.77% of the scores fall between three standards deviations below

and above the mean
Figure 8 illustrates the theoretical model.

Figure 8. The Areas under the Normal Curve

From the above figures, we can state the properties of the normal
distribution:
1. The mean, median, and more are all equal
2. The curve is symmetrical. As such, the value is a specific area on the
left is equal to the value of its corresponding area on the right.
3. The curve changes from concave to convex and approaches the X-
axis, but the tails do not touch the horizontal axis.
4. The total area on the curve is equal to 1.

Standard Scores
In the preceding sections, we discussed raw scores, which are the
original scores collected from an actual testing activity. However, there are
situations where computing measures from raw scores may not be enough.
Consider a situation where we, as a student, want to know in what subjects
we performed best and poorest to determine where you need to exert. Or

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 24

 SULTAN KUDARAT STATE UNIVERSITY

maybe in the past, you took entrance examinations in more than one
university and asked yourself in which university you performed best. In case
like these we cannot find the answer by merely relying on a single score.
More concretely, if you get a score of 86 in English and 90 in Physics, you
cannot conclude that you perform better in Physics than in English. This is
ridiculous because 90 is higher than 86. Say, you later learned that the mean
score of the class in English was 80, and in Physics, the mean score like 86
or 90 is not meaningful unless it compared with other test scores. In particular,
a score can be interpreted more meaningfully if we know the mean and
variability of the other scores where that single score belongs. Knowing this, a
raw score can be converted into standard scores.

Z-score. There are many kinds of standard scores. The most useful is the z-
score, which is often used to express a raw score relation to the mean and
standard deviation. This relationship is expressed in the following formula:
̅

Let us recall that X – ̅ is a deviation score. With this difference, we

are able to know whether your test score, say, X is above or below the
average score. However, you cannot say whether your test score or grade is
better or worse than the average score. Again, the importance of knowing the
standard deviation is highlighted here. The standard deviation helps you
locate the relative position of the score in a distribution. The equation above
gives you the z-score, which can indicate the number of standard deviations
the score is above or below the mean. A z-score is called a standard score,
simply because it is a deviation score expressed in standard deviation units.
If raw scores are expressed as z-scores, we can see their relative
position in their respective distribution. Moreover, if the raw scores are already
converted into standard scores, we can now compare the two scores even
when these scores come from different distributions or when scores are
measure two different things, like knowledge in Physics or English. The
following figures illustrate this point.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 25

 SULTAN KUDARAT STATE UNIVERSITY

Figure 9. A Comparison of Score Distribution with Different

Means and Standard Deviations

In Figure 9, a score of 86 in English indicates better performance than

a score of 90 in Physics. Let us suppose that standard deviations in English
and Physics are 3 and 2, respectively. Thus, we can express the z-scores as:

From the above, if 86 and 90 are your scores in the two subjects, you
can confidently say that, compared with the rest of your class, you performed
better in English than in Physics. That is because in English, your
performance is 2 standard deviations above the mean, while in Physics, you
are 2.5 standard deviations below the mean. While 90 is numerically higher,
this score is more than half standard deviation below the average
performance of the class where you belong, while 86 is above the mean and
even 2 standard deviations above it. Having been transformed in the same
scale, the two graphs can be taken as one standard distribution of score.

Figure 10. Different Raw in one Z-score Distribution

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 26

 SULTAN KUDARAT STATE UNIVERSITY

We may want to look deeper into the scores. Note that in Figure 9, the
shaded areas in the two graphs indicate the proportion of scores below yours.
This is the same as saying that this proportion is the number of students in
your class who scored lower than you. But to be more specific on this
proportion, you need to look back at Figure 8 and the converted z-scores.
Examining the areas under the normal curve, we can say that about 98% in
the class scored below you in English, while only 1.2% scored below you in
Physics. We assume here that the scores are normally distributed. So, in what
subject makes you feel better?

