Part 1

Descriptive statistics

1.1. Exercises
1. Determine, for each case, the population and the variable studied. Specify if the variable
is qualitative or quantitative; in the first case, tell whether it is categorical or ordinal,
and, in the second, if it is discrete or continuous:

a) Time devoted to housework by people working outside.

b) Type of studies that students from a school want to tackle after Secondary school.
c) Voting intention in some local elections.
d ) Time spent watching TV by university students.
e) Number of radio receivers in the Spanish homes.

2. Give two examples of the following types of statistical variables: qualitative categorical,
qualitative ordinal, quantitative discrete and quantitative continuous.

3. What type of statistical variable is the temperature? And if we are only able to observe
the values 10o C, 30o C and 50o C?

4. In a quality control study of a machine the number of breakdowns in the last 50 weeks
has been noted, and weeks with 2, 3, 4 and 5 breakdowns have been observed. Compute
the number of weeks with 4 breakdowns in each of the following situations:

a) 56 % of the weeks it had 2 breakdowns, 32 % it had 3 breakdowns and 2 % it had 5

breakdowns.
b) In 43 weeks it had at most 4 breakdowns and in 20 weeks it had at least 4 breakdowns.
c) In 54 % of the weeks it had at most 3 breakdowns and in 38 % of the weeks it had
exactly 5 breakdowns.

1
2 PART 1. DESCRIPTIVE STATISTICS

5. Complete the following frequency table:

xi ni Ni fi Fi
1 4 0.08
2 4
3 16
4 7
5 28
6 38
7 0.14
8

6. Give an adequate graphical representation for data from the following variables: oven
brand, lifetime, number of breakdowns and exam mark (assume this one takes the values
A,B,C,D,E).

7. The following pie chart describes the number of accelerated wearing tests which were
necessary to break 2000 tyres.

a) Interpret the pie chart.

b) How many tyres broke in the first test? And in the third one?
c) Obtain the frequency table of the variable “number of tests necessary to break the
tyre”.
d ) Represent this variable in a bar chart.
1.1. EXERCISES 3

8. The following bar charts represent the marks obtained in an exam in four different groups
of the same subject (the marks range from zero to ten).

a) In one of these data the mean is 3.5 and the median is 3, which one is it?
b) In which of these charts the mean and the median are alike?
c) Among these, which one is better represented by the mean?
d ) In which group the median is greater than the mean?

9. The manufacture temperature of 200 pieces has been noted, with the results represented
in the following pie chart:

According to these data,

a) What is the average manufacture temperature?

b) How much is the median of the manufacture temperature?
c) Which of these two measures is less affected by data errors or outliers?

10. The following table gives us the yearly rainfall in Oviedo between 1976 and 1995:

Year 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
Rainfall (mm) 837 1079 1105 1294 1158 741 989 934 1174 726

Year 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
Rainfall (mm) 934 909 905 926 908 936 1250 1031 1075 798

a) Compute the average rainfall, and the median.

4 PART 1. DESCRIPTIVE STATISTICS

b) Group these data in the classes [700, 900], (900, 1000], (1000, 1100], (1100, 1300], and
determine their frequency table.
c) Represent the data in (b) in a histogram of densities.

11. The following frequency table gives us the number of correct answers obtained in a test
of 5 questions by a group of 50 students:

xi ni fi Ni Fi
0 4
1 0.1
2 0.36
3 40
4 6
5

(a) Complete the frequency table.

(b) Determine the mean, median and mode.

12. The following sample represents the daily number of absences in a factory in the last 20
days:
2 0 0 1 3 1 0 0 0 2 3 3 3 2 1 0 0 0 1 0

Represent these data in a frequency table and a pie chart, and compute the percentile 60.

13. The number of contaminant particles in a silicon chip before a washing process has been
observed in a sample of size 100, with the following frequencies:

Number of particles 0 1 2 3 4 5
Number of chips 3 15 44 26 11 1

a) What is the percentage of chips with at least one particle? And with at least four
particles?
b) What percentage of chips had more than one and less than four particles?
c) Represent these data in a bar chart.
d ) Determine the mean of the number of contaminant particles in a silicon chip.
e) Compute the mode, range and percentile 62 of this sample.
1.1. EXERCISES 5

14. The following bar chart represents the minimum daily temperature in a specific place
during a 30-day month:

Compute the median, range, mode, mean temperature and the 35-th percentile.

