TIME SERIES ANALYSIS ON RECORDED CASES OF DRUG ABUSE AMONG THE

YOUTHS IN KEBBI STATE

Ladan S.1*, Usman A.2, Ibrahim H. W.3 Barade B.4

1,2,3
Department of Statistics, Waziri Umar Federal Polytechnic, Birnin Kebbi, Nigeria.
4
Department of Industrial Education, Waziri Umar Federal Polytechnic, Birnin Kebbi, Nigeria.
*[email protected], [email protected],
Tel: +234(0)8062352452, +234(0)8037586145

ABSTRACT: This research work is aimed to develop an Autoregressive integrated moving average
(ARIMA) model for the monthly recorded cases of drug abuse among the youths in Kebbi State. The data
is for the period of (72) months only from January, 2015 to December, 2020 were used. The empirical
study reveals that the best model for the recorded cases of drug abuse among the youths in Kebbi State for
the period under study is ARIMA (2,1,1). Using the model we developed a forecast for the period of 36
months i.e January, 2021 to December, 2023. The result revealed a downward movement throughout the
forecast period, with about 2 cases to be recorded by December, 2023.

Key words: ARIMA, PACF, Drug Abuse, ACF, KPSS.

1.1 INRODUCTION

The concept of drug has been defined as any chemical substance which affects living organisms. And also
such substance may be used to fight infection and illness or minimize pain, fatigue anxiety or at times to
achieve certain level of euphoria. On the other hand “Drug Abuse” is defined as any use of drug for non-
medical purpose also always for altering consciousness. Drug abuse denotes substance that change the
mental or physical state of a person and that may be used repeatedly for the effect leading to abnormally.

Drug Abuse has reach a certain limit that the youth always had a desire to eat or drink substance (Drugs)
that make them feel relaxed or stimulate. The discovery of fomentation and forming since 6000BC is
when people started to use drugs. The first drug to be abused was homemade alcohol wine and the level
could read up to 14-16%. Drug abuse started to be high through a various experimentation and people
diverted the use of drugs and also used for money generating.

1
The Samarian people of Asia, minor used joy plant which is believed to be opium poppy in 500BC, china
used ephedrine inhalant in 3000BC. Cannabis was introduced in India in 2000BC and American were
used to chew coco leaves, thus, by the end of sixteenth century drugs like cocaine, tobacco, cannabis,
hallucinogen and nicotine ware used globally even in Africa whereby Egypt and Tanganyika were both
noted.

Cannabis, marijuana and hashish are the most widely abused drugs in the world. Around 141 million
people consume cannabis. The use of stimulates such as amphetamine and ecstasy is also widespread,
with nearly 30 million people abusing these drugs cocaine is used by around 13 million people across the
globe, with the highest number of users in the United states. Abuse of heroin and other opioids is less
common than with other drugs and is taken up by around 8 million worldwide mainly in south-east and
south-west Asia and Europe.

In addition, on looking of the problem of Drug Abuse, locally and globally and measures which have
been taking to solve the problem, there are some well-known risk, nevertheless children continue to abuse
Drugs and such Abuses continue to tear down their lives. Therefor there is a need of “Modernity” ways of
solving rather than relying on immediate short run solutions.

It need investigation by visiting the individual Drug users, the groups and communities of large and
interviews them. Thus, they can give out their experience as response to their environments and come out
with social diagnosis which will help in provision of solutions or treatments according to the information
given from clientele system. By doing so social workers will manage to solve the problem of Drug Abuse
among the youth by using African theory which has been groomed from their own locality hence they can
meet the local needs of the people.

Drug abuse among the youth has become a serious problem affecting everyone. Addiction lead many
people, young people prominent among then, into downward spiral of hopelessness that in some case end
fetal. Drugs is responsible for lost wage, distraction of properly in school, soaring health care cost and
broken families, t is a problem which affect us as parent’s children, teachers, government officials.

Although excessive use of illicit drugs is prevalent in all societies, students with exception, none of
critical studies on factors influencing its prevalence comparing in school to school, youth has yet been
conducted.

On looking at the above theorem and knowledge, there is a strong need of solving the problem of drug
abuse among the youth by going direct to the users that are individuals, group communities and society at
large, interviewing them and develop theory and knowledge basing from the results of interviews rather

2
than relying on knowledge borrowed from western countries which does not meet the local need of
indigenous populations.