T-Score. As you can see in the computation of the z-score, it can give us a
negative number, which simply means the score is below the mean. However,
communicating negative z-score as below the mean may not be
understandable to others. We will not even say to students that they got a
negative z-score. Also, a z-score may also be a repeating or non-repeating
decimal, which may not be comfortable for others. One option is to convert a
z-score into a T-score, which is a transformed standard score. To do this,
there is calling in which a mean of 0 in a z-score is transformed into mean of
50, and the standard deviation in a z-score is multiplied by 10. The
corresponding equation is:
T-score = 50 + 10z

For example, a z-score of -2 is equivalent to a T-score of 30. That is:

T-score = 50 + 10 (-2)

= 50 - 20

= 30

Looking back at the English score of 86, which resulted in a z-score of

2 as shown in Figure 9, T-score equivalent then is:
T-score = 50 + 10 (2)

= 50 + 20

= 70

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 27

 SULTAN KUDARAT STATE UNIVERSITY

Examining further Figures 9 and 10, where z-scores range from -4 to

+4, more than 99% of the scores are between 20 and 80.

Stanine Scores. Another standard score is stanine, shortened from standard

nine. With nine in its name, the scores are on a nine-point scale. In a z-score
distribution, the mean is 0, and the standard deviation is 2. Each stanine is
one-half standard deviation-wide. Like the T-score, stanine score can be
calculated from the z-score by multiplying the z-score by 2 and adding 5. That
is,
Stanine = 2z + 5

Going back to our example on a score of 86 in English, that is

equivalent to a z-score of 2 (Figure 10), its stanine equivalent is
Stanine = 2(2) + 5
= 9.

Example:
Scores in stanine scale have some limitations. Since they are in a 9-
point scale and expressed as a whole number, they are not precise. Different
z-scores or T-scores may have stanine score equivalent.

Z-score T-score Stanine

2.1 71 9
2.0 70 9
1.9 69 9

On the assumption that stanine scores are normally distributed, the

percentage of cases in each band or range of scores in the scales is as
follows:

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 28

 SULTAN KUDARAT STATE UNIVERSITY

With the above percentage distribution of scores in each stanine, you can
directly convert a set of raw scores into stanine scores. Simply arrange the
raw scores from lowest to highest, and with the percentage of scores in each
stanine, we can directly assign the appropriate stanine score in each raw
score. On interpretation of stanine score, let us say Kate has a stanine score
of 2. We can see that her score is somewhere at the low or bottom 7 percent
of the scores. In the same way, if John’s score is the 6 th stanine, it falls
between the 60th and 77th percentile, simply because 60 percent of the scores
are below the 6th stanine and 23 percent of the scores are above the 6 th
stanine. For qualitative description, stanine scores of 1, 2, and 3 are
considered as below average; 4, 5, and 6 are average, and 7, 8, and 9 are
above average. Thus, we can say that your score of 86 in English is above
average. Similarly, Kate’s score is below average while that of John is
average. Figure 11 illustrates the equivalence of the different commonly-used
standard scores.

Figure 11. The Normal Distribution and the Standard Score

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 29

 SULTAN KUDARAT STATE UNIVERSITY

Measures of Covariability
There are situations when we look at examinees’ performance
measures, and ask ourselves what could explain such scores. Measures of
covariablity tell us to a certain extent the relationship between two tests or
factors. Admittedly, a score one gets may not only be due to a single factor
but with other factors directly or indirectly observable which are also related to
one another. This section will be limited to introducing two scores that are
hypothesized to be related to one another.
When we are interested in finding the degree of relationship between
two scores, we are dealing with the correlation between two variables. The
statistical measure is the correlation coefficient, an index number that
ranges from -1.0 to +1.0. The value -1.0 indicates a negative perfect
correlation, 0.00 no correlation at all, and 1.00 a perfect positive correlation.
There have been many correlation studies conducted in the field of
assessment and research, but correlation coefficient values are either closer
to +1.0 or -1.0.
The various types of relationships are illustrated in the following
scatter-plot diagrams:

Figure 11. Various Types of Relationships

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 30

 SULTAN KUDARAT STATE UNIVERSITY

Some examples of interpreting Correlation coefficients are as follows:

To further analyze the relationship between two scores, we need a

statistical test to determine the correlation coefficient of two variables. In
actual practice, we deal with test scores, so the dataset is at interval and ratio
level of measurements. Hence, the commonly used statistical analysis is the
Pearson Product-Moment Correlation or Pearson r. The formula for this has
been earlier discussed and used when determining the reliability of test
through test-retest method and parallel forms.