15. In a quality control study over four factories, we have noted the number of flawed products
per batch in a day. Determine the median number of flawed products for each factory,
from the following data:

1. The first factory produced 5 batches with the following number of flawed products:
1, 3, 7, 9, 14.
2. The second factory produced 4 batches with the following number of flawed products:
1, 3, 6, 9.
3. The third factory produced 50 batches which had the following relative frequencies
of flawed products:
Number of flawed products (xi ) 1 3 7 9
Percentage of batches 20 % 40 % 30 % 10 %
4. The fourth factory produced 100 batches which had the following frequencies of
flawed products:
Number of flawed products (xi ) 1 3 7 9
Number of batches 20 10 20 50
6 PART 1. DESCRIPTIVE STATISTICS

16. The shear resistance, in MPa, of a joint, has been observed in a sample, obtaining the
following ordered data:

4.4 − 8.7 − 16.4 − . . . . . . . . . . . . . . . . . . . . . − 79.6 − 81.5 − 209.9

Analyzing these data with R, we have obtained the following results:

mean sd 0% 25% 50% 75% 100% n

55.9 56.53001 4.4 22.2 36.6 73.7 209.9 101

a) Determine the mean and the median.

b) If the third observation was 20.0 instead of 16.4, would the mean or the median
change?
c) Which are the quartiles?
d ) Represent these data in a box-plot and interpret it. Say if there are any outliers.

17. The following box-plot provides a graphical representation of some data.

17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1

Determine, if possible, the median, the percentile 75, the first quartile, the inter-quartile
range, the percentile 25, the third quartile, the mean and the range. Is 16 an outlier? In
that case, should it be eliminated from the sample?

18. The number of new contracts signed by a company in the last 20 days is given by:

3 0 1 2 3 5 2 0 1 0 5 3 2 0 3 0 1 2 2 4

Compute the variance, the range and the interquartile range.

1.1. EXERCISES 7

19. The following histogram of frequencies summarizes a sample of petrol consumption of

some cars.
10
9
8
7
6
5
4
3
2
1

23’5 24 24’5 25 25’5 26 26’5

Say if the following statements are true or false:

a) The median is 0.5.

b) The second quartile is 0.5.
c) The first quartile is 0.25.
d ) The mean is 18.
e) The median is 26.
f ) The mean lies between 23.5 y 26.5.
g) The mean is 27.
h) The variance is -3.
i ) The sample size is 45.

20. The following data summarize the salaries of a sample of 100 people from a population:

> numSummary(Data[, "salary"], statistics = c("mean", "sd",

+ "quantiles"), quantiles = c(0, 0.25, 0.5, 0.75, 1))

mean sd 0% 25% 50% 75% 100% n

1100 129 400 550 1200 1400 4000 100

Moreover, the ordered sample is given by

400 420 425 . . . 2500 2650 4000.

Represent these data in a box-plot, and interpret the graph. Are there outliers? What is
the coefficient of variation?

21. The best 10 marks in Statistics and Algebra in the past year are given by:

Statistics 9.42 9.12 8.65 8.6 8.55 8.2 8.2 8.1 8 7.8
Algebra 9.06 8.94 8.86 8.45 8.45 8.31 8.2 8 7.95 7.9
8 PART 1. DESCRIPTIVE STATISTICS

a) Compute the quantile 0.3 of the marks of Statistics.

b) In which of the two cases is the mean more representative?
c) Represent the mark of Statistics in a box-plot, and discuss the result.

22. The following data tell us the number of times the first prize of the Christmas lottery has
finished on each digit:

Digit 0 1 2 3 4 5 6 7 8 9
Absolute frequency 20 8 13 20 27 32 27 20 21 16

a) Compute the proportion of times the first prize has gone to a multiple of 3.
b) Which are the adequate graphical representations for this variable?
c) Complete the frequency table.
d ) Determine the median.