This study is aimed of investigating the courses and effect of substance abuse among the youths with a
view to achieve the following objectives. To fit the ARIMA model to the data, to determine the best fitted
ARIMA model and to use the model developed to forecast the case of drug abuse for the next 4 years

2.1 LITERATURE REVIEW

A study by Oteno (2009) found that 58% of middle school students in Muyu District had drunk alcohol at
some point in their lives. The study interviewed nearly 458 students from nine middle schools in Kisumi
and concluded that drug use, including alcohol, tobacco, marijuana, and cocaine, has risen sharply in the
past decade. The results of the study showed that age, maleness, high-risk behaviors, whether there are
smokers in family or friends, drug abuse experience, smoking tendency, and positive smoking thoughts
are related to adolescent smoking. The study also confirmed that men and women have almost equal
chances of becoming smokers (11.2%); however, the prevalence of regular alcohol consumption in men
(22.4%) is slightly higher than that of women (19.3%).

Ngesu, (2008) explains that young people are abusing drugs for a variety of reasons. One of them was
that drugs were easily available in the neighborhood of the school. This will allow students to go buy
medicines and use them on school grounds and in mythology, helping them to study for long periods of
time without considering the effects. Drug availability and cost were associated with substance abuse.
Age factors also contribute to substance abuse. This is because adolescents want young people to have
their own identity and discover about themselves and their growth. Young people are also curious about
young people trying beer to abuse or even reduce their effectiveness, for example, when they see a drunk
person staggering on the road and feel the same effect. Was born (Walter, 2011).

In another study, investigating family and use-related issues, Darcis (2012) conducted a rapid situational
analysis study in Nigeria, and that being a man in an unstable family was a high risk of substance abuse. I
found it to be related. There was support for discussions from clinical findings in Nigeria, indicating that
cannabis abuse tends to be young men, including students who were deprived of parental supervision and
warmth when they were young (Obot, 2010). ).

According to an international study by Doll, (2004), half of long-term smokers will give birth
prematurely. Half of them are middle-aged. The study concluded that smokers are four times more likely
to develop heart disease before the age of 40 than non-smokers (Mahonen, 2004). In addition, the earlier

3
young people start smoking, the more they smoke in their lives, and the more likely they are to develop
smoking-related diseases. Sometimes drugs replace satisfying relationships, educational achievement, or
self-realization, making college and high school students often forget why they are in school.

According to the Ministerial council on Drugs strategy (2005), drugs abuse, including smoking and
drinking alcohol, imposes substantial cost on uses and their families, taxpayers, on the national economy
and the commonly as a whole. An initial search into the data bases yielded 11 articles, two of which were
related to years before the study time frame (1997 and 1998). In addition, two articles were ignored, one
because of their different age groups (lower age), and the other because it targeted a specific area of
Tehran with a small sample size. These results are based on 7 articles. All studies are on people aged 14-
19 years, and only three studies distinguish between genders. All the seven studies considered in this
article are cross-sections of the adverse effects of drug abuse among adolescents and college students on
academic performance, which is evidence in national examinations, especially in this research field. The
student was found to be in possession of drugs during the school season and therefore appealed to
education stakeholders to take immediate action (Montazi, 2010).

A study by et al. (2002) showed that drug abuse can lead to antisocial peer groups, which can reduce
school participation and increase other behaviors and social problems.

Horwood, (2010) also pointed out that drug abuse is related to other social factors, which are related to
campus, attendance, and completion or dropout. It may not be until the 1990s that investigations on drugs
and drug abuse attracted much attention in Kenya.

Substance (drug) abuse can seriously affect academic performance, and part of long-term addiction can
lead to decreased performance (Montazi, 2010). Substance use affects your entire body, including the
brain.

3.1 MATERIALS AND METHOD

Data use in this research were monthly figures cases of drug abuse among the youth covering the period
from 2015-2020, the analysis was performed using Gretel. The data were obtained from National Drug
and Law Enforcement Agency (N.D.L.E.A) Kebbi State from January, 2015 to December, 2020.

However, for the model we used, there is need to test the heteroskedasticity and more importantly the
covariance stationary of each of the series involved in order to obtain valid and accurate results. These
test and all others that will be implemented on the course of the research are introduced and explain as
follows:

4
3.2 AUGMENTED DICKEY FULLER TEST (ADF)

Augmented Dickey Fuller Test (ADF Test) is a common statistical test used to test whether a given time
series data is stationary or not. It is also most commonly used statistical test when it comes to analyzing
the stationary of a series. We check stationary by two ways. One is a manually check of mean and
variance of time series and another way is a using ADF test function. As we know that stationary time
series don’t have change mean or variance over time.

The testing procedure for the ADF test is the same as the Dickey Fuller test but it is applied to the model.