Use of CORREL and PEARSON Functions

In this section, let us compute the correlation coefficient using Excel so
as avoid redundancy of lesson. Considering the data set in Table 3, we will
illustrate how the software works.

Table 3. Scores in Reading and Problem Solving Performance

X Y
Reading Problem Solving
4 6
9 11
4 5
11 10
12 8
5 5
7 10
5 8
6 9
2 5

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 31

 SULTAN KUDARAT STATE UNIVERSITY

First, we will encode the two (2) sets of raw data representing X and Y
in either two columns or rows in Excel as shown in the next illustration. At any
free cell, type =CORREL(A3:A12,B3:B12) as reflected in the worksheet. The
A3:A12 are the cell addresses of the first variable X, while B3:B12 are the cell
addresses of the second variable Y. Take note that the cell ranges of the 2
variables are separated by a comma.

Another way of finding the correlation coefficient in Excel is using

PEARSON function such as the one below, that is, =PEARSON(I1:I12,I2:R2).
As cited earlier, the correlation coefficient can be calculated even when data
are in array by column or row.

In contrast, the manual process will give us this complex process and
the same result.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 32

 SULTAN KUDARAT STATE UNIVERSITY

Obviously, the mathematical processes gave a correlation coefficient of

0.705 between performance scores in Reading and Problem-Solving. This
coefficient indicates a moderate correlation between performance in reading
and problem-solving.
There are some precautions to be observed with regard to the
computed r as correlation coefficient. First, it should not be interpreted as
percent. Thus, 0.705 should not be interpreted as 70%. If we want to extend
the meaning of 0.705, we need to compute the value for r2, which becomes
the coefficient of determination. This coefficient explains the percentage of
the variance in one variable that is associated with the other variable.
Remember how variance and standard deviation were explained in the
precious sections?
Now, with reference to the two variables indicated in Table 3, and the
computed r of 0.705, it results to an r2 of 0.49, which can be taken as 49%.
This is interpreted as 49% of the variance which is equal to 1 or 100%. If 49%
of the variance observed in problem-solving scores is attributable to reading
scores, then the other 51% of the variance is due to other factors. This
concept is depicted in Figure 12. Second, while the correlation coefficient
shows the relationship between two variables, it should not be interpreted as
causation. Considering our examples, we could not say that the scores in
reading test cause 49% of the variance of the problem-solving test scores.
Relationship is different from causation.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 33

 SULTAN KUDARAT STATE UNIVERSITY

Figure 13. Covariation of Performance in Reading and Problem-

Solving

The example in Table 3 presented only the computation of 10

observations. In actual class setting, we deal with bigger numbers of scores,
especially as classes become larger in public schools. As such, formulas (a)
and (b), as stated above, will be difficult to use, even with the scientific
calculator. Hence, statistical software as Excel and others will help us out in
finding the correlation coefficient.
Besides, if we want to supplement our analysis with a scatterplot
diagram, it is equally easy. With Excel application, here are few of the general
steps regardless of the version.
1. Encode the 2 sets of raw scores either in row or column.
2. Select the 2 columns or rows using the pointer.
3. Click the Insert command and select Chart.
4. From the options, choose XY SCATTER then click the chosen variant.
5. Once the Scatterplot graph is displayed, choose the preferred Layout.
6. Edit the title, axis labels, legend, gridlines, colors, fonts, etc. according
to your needs.
7. Finally, you get the desired graph as shown in Figure 14.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 34