23. For these variables, say in which cases it makes sense to compute the mode, median,
mean, third quartile and standard deviation:

DESCRIPTIVE Oven Lifetime Number of Exam mark

MEASURE brand breakdowns (A,B,C,D,E)
Mode
Median
Mean
Third quartile
Standard deviation

1.2. Test
(There may be more than one correct answer)

1. In a sample of a variable taking the values 0, 1 and 2, 20 % of the observations are equal
to zero and 70 % of the observations are smaller or equal to one. Then the percentage of
observations equal to 2 is:

a) 20 %
b) 30 %
c) 50 %
d ) 70 %
1.2. TEST 9

2. With the previous data, we have that:

a) The sample mean is 1

b) The sample median is 1
c) The sample mean is 1.1
d ) The sample median is 1.1

3. From the following frequency table,

xi ni
0 3
1 2
2 2

we can observe that:

a) The sample mean is 1

b) The sample median is 1
c) The sample mean is 6/7
d ) The third quartile is 6/7

4. If the mean of a sample of size 10 is equal to 50, we have that:

a) Five data are greater or equal to 50 and five data are smaller or equal to 50
b) At least one element of the sample must be smaller or equal to 50
c) The most frequent value is 50
d ) At least one element of the sample is greater or equal to 50

5. If a sample mean is x = 6, then the sample median must be:

a) Greater than 6
b) Smaller than 6
c) Equal to 6
d ) It may be greater, smaller or equal to 6

6. Given the sample 1,4,4,6,6,7,7, we have that:

a) Its median is 6
b) Its 50-th percentile is 6
c) Its mean is 5
d ) Its range is 6
10 PART 1. DESCRIPTIVE STATISTICS

7. A box-plot is a graphical representation which always shows:

a) The coefficient of variation

b) The mean and variance
c) The median and the first and third quartiles
d ) The median and the variance

8. The inter-quartile range of a variable:

a) May coincide with the data range

b) May be greater than the data range
c) May be smaller than the data range
d ) Must be smaller than the data range

9. Given the sample 3, 5, 6, 6, 7, 10, 12, we have that:

a) The inter-quartile range is 35

b) The median is 6
c) The second quartile is 6
d ) The range is 12

10. The voltage resistance (in psi) in a sample of aluminium-lithium alloy are 92, 120, 200
and 140. With these data, 2096 is the value of:

a) The sample variance

b) The range
c) The inter-quartile range
d ) The standard deviation

11. What is the criterion to consider an observation an outlier?

a) It does not belong to the interval [Q1 − 1.5IQR, Q3 + 1.5IQR].

b) It does not belong to the interval [Q1 , Q3 ].
c) It does not belong to the interval [Q1 − 3IQR, Q3 + 3IQR].
d ) It is at a distance 1.5IQR from any other observation.

12. How do we determine the number of degrees of a category in a pie chart?

a) It is the relative frequency, fi .

b) It is the absolute frequency, ni .
c) It is the relative frequency multiplied by 360.
1.3. SOLUTIONS TO THE EXERCISES 11

d ) It is the absolute frequency multiplied by 360.

13. In which type of histogram the areas of the rectangles are equal to the relative frequencies?

a) Histogram of percentages.
b) Histogram of densities.
c) Histogram of frequencies.
d ) Histogram of areas.

14. Which of the following conditions does NOT necessarily hold when the sample variance
is 0?:

a) That all the data coincide.

b) That the coefficient of variation is equal to 0.
c) That the interquartile range is equal to 0.
d ) That the mean and the standard deviation coincide.

15. If we know the relative frequencies and the sample size, can we determine the cumulative
absolute frequencies?

a) Yes, adding the relative frequencies corresponding to the previous data and mul-
tiplying by the sample size.
b) Yes, multiplying by the sample size.
c) Yes, adding the relative frequencies associated with the previous data.
d ) No.

1.3. Solutions to the exercises

1. a) Population: people working outside; Variable: time devoted to housework; Type:
quantitative continuous.
b) Population: students from the school Variable: studies they want to tackle after
secondary school; Type: qualitative categorical.
c) Population: voters in the local elections; Variable: political option; Type: qualitative
categorical.
d ) Population: university students; Variable: time devoted to watching television; Type:
quantitative continuous.
e) Population: Spanish homes; Variable: number of radio receivers; Type: quantitative
discrete.
12 PART 1. DESCRIPTIVE STATISTICS

2. Qualitative categorical: type of product (conforming and non-conforming); type of con-

crete (A, B or C).
Qualitative ordinal: mark in an exam (A, B, C, D, E); temperature of an oven (low, me-
dium, high).
Quantitative discrete: number of flawed screws in a batch; number of siblings in a family.
Quantitative continuous: diameter of a piston; width of a component.