Δyt = α + βt + ϒyt -1 + σ1 Δyt-1

Where α is a constant, β the coefficient on a time trend and Ῥ the lag order of the autoregressive process.
Imposing the constraints α = 0 and β = 0 corresponds to modeling a random walk and using the constraint
β = 0 corresponds to modeling a random walk with a drift.

For the ADF the hypothesis is:

Ho: the series contain a unit root and H1: the series is stationary

3.3 KPSS TEST

Kwiatkowski, Phillips, Schmidt and shin (1992) proposed a test of the null hypothesis that an observable
series is trend stationary (stationary around a deterministic trend). The integration properties of a series
yt may also be investigated by testing the null hypothesis that the series is stationary against a unit root.

Assuming no linear trend term, the data generating process is given as:-

yt  xt  zt Where xt a random is walk, xt  xt 1  vt , vt iid (0,  v 2 ) and is a stationary process.

1 r
St 2 t
  St   w j
Kwiatkowski (1992) proposed the following test statistic KPSS
T2 t 1 
2
Where j 1
With

 T 
 zt ,  2  lim T 1 var   Z t 
w j  yt  y  2 T 
 t 1  .
and an estimator of the long run variance of

The null hypothesis of the test is

H 0 :  v 2  o against the alternative hypothesis H1 :  v 2  0 . Reject the

null hypothesis if the test statistic is greater than the asymptotic critical values.

The theoretical model, which also serve as basic frame work of our statistical analysis, is Autoregressive
integrated moving average (ARIMA)

5
3.4 ARIMA MODEL

Autoregressive integrated moving average (ARIMA) model in statistics and econometrics, and in
particular in time series analysis, an Autoregressive Integrated Moving Average (ARIMA) model is a
generalization of an Autoregressive Integrated Moving Average (ARIMA) model. In theory, the most
general class of models for forecasting a time series which can be stationary by transformations such as
differencing and logging. ARIMA model form an important part of the Box- Jenkins approach to time
series modeling.

Knowing that stationary time series is integrated of the order d and if by differencing the terms it becomes
an ARIMA (p,d,q) process, then the difference process can have an ARIMA (p,d,q) representation. In this
{X t }
case the time series can be expressed as

 ( L)  d X t   ( L)et Where
 d  (1  L) d
A non- seasonal ARIMA model is classified as an ARIMA (p,d,q) model with:

P is the number of autoregressive terms, d is the number of non- seasonal differences

q is the number of moving average lags

(Yt) is said to be ARIMA (p,d,q) if (1- L)d ϕ.( L) Yt = C + ϴ( L)Ɛt

Where ϕ (L) is defined as in ϕ (L) = (1-L) ϕ (L), ϕ (Z) ≠ 0 for all ̸̸̸Z̸ ≤ 1.and ϴ( L) is defined as ϴ( Z) ≠ 0
for ̸̸̸Z̸ ≤ 1.

3.4.1 AUTO REGRESSIVE PROCESS AR (p)

The auto regressive process uses weighted time lagged (previous) values to generate new current values

for the time series. A time series {

X t } is an AR (p) process if it has the following representation.

p
Xn  m  en   k X n  k
i

k 1

 mi  en  1 X n 1  2 X n  2  ...........   p X n  p
e
For n  0, where { n } n  0, is a series of independent identically distributed (iid) random variables,

i
and m is some constant.

6
3.4.2 MOVING AVERAGE PROCESS, MA (q)

The moving average technique is often used for linear fitting. A moving average process of order q

denoted by (MA)q is a stationary time series process {

X t } by box and Jenkins (1976), if it has

representation of the form

X n  m  en  1en 1   2en  2  .....   q en  q

q
 m  en    k en  k
k 1
4.1 ANALYSIS AND RESULTS

The time series data collected for this study are recorded cases of Drug Abuse among the youths in Kebbi
State from January, 2015 to December, 2020 consisting of 72 monthly observation . These were used to
build a suitable ARIMA (p,d,q) model to forecast the feature cases of recorded cases of Drug Abuse over
a period of 36 month.

4.2.1 STATIONARY TEST RESULT

The process will start with testing for stationarity using the time plot diagram and also performing unit-
root test (ADF and KPSS). The graphical representation of the series against time in it levels (Fig. 4.1)
indicate that the series is experiencing an upward and downward movement through the series which
means that the series is not stationary because the mean and variance change with time.

7
 10

8
DrugAbuse

0
2015 2016 2017 2018 2019 2020 2021

Figure 1 a time series plot display of the reported cases of Drug Abuse among the youths in Kebbi State.