 SULTAN KUDARAT STATE UNIVERSITY

12
10

Problem-solving
8
6
4
2
0
0 5 10 15
Reading

Figure 14. Correlation between Performance in Reading and

Problem Solving

Summary
 Measures of central tendency include the mean, median and mode.
 Mean is the average of given sets of scores. You should add the
numbers up then divide by the number of the cases.
 Median is the number in the middle when you order the numbers in an
ascending order. If there are two numbers in the middle, you should
take the average of those two numbers.
 Mode is the number which is repeated the most in the set.
 Measures of dispersion include the range, interquartile range, semi-
interquartile range, quartile deviation, variance and standard deviation.
 The range of a dataset is the difference between the largest and
smallest values in that dataset.
 The interquartile range is the middle half of the data that is in between
the upper and lower quartiles. Dividing this by 2 gives us the quartile
deviation.
 Variance is the average squared difference of the values from the
mean. Unlike the previous measures of variability, the variance
includes all values in the calculation by comparing each value to the
mean.
 The standard deviation is the standard or typical difference between
each data point and the mean.
 Measures of location can be categorized as percentile, decile and
quartile.
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 35
 SULTAN KUDARAT STATE UNIVERSITY

 A normal distribution is sometimes called the bell curve because its

distribution occurs naturally in many situations. The bell curve is
symmetrical where half of the data falls to the left of the mean and
another half falls to the right.
 Standardized scores can be a z-score, T-score or stanine.
 Measures of covariablity tell us to a certain extent the relationship
between two tests or factors.

Assessment
A. Answer the following questions orally.
1. What are the measures of central tendency?
2. What are the measures of dispersion?
3. What are the measures of position?
4. What is covariability?
5. Differentiate the different ways of converting raw scores into standard
scores?
B. Given the following test scores in Math: 5, 20, 13, 15, 12, 19, 20, 10, 17,
11& 16.
1. What is the mean? How did you find it?
2. What is the median? Why did you say it is the median?
3. What is the mode? How did you determine this measure?
C. Examine the following frequency polygons below:

1. Which distribution has the highest mean, X, Y or Z? What is the

estimated mean of each distribution?
2. Which distribution is symmetrical? Which is asymmetrical? Why?

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 36

 SULTAN KUDARAT STATE UNIVERSITY

3. In which distribution are the scores most dispersed? What range of

scores is indicated by this distribution?
4. Which distribution has the smallest standard deviation? Why did you
say so? How do you describe this distribution?
D. Let me see further how you understand the earlier discussion by
considering the following class scenario:
Ms. Dioneza’s best students in Spelling obtained the following scores in a
competitive test:
89, 90, 91, 92, 93, 94, 95, 95, 96, 98
Dr. Protacio’s best students in Spelling got the following scores in the
same competitive test:
60, 90, 92, 93, 93, 94, 96, 96, 97, 98
1. Which measure of central tendency is most appropriate for the
students of Ms. Dioneza? Why did you say so? Then, what is the value
of that measure?
2. Which measure of central tendency is most appropriate for the
students of Dr. Protacio? Why? Then, what is the value of that
measure?
3. For each group, find the following:
a. Range
b. Q1, Q2 & Q3
c. Semi-interquartile range
d. Variance
e. Standard deviation
E. Determine the type of distribution depicted by the following measures:
1. ̅ = 80.45 Mdn = 80.78 Mode = 80.25
2. ̅ = 120 Mdn = 130 Mode = 150
3. ̅ = 89.78 Mdn = 82.16 Mode = 82.10

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 37

 SULTAN KUDARAT STATE UNIVERSITY

F. The following is frequency distribution of scores of 10 persons

X f
30 1
28 2
20 3
15 1
13 2
10 1

1. What is the mean of the distribution?

2. What is the median?
3. What is the mode?
4. What is the range?
5. What is the standard deviation?
G. After all what you have done, can you now identify the elements that are
implied in the empty boxes below?