3. Quantitative continuous, but if we can only observe the values 10ºC, 30ºC and 50ºC it
becomes quantitative discrete.

4. The number of weeks with 4 breakdowns in each of the situations is the following:

a) 5 weeks.
b) 13 weeks.
c) 4 weeks.

5. The frequency table is:

xi ni fi Ni Fi
1 4 0.08 4 0.08
2 4 0.08 8 0.16
3 8 0.16 16 0.32
4 7 0.14 23 0.46
5 5 0.1 28 0.56
6 10 0.2 38 0.76
7 7 0.14 45 0.9
8 5 0.1 50 1

6. Oven brand: it is qualitative categorical, so we may use a pie chart or a bar chart.
Lifetime: it is quantitative continuous, so we may represent it by means of a histogram
or a box-plot.
Number of breakdowns: it is quantitative discrete, so we may use a pie chart or a bar
chart.
Mark in an exam: it is qualitative ordinal, so we may use a bar chart or a pie chart.

7. a) Out of the tyres studied, only 15 % broke in the first or the second test, while in
the third test this percentage increases to 80 %, which makes it difficult for a tyre
to overcome the third test. Indeed, only 5 % of them do it and all of them break in
the fourth test. Depending on the resistance level desired for the tyres, these data
would have different interpretations in terms of the quality of the tyres.
b) Out of the 2000 tyres studied, 100 of them broke in the first test and 1600 in the
third one.
1.3. SOLUTIONS TO THE EXERCISES 13

c) The frequency table of this sample is:

xi n i fi Ni Fi
1 100 0.05 100 0.05
2 200 0.1 300 0.15
3 1600 0.8 1900 0.95
4 100 0.05 2000 1

d ) The bar chart associated with these data is:

8. a) Group D. In the others, the mean and median are greater or equal to 5.
b) Groups A and C are the more symmetrical ones.
c) Group A, because the values closer to the mean are more frequent than the ones
away from it (it would have a smaller coefficient of variation).
d ) Group B.

9. a) 47.5o C.
b) 50o C, because F50o C = 0.8 and F40o C = 0.45.
c) The median.

10. a) The average yearly rainfall is given by 837+1079+···+1075+798

20
= 985.45 mm. In order
to compute the median, we must order the data from the smallest to the greatest
value, and take the central one. Since the sample size is n = 20, we take the average
of the interval [X(10) , X(11) ] = [934, 936], that is, 935.
b) The frequency table when the data are grouped in these intervals is given by:

Ii ni fi Ni Fi
[700,900] 4 0.2 4 0.2
(900,1000] 8 0.4 12 0.6
(1000,1100] 3 0.15 15 0.75
(1100,1300] 5 0.25 20 1
14 PART 1. DESCRIPTIVE STATISTICS

c) In order to draw the histogram of densities, we must take into account that the area
of each rectangle must be proportional to the frequency of the interval determined
by its base. In this way, we obtain the following heights:
Ii fi hi
[700,900] 0.2 0.001
(900,1000] 0.4 0.004
(1000,1100] 0.15 0.015
(1100,1300] 0.25 0.0125
and thus the following histogram:

11. (a) The complete frequency table is:

xi ni fi Ni Fi
0 4 0.08 4 0.08
1 5 0.1 9 0.18
2 9 0.18 18 0.36
3 22 0.44 40 0.8
4 6 0.12 46 0.92
5 4 0.08 50 1
ni
In order to obtain it, we must take into account that fi = n
, Ni = n1 + · · · + ni ,
Fi = f1 + · · · + fi and that n = 50.
(b) The mean is given by

x = 0 · 0.08 + 1 · 0.1 + · · · + 5 · 0.08 = 2.66

1.3. SOLUTIONS TO THE EXERCISES 15

Since the cumulative relative frequency of 2 is 0.36 < 0.5 and that of 3 is 0.8 > 0.5,
we deduce that the median is equal to 3.
The mode is the most frequent value. By looking at the column of the absolute
frequencies, we conclude that the mode is equal to 3.