Regression residuals (= observed - fitted DrugAbuse)

5

2
residual

-1

-2

-3

-4
2015 2016 2017 2018 2019 2020 2021

Figure 2 a time series plot display of the reported cases of Drug Abuse among the youths in Kebbi State at first
difference.

8
 ACF for DrugAbuse

0.6 +- 1.96/T^0.5

0.4

0.2

-0.2

-0.4

-0.6

0 5 10 15 20 25 30
lag

PACF for DrugAbuse

0.6 +- 1.96/T^0.5

0.4

0.2

-0.2

-0.4

-0.6

0 5 10 15 20 25 30
lag

Figure 3 ACF and PACF for the reported Drug Abuse cases at levels

From ACF it can be deduced that the series is highly positively auto correlated with highly persistent
correlation

Looking at the PACF on the other hand, it can be observed clearly that the model for the data is mixed
one. This is to say that the model for the data is mixed with ARIMA component.

ACF for d_DrugAbuse

0.3 +- 1.96/T^0.5
0.2

0.1

-0.1

-0.2

-0.3
0 5 10 15 20
lag

PACF for d_DrugAbuse

0.3 +- 1.96/T^0.5
0.2

0.1

-0.1

-0.2

-0.3
0 5 10 15 20
lag

Figure 4: ACF and PACF for the reported Drug Abuse cases at first difference

The plot for the ADF and PACF above shows that the series is stationary at first difference

9
4.2.2UNIT ROOT TEST

Due to the stationarity of the data which we observed from the time series plot of ADF and KPSS at first
difference above we also apply the Unit-root test which is shown in table 4.2. Evidently from figure 4.1,
the variance and mean looks stationary but this is not enough to test for the stationarity with respect to the
mean and variance.

A confirming test on stationarity is followed through using the Augmented Dickey Fuller (ADF) and
KPSS Unit-root test at levels as shown in table 4.2

TABLE 1 Result of Unit Root Test using ADF and KPSS

UNIT ROOT TEST COMPUTED VALUE CRITICAL VALUE AT 5%

ADF -4.27241 -0.415808
KPSS 0.233631 0.463

TABLE 2: Test for unit roots for the first difference

UNIT ROOT TEST COMPUTED VALUE CRITICAL VALUE AT 5%

ADF -6.7732 -9.34814
KPSS 0.193071 0.463

4.3 MODEL FITTING

In order to identify ARIMA Model for recorded cases of drug abuse among the youths in Kebbi State the
steps is describe by Box and Jenkins have been followed

For ARIMA Models with tentatively selected various values of p,d and q are estimated by using Gretel
Software the models are; ARIMA (2,1,1), ARIMA (2,2,1), ARIMA (2,2,2), ARIMA (3,2,1)

TABLE 3 VALIDATION AND DIAGONESTIC CRITERIA FOR THE MODEL

MODEL AKAIKE (AIC) SCHWARZ (SC) HANNAN- QUINN

ARIMA (2,1,1) 302.9511 314.1936 307.4168

ARIMA (2,2,1) 333.4884 344.6589 337.9201

ARIMA (2,2,2) 311.8754 325.2801 317.1935
ARIMA (3,2,1) 330.2706 343.6753 335.5887

10
From the above table the ARIMA with p=2, d=1 and q=1 has the lowest value of all the selected criteria
AKAIKE (AIC), SCHWARZ (SC) And HANNAN –QIUNN. Hence ARMA (1,0,1) has been selected as
the best model to forecast the cases of Drug Abuse among the youth in Kebbi State.

TABLE 4 PARAMETER OF ESTEMATE OF ARIMA (2, 1, 1)

Variable Coefficient Std. Error t- ratio p-value

Const -0.00172219 0.00793495 -0.2170 0.8282
phi_1 -0.198740 0.120499 -1.649 0.0991 *
phi_2 -0.208026 0.120013 -1.733 0.0830 *
theta_1 -1.00000 0.0380756 -26.26 5.00e-152 ***
From the above table, ARIMA (2, 1, 1) model looks the most appropriate model.