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 38

 SULTAN KUDARAT STATE UNIVERSITY

H. Refer to the figure below as the frequency polygons representing entrance

test scores of three groups of students in different programs.
1. What is the mean score of Education students?
2. What is the mean score of Engineering students?
3. What is the mean score of Business students?
4. Which group of students has the most dispersed scores in the test?
Why?
5. What distribution is symmetrical? What distribution is skewed? Why?
I. The following is a frequency distribution of year-end examination marks in
a certain secondary school.
Class Interval f
60 - 65 2
55 - 59 5
50 - 54 6
45 - 49 8
40 - 44 11
35 - 39 10
30 - 34 11
25 - 29 20
20 - 24 17
15 - 19 6
10 - 14 4
N = 100

1. Compute the mean, median, and mode of the frequency distribution.

2. Compare the measure of average in (a) from this comparison, is the
distribution positively skewed, negatively skewed, or symmetrical?

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 39

 SULTAN KUDARAT STATE UNIVERSITY

3. Make a sketch of the graph of the frequency distribution. Describe the

graph of the distribution as to its skewness.
4. Find:
a. Third quartile or the 75th percentile (P75)
b. First quartile or the 25th percentile (P25)
c. Semi-interquartile range
J. A common exit examination is given to 400 students in a university. The
scores are normally distributed and the mean is 78 with a standard
deviation of 6. Paul had a score of 72 and Vic a score of 84. What are the
corresponding z-score of Paul and Vic? How many students would be
expected to score between the scores of Paul and Vic? Explain your
answer.
K. Jayson obtained a score of 40 in his Mathematics test and 34 in his
Reading test. The class mean score in Mathematics is 45 with a standard
deviation of 4 while in Reading, the mean score is 50 with a standard
deviation of 7. On which test did Jayson do better compared to the rest of
the class? Explain your work.
L. Following are sets of scores on two variables: X for Reading
Comprehension and Y for Reasoning Skills administered to sample of
students
X 11 9 15 7 5 9 8 4 8 11
Y 13 8 14 9 8 7 7 5 10 12

1. Compute the Pearson Product-Moment Correlation for the above data.

2. Describe the direction and strength of the relationship between reading
comprehension and reasoning skills
3. Compute the coefficient of determination. Interpret the results.
M. Your learning in this chapter will be more meaningful if you will be
engaged in actual situations beyond the classroom. The examples that
have been presented are basically hypothetical for the purpose of
concretizing the concepts and principles intended to be learned. This time,
the following activities will require your engagement in gathering of

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 40

 SULTAN KUDARAT STATE UNIVERSITY

empirical data that are experiential and will depict the true picture in the
field of education.

Activity 1
Identify yourself in a group of four classmates. Here is the task you need to
work your team members.
 Secure a set of old test papers that have been scored by a teacher or a
student teacher. It is advised that number of cases be at least 100. It is
understood that the 100 scores came from the same test. See to it that
you observe confidentially of the documents you have requested. No
name should be identified in your written work, but use codes to
identify the observations or cases.
 Here are the tasks:
1. Prepare a data set for the test scores.
2. With the aid of your scientific calculator, or Microsoft Office Excel,
find the following:
a. Mean
b. Median
c. Mode
3. Describe the type of distribution of test scores.
4. Which measure of central tendency is most appropriate to describe
the distribution? Why do you say so? Explain.
5. Select three (3) students in the list. Describe the test performance
of each student relative to the performance of all the students in the
whole class.

