Study Location and population
This cross-sectional study was carried out in Benin City, which is the capital city of Edo State, located in the South-South geopolitical zone of Nigeria. The city is the residence of diverse ethnic groups, and it has a population of more than a million persons. Within the city are schools either owned by the government (public) or by individuals (private). The privately-owned schools obtain approval from the government to operate, and they far outnumber the public schools. The school system is such that the secondary school education is divided into junior and senior, and each is a three-year program. The study population consisted of students from both levels of secondary schools.
Ere the collection of data, approval was obtained from the research and ethics department, Edo State Ministry of Education, and the authorities of the schools involved. Written informed consent was obtained from the participants and their parents/teachers if below the age of 18 years. Only the students who could give consent or assent and understand the English language enough to fill the questionnaire were selected.
A total of 420 questionnaires were distributed even though a sample size of 195 was calculated using a Nigerian study in which 15.5% of children and adolescents had sleep problems . Sampling was based on convenience; four gender-mixed schools were picked from the numerous schools in the city. The selection was purposefully made to include two schools each from the public (government-owned) and privately owned schools. Students were thereafter randomly selected from both the junior and senior secondary levels from the chosen schools.
Two research assistants who were adequately trained were employed for the collection of data. The training entailed understanding the content of the questionnaire enough to be able to assist the students when necessary. The administration of the questionnaire was done in the classroom, mostly during the break period. Data were collected between January 22 and February 2, 2018.
A set of questionnaires which consisted of 5 sections was used for collecting data.
The first section, which is the sociodemographic characteristics, was used to collect information such as the gender, age, family structure, school type, religion, ethnicity, parental figure, and academic performance of the students. It was purposefully designed and structured to meet the objectives of the study.
The Sleep Quality Questionnaire (SQQ)  was used to assess sleep quality in the participants. It is a 9-item scale that is scored on a 5-point scale from 0 to 4 and added together. Possible scores range from 0 to 27; the lower the score, the more severe the sleep problem, while high scores indicate sleep to be in good shape. A score of 18 and below is highly suggestive of problems with sleep using SQQ, and this was adopted as the cut off in this study (Huffington, 2016). The scale has a strong correlation (r=0.925) with another sleep scale, the Sleep Condition Indicator, which has been validated and found to have good psychometric properties . The reliability score of the SQQ in the current study was 0.71.
The Internet Addiction Test  is a 20-item scale that assesses the general pattern of internet use and its impact on the respondent's life at work, school, home, and social. Responses are made on a 6-point Likert measure (0-5); thus, the possible score ranges between 0 and 100. A score of 50 and above indicates a potential internet addiction which was chosen in this study . It has been widely used among students, and the scale recorded a Cronbach alpha of 0.93 in the current study.
The Hospital Anxiety Depression Scale (HADS) was used to screen for the presence of anxiety and depression in the participants. HADS can be used in clinical and non-clinical populations, and it consists of seven depression and anxiety items. It has been validated for use in Nigeria . A higher score on each scale indicates increased psychological distress. A cut-off point of eight and above on either of the two components is regarded as a probable case diagnosis.
The World Health Organisation Questionnaire for Student Drug Use Surveys was developed from different parts of the world, including Nigeria . For this research, only the tobacco and caffeine sections of the instrument were used to obtain information on the commonly used stimulants. Respondents are required to indicate whether they have used them in the past thirty days (current use) and the frequency of use within this period. The instrument has been used in different cultures and countries, including Nigeria.
Data collected were analyzed using the Statistical Package for Social Sciences (SPSS), version 22. Uni-variate analysis, such as frequencies, was determined in percentages and displayed in tables. Some variables, such as religion, current stimulant use, were dichotomized to enable a sufficient number of participants for analysis. Cut off points were utilized to determine the prevalence of cases of depression, anxiety, problematic internet use, and PSQ. However, the continuous scores of some of these variables were employed in further bi-variates and multi-variates analyses.
The Chi-square tests were used to determine the association between sleep quality and categorical variables such as gender, family structure, school type, academic performance, current use of stimulants (caffeine and cigarette). The sleep quality scores were tested for normality using the one-sample Kolmogorov-Smirnov tests and were found to be non-normally distributed. Hence, the non-parametric Mann-Whitney U test was applied to examine the differences in the rank mean scores of age, depression, anxiety, and internet use between the normal and poor group of sleep quality. Variables such as current tobacco use, internet use, depression, and anxiety, which were significant on bi-variates analyses were included in the logistic regression model. A p-value of less than 0.05 was accepted as the level of statistical significance