Estimated equation for ARIMA (2, 1, 1) model

Yt= a + b1X t-1 + b2X t-2 + d1U1 + et

Yt = -0.00172219-0.198740 yt-1 -0.208026 yt-2 -1.00000yt-1 et

Analysis of residual

The analysis of ARIMA residual constitutes an important test of the model. The estimation procedure
assumes that the residual are not auto correlated and normally distributed. The figure below show the
analysis of the residual:

ACF for d_DrugAbuse

0.3 +- 1.96/T^0.5
0.2

0.1

-0.1

-0.2

-0.3
0 5 10 15 20
lag

PACF for d_DrugAbuse

0.3 +- 1.96/T^0.5
0.2

0.1

-0.1

-0.2

-0.3
0 5 10 15 20
lag

Figure 5 Residual ACF and PACF for reported cases of Drug Abuse

11
To be certain that there is no information in the selected model that is left outside. The residual
correlogram of the ARIMA (2, 1, 1) is plotted. An ideal residual correlogram should be flat, meaning all
the lags structure should lie within 95% confidence interval. Figure 4.3 above has shown that no
information is left behind in the 20 lags. So the forecast would be on this model.

4.5 FORECAST

The essence of fitting an ARIMA model is to forecast future event of a particular series. In other words,
we are only using past values of the series to explain its future occurrence. However, forecast is based on
the final selected model. When the model is has been identified and forecasted, then the forecast value
and the graph would be shown and its corresponding values.

It could be observed from the forecast results above that the recorded cases of Drug Abuse among the
youth in Kebbi State for the period of three (3) years starting from January, 2021 to December, 2023. It
would be experiencing a downward movement throughout the forecast period, with about 2 cases to be
recorded by December, 2023.

10
DrugAbuse
forecast
95 percent interval

-2
2018 2019 2020 2021 2022 2023 2024

Figure 4.6: Comparison of actual and forecast values of reported cases of Drug Abuse among the
youth in Kebbi State.

12
4.6 CONCLUSION

The forecast from January, 2021 to December, 2023 predict experience an decreases for new cases of
Drug Abuse among the youths in Kebbi State. The fact that the decrease was slights does not mean the
stakeholders, students and also government should relax on the result. But rather, the National Drug Law
and Enforcement Agency (NDLEA) and State Ministry of Health (SMoH) should come together to
provide intensive education to people on the effect and the risk factors of Drug Abuse.

4.7 RECOMMENDATIONS

1. A lot of programs should be organized by NDLEA and Health workers so as to create awareness on the
effect of taking illegal drugs. So that, they can start treatment early before it results into complications
such as heart disease and also lungs problems. Therefore such programs must throw more lights on how
drug abuse can lead to other dangerous conditions.

2. The health facilities should advice people to eat good foods and fruits and they should avoid taking
cigarette and many harmful things in other to keep their self-healthy.

3. I also recommend that any further studies on this topic and its related compilations using multivariate
approach should be conducted to observe the relation between the variables.

REFERENCE

Oteno (2009), found that 58% of the secondary school students in Kisumi District had consumed alcohol
at some point in their lives.

World Drug Report in (2006), the use of illicit drug has increased throughout the world in recent years.

World Health Organisation (WHO), tobacco causes over 4 million deaths annually, not including prenatal
morbidity and mortality.

World Health Organisation maid, (INCB, 2007).This figure is projected to rise to 1.4 million by the year
2025, with 70 % of what will occur in the developed countries if current trends continue according
to the research.

Masita (2004) a cigarette smoker among males and females is almost equal (11.2%); however, the
prevalence of regular alcohol consumption in males (22.4%) is slightly higher than in females
(19.3%).

13
Ngesu, (2008) explain that young people abuse Drugs for various reason.One of them was that Drugs
were easily available in the neighborhoods of the school.

Walter (2011).Young people also abuse Drugs or rather to fell the effect, for instance seeing a drunker
person staggering on the road created curiosity to the youth to go and try beer so as to feel the same
effect (Walter, 2011).

Darcis (2012), carried out a rapid situations Analysis study in Nigeria and found that being male in an
unstable family was associated with high risk for substance (Drug) abuse.

Obot(2010).There was support to the argument from clinical findings in Nigeria, which showed that
cannabis abuse tend to be young men, including students, who had deprived of parental supervision
and warmth when they were young.

Mahonen(2004). In addition, the earlier young people start smoking, the more they smoke over their
lifetime, and the more likely they are to suffer from smoking related diseases.

According to the Ministerial council on Drugs strategy (2005), drugs abuse, including smoking and
drinking alcohol, imposes substantial cost on uses and their families, taxpayers, on the national
economy and the commonly as a whole.

World Health Organisation (WHO, 2010), tobacco causes over 4 million deaths annually, not including
prenatal morbidity and mortality.

WHO (2008) Drug abuse among teens and college students have a detrimental effect on academic
performance which is evidence in the national examination especially in the area of this study.

14