Activity 2
Interview a teacher on how she decides to pass or fail a student. Report the
specific questions that you have asked and the corresponding responses you
have captured in this interview activity. Your analysis and presentation should
reflect the application of the measures of central tendency, measures of
variability, and of the measures of location.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 41

 SULTAN KUDARAT STATE UNIVERSITY

Activity 3
With your team members, make a visit to any of the following offices of a
school:
1. Office of the Guidance Councilor
2. Testing Center
3. Office of Student Affairs
Request for any of the following data which you can avail of:
1. IQ test scores
2. Aptitude test scores
3. Admission test scores
4. Qualifying examination scores
5. Any psychological test scores
It is advised that the number of observations should at least be 50. The
larger the number of cases would be better.
Please emphasize the data will be held in full confidentiality. They have
the option not to give the names of the examinees as ethical consideration.
Do not apply coding technique to label the different observation.
From the test scores that you have gathered, do the following:
1. Make a frequency distribution of the test scores.
2. Find the following measures: mean, median, and mode.
3. Describe the type of distribution.
4. Determine the most appropriate measure of central tendency to
describe the average IQ/Aptitude/Performance of the examinees.
5. For those who gather IQ test scores, interview the personnel in-charge
on how they interpret the numerical value of IQ test scores (as above
average, average, etc. whatever applies).
6. Find the standard deviation.
7. Consider at least 5 student scores in you data set and try computing
the ff:
a. z-scores
b. T-scores
c. Percentile equivalent
d. Stanine scores
8. Make a report you your class on the following:
| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 42
 SULTAN KUDARAT STATE UNIVERSITY

a. The methods and procedures in the test data collection

b. Summary of the data set to your class using tables, figures, and
graphs
c. Summary of your findings on question 1-7 above.
N. Test Yourself.
In each item, choose the letter which you
think can best represent the answer to
given problem situation. Give a statement
to justify your choice.
1. Which distribution is negatively skewed
in the figure shown?
A. Distribution X
B. Distribution Y
C. Distribution Z
D. Distribution W
2. What is the preferred measure of central tendency in a test distribution
where there are small number of scores that are extremely high or low?
A. Median C. Variance
B. Mean D. Mode
3. The following scores were obtained from a short spelling test:
10, 8, 11, 13, 12, 13, 8, 16, 11, 8, 7, 9.
What is the modal score?
A. 8 C. 11
B. 10 D. 13
4. What does it mean when a student got a score at the 70th percentile on
a test?
a. The performance of the student is above average
b. The student answered 70 percent of the items correctly
c. The student got at the least 70 percent of the correct answers
d. The student’s score is equal or above 70 percent of the other
students in the class
5. Which best describes a normal distribution?
A. Positively skewed c. symmetric
B. Negatively skewed D. bimodal
6. What does a large standard deviation indicate?
A. Scores are not normally distributed
B. Scores are not widely spread, and the median is unreliable
measure of central tendency.
C. Scores are widely distributed where the mean may not be a reliable
measure of central tendency.
D. Scores are not widely distributed, and the mean is recommended
as more reliable measure of central tendency

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 43

 SULTAN KUDARAT STATE UNIVERSITY

7. In a normal distribution, approximately what percentage of scores is

expected to fall within three-standard deviation from the mean?
A. 34% C. 95%
B. 68% D. 99%
8. Which of the following is interpreted as the percentage of scores in a
reference group that falls below a particular raw score?
A. Standard scores C. Reference group
B. Percentile rank D. T-score
9. For the data illustrated in the scatter plot
below, what is the reasonable product-
moment correlation coefficient?
A. 1.0
B. -1.0
C. 0.90
D. -0.85
10. A Pearson test statistics yields a
correlation coefficient (r) of 0.90. If X
represents scores on vocabulary and Y,
the reading comprehension test scores, which of the following best
explains r=0.90?
A. The degree of association between X and Y is 81%.
B. The strength of relationship between vocabulary and reading
comprehensive is 90%.
C. There is almost perfect positive relationship between vocabulary
test scores and reading comprehension.
D. 81% of the variance observed in y can be attributed to the variance
observed in X.
O. Refer to these test data:
38 15 10 27 18 26 27 30 17 43
21 36 27 34 49 31 40 20 30 36
42 23 34 39 46 21 39 33 37 34
26 43 27 28 34 36 27 35 47 45
18 28 22 42 20 18 35 27 47 34
60 51 63 37 43 60 38 25 45 18
31 29 64 18 31 28 32 22 18 49
22 21 21 31 26 27 27 32 41 40
35 38 34 24 37 35 35 22 38 25
19 28 28 42 33 25 27 25 50 18

a. How do you find the mean if data are grouped? What is the most
appropriate class interval for this set of data? What other information
you need to generate and achieve the task of what is the mean?
b. How do we determine the median of the scores?

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 44

 SULTAN KUDARAT STATE UNIVERSITY

 In what class interval did the median fall?

 What is the cumulative frequency below the median class?
 What is the frequency of the median class?
 What is the lower limit of the median class?
 What is the median equal to?
P. Take hold of a specific calculator. Use the raw score formula in finding the
standard deviation. If you have Excel or whatever software, you can
directly work on the data.
1. What is the variance?
2. What is the standard deviation?
Q. Assuming that the scores are normally distributed, what is the range of
scores the would fall between:
1. ±1σ from the mean. Explain your answer.
2. ±2σ from the mean. How did you determine this ranger of scores?
3. ±3σ from the mean. Explain your answer.
4. Define the 25th percentile. How did you determine it?
th
5. Where is the 75 percentile? How did you locate it? How many fall on
this percentile?
6. How many got a percentile rank of 99? How did you determine it?
Explain.
R. With the help of your scientific calculator, Excel, or whatever, load a data
set - either hypothetical data or an actual data - you have accessed from
additional documents or any research studies. Analyse the data and enter
the values into the following table:
Mean
Median
Mode
Range
Variance
Standard Deviation
Skewness
Kurtosis

Select an appropriate type of graph (bar chart, histogram, pie graph,

box plot, and line) for this variable and have Excel or SPSS draw the
graph. Why did you select this particular type of graph?

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 45

 SULTAN KUDARAT STATE UNIVERSITY

Examine the values you have written in the table. Discuss the graph in
relation to the values in the table.

References
Cheusheva, Svetlana (September 4, 2020). Mean, median and mode in
Excel. Retrieved from https://www.ablebits.com/office-addins-
blog/2017/05/24/mean-median-mode-excel/
David et al. (2020). Assessment in Learning 1. Manila: Rex Book Store.
De Guzman, E. and Adamos, J. (2015). Assessment of Learning 1. Quezon
City: Adriana Publishing Co., Inc.
Frost, Jim (n.d.).Measures of Variability: Range, Interquartile Range,
Variance, and Standard Deviation. Retrieved from
https://statisticsbyjim.com/basics/variability-range-interquartile-
variance-standard-deviation/.
McLeod, Saul (2019). Introduction to the Normal Distribution (Bell Curve).
Retrieved from https://www.simplypsychology.org/normal-
distribution.html
Mean, Median, Mode: What They Are, How to Find Them. Retrieved from
https://www.statisticshowto.com/probability-and-statistics/statistics-
definitions/mean-median-mode/
Statisticsfun (September 20, 2009). How to calculate Standard Deviation,
Mean, Variance Statistics, Excel. [Video]. YouTube:
https://www.youtube.com/watch?v=efdRmGqCYBk.
Study.com (2020). Skewness in Statistics: Definition, Formula & Example.
[Video]: https://study.com/academy/lesson/skewness-in-statistics-
definition-formula-example.html
Surbhi S. (February 27, 2017). Differences Between Skewness and Kurtosis.
Retrieved from https://keydifferences.com/differences-between-
skewness-and-kurtosis.html.
The Organic Chemistry Tutor (May 29, 2018). Calculating The Standard
Deviation, Mean, Median, Mode, Range, & Variance Using Excel.
[Video]. YouTube: https://www.youtube.com/watch?v=k17_euuiTKw.

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 46

 SULTAN KUDARAT STATE UNIVERSITY

| ERNIE C. CERADO, PhD & MA. DULCE P. DELA CERNA, MIE 47