12. The frequency table for this sample is:

xi ni fi Ni Fi
0 9 0.45 9 0.45
1 4 0.2 13 0.65
2 3 0.15 16 0.8
3 4 0.2 20 1

The pie chart of these data is:

Pie chart

0
Frequency

In order to build it, we must take into account that the number of degrees of the pie
associated with value xi is oi = 360 · fi .
Finally, since the cumulative relative frequency of 0 is 0.4 < 0.6 and that of 1 is 0.65 > 0.6,
the percentile 60 is 1. We can also see this noting that the values 12-th and 13-th in the
ordered sample are both equal to 1.

13. a) The percentage of chips with at least one particle is 97 %. The percentage of chips
with at least four particles is 12 %.
b) 70 %.
16 PART 1. DESCRIPTIVE STATISTICS

c)

d ) The mean is
100
1 X 1
x= xi · n i = (0 · 3 + 1 · 15 + 2 · 44 + 3 · 26 + 4 · 11 + 5 · 1) = 2.3 particles.
100 i=1 100

e) The mode is the value with the greatest absolute frequency, in this case M o = 2.
The range is the difference between the greatest and the smallest observation, so in
this case we obtain R = 5 − 0 = 5.
Since the cumulative relative frequency of 2 is 0.62 and that of 3 is 0.78 > 0.62, we
conclude that the percentile 62 is equal to 2+3
2
= 2.5.

14. According to the bar chart, the frequency table is:

xi ni fi Ni Fi
−1 1 0.0333 1 0.0333
0 2 0.0667 3 0.1
1 4 0.1333 % 7 0.2333
2 5 0.1667 % 12 0.4
3 6 0.2 18 0.6
4 4 0.1333 22 0.7333
6 4 0.1333 26 0.8667
9 1 0.0333 27 0.9
10 3 0.1 30 1

With these data we obtain that the median is M e = 3, the range is R = 10 − (−1) = 11,
the mode is M o = 3 and the mean is x = 110/30 = 3.67, i.e., the average minimum tem-
perature is of 3.67 Celsius). The 35th percentile is the smallest element whose cumulative
percentage is greater than 0.35, that is, P35 = 2.
1.3. SOLUTIONS TO THE EXERCISES 17

15. 1. 7 flawed products.

2. 4.5 flawed products.
3. 3 flawed products.
4. 8 flawed products.

16. a) The mean is x̄ = 55.9 and the median is M e = 36.6.

b) The mean is now 56.06; the median does not change.
c) The first quartile is 22.2 and the third is 73.7.
d ) 25 % of the observations lie between 22.2 and 36.6, and another 25 % lie between
36.6 and 73.7. All the values can be considered typical except for 209.9MPa, which
goes beyond the upper bound of “normality” (the interval of normal values would be
[22.2 − 1.5 · 51.5, 73.7 + 1.5 · 51.5] = [−55.05, 150.95]). Moreover, the box-plot shows
that the amount of values between 22.2 and 36.6 is very high.
200

6
209'9
150
100
50
0

17. The median is 4, the 75-th percentile is 6, the first quartile is 3, the inter-quartile range is
3, the 25-th percentile is 3, the third quartile is 6, the mean cannot be obtained and the
range is 14. 16 is an outlier, but it should only be eliminated if we are sure that there has
been an error in the observation (and not if it is a rare event that has been observed).

18. The mean of these data is given by:

3 + 0 + ··· + 4
x= = 1.95 contracts
20
and as a consequence the variance is

(3 − 1.95)2 + (0 − 1.95)2 + · · · + (4 − 1.95)2

s2 = = 2.57 contracts.
19
Since the greatest value is 5 and the smallest one is 0, we deduce that the range is equal
to 5.
Finally, the frequency table is given by:
18 PART 1. DESCRIPTIVE STATISTICS

xi ni fi Ni Fi
0 5 0.25 5 0.25
1 3 0.15 8 0.4
2 5 0.25 13 0.65
3 4 0.2 17 0.85
4 1 0.05 18 0.9
5 2 0.1 20 1

0+1
We deduce that Q1 = 2
= 0.5 and Q3 = 3, whence the interquartile range is IQR =
3 − 0.5 = 2.5.

