Factors Affecting People’s Environmental Awareness In The Urban Areas: A Case of Addis Ababa, Ethiopia


 Background: In developing countries, the urban environment is deteriorating over time. In the meantime, people's demand for clean and green residential and recreational places has increased. If so, why has it been hard to keep clean and green cities? This paper investigates the level and determinants of environmental awareness in Addis Ababa. From three sub-cities, a three-stage sampling procedure applies to select 293 sample respondents. The data collection applies a structured questionnaire. We applied a five-point Likert scale to classify the levels of awareness. Besides, an ordered logit model was applied to analyze factors that affect the level of awareness.Results: The result shows that the knowledge of forest degradation is high, while the attitude to reduce the use of glass bottles is medium. The cognitive skill on the cause of acid rain is medium. The probability of low (13%) and medium (26%) levels of environmental awareness increases for the income group of 601 to 1650. Likewise, the likelihood of having low levels of environmental awareness rises by 9%; in contrast, the probability of having a moderate level of environmental awareness increases by 12% for the age of 50 to 59. The TVET educational level has a low chance of having low (8%) and medium (12%) levels of awareness.Conclusions: An income-generating activity raises employment opportunities and creates a better income, which would influence the environmental mindset. So, improving the living standard assures clean and green cities. Besides, the higher the education, the better would be environmental knowledge, cognitive skills, and attitude. In the meantime, besides formal education, adult education, training, and workshops are alternatives to enhance environmental awareness.


INTRODUCTION
Environmental quality links to the use of resources and waste disposal. It relates to the effluent systems and waste management (Hoornweget al., 2011), greenhouse gas and particulate matter (Shanmugam and Hertelendy, 2011), and poor infrastructure, and meager urban planning (Colombani et al., 2018;Liu et al., 2015). In developing countries, lack of standard inbuilt sewerage systems, poor waste management (Gondo et al., 2010), and volatile gases (Kaushal and Sharma, 2016;Kumar et al., 2016) provoke environmental pollution. Environmental pollution links people's environmental awareness (ECLAC, 2004;Momoh and Oladebeye, 2010) and their consumption behavior (Xu et al., 2019). Hence, working on people's awareness gives strength to manage the environment (Giudici et al., 2019). Awareness also links to education (Mutisya and Barker, 2011), residential places (Bickersta and Walker, 2001), and technological knowledge (Giudici et al., 2019).
Rivers and groundwater deterioration (Ademe and