19. a) False, the median must line in the range of values in the sample, that is, between
23.5 and 26.5.
b) False, the second quartile should also lie in the range of values in the sample.
c) False, for the same reason.
d ) False, for the same reason.
e) False, in the histogram we see that 26 does not leave 50 % of the observations on
each side.
f ) True, from the form of the histogram.
g) False, because 27 ∈
/ [23.5, 26.5].
h) False, because the variance cannot be negative.
i ) True, it suffices to add the frequencies represented in the graph.

20. The box plot associated with these data is given by:

An element is considered an outlier when it does not belong to the interval [Q1 −
1.5IQR, Q3 +1.3IQR]. In this case, the interval becomes [550−1.5·850, 1400+1.5·850] =
1.3. SOLUTIONS TO THE EXERCISES 19

[−725, 2675]. Using the information from the ordered sample, we deduce that the extre-
mes of the whiskers are 400 (the minimum) and 2600 (the maximum observation that
does not exceed 2675). Since the maximum of the data is 4000, we see that there is at
least one outlier.
Note also that it makes sense to compute the coefficient of variation since all the data are
positive. We obtain
sd
CV = = 1291100 = 0.117,
x
that we can interpret saying that in this case the mean is very representative.
21. a) Since the have 10 data, the quantile 0.3 will correspond to the average of those that
occupy the third and the fourth positions in the ordered sample (from smallest to
greatest value). In this case, we obtain q0.3 = 8.1+8.2
2
= 8.15.
b) To see in which of the two cases the mean is more representative, we must compute
the coefficients of variation. If we denote by X the mark in Statistics and by Y the
mark in Algebra, we obtain
0.509
x = 8.464, sX = 0.509, CV (X) = = 0.06
8.464
0.422
y = 8.412, sY = 0.422, CV (Y ) = = 0.05
8.412
Since the coefficient of variation is smaller for the marks of Algebra, we conclude
that the mean is more representative in that case.
c) The box-plot of the mark of Statistics is given by:

Note that it makes sense to make this graphical representation because the variable
mark is quantitative and continuous. We see that there are no outliers: the quartiles
are given by Q1 = 8.1, Q3 = 8.65, and as a consequence the interval of normal data is
given by [8.1 − 1.5 · 0.55, 8.65 + 1.5 · 0.55] = [7.275, 9.475], that includes in particular
the minimum and maximum values (7.8 and 9.42, respectively).
20 PART 1. DESCRIPTIVE STATISTICS

22. (a) The proportion of times the prize has gone to a number finishing in 0,3,6 or 9 is

20 + 20 + 27 + 16 83
= = 0.407.
20 + 8 + · · · + 21 + 16 204

(b) The variable is quantitative (it takes numerical values) and discrete (it takes a finite
number of different variables). As such, the adequate graphical representations are
the pie chart and the bar chart.

(c) The frequency table for these data is:

xi ni fi Ni Fi
0 20 0.098 20 0.098
1 8 0.039 28 0.137
2 13 0.064 41 0.201
3 20 0.098 61 0.299
4 27 0.132 88 0.431
5 32 0.157 120 0.588
6 27 0.132 147 0.720
7 20 0.098 167 0.818
8 21 0.103 188 0.921
9 16 0.078 204 1

(d) From the column of the cumulative relative frequencies we deduce that the median
is equal to 5.

23. In a previous exercise we have determined the type of each of these variables, which
determined the kind of measures we may use.

DESCRIPTIVE Oven Lifetime Number of Exam

MEASURE brand breakdowns mark
Mode X X X
Median X X X
Mean X X
Third quartile X X X
Standard deviation X X

Note that since the exam mark is qualitative and ordinal, we may not always be able to
compute the quantiles (only when the cumulative relative frequency of the quantile is not
attained by any category).
1.4. TEST SOLUTIONS 21

1.4. Test solutions

1. b) 6. a), b), c) and d) 11. a)

2. b) and c) 7. c) 12. c)

3. b) and c) 8. a) and c) 13. b)

4. b) and d) 9. b) and c) 14. d)

5. d) 10. a) 15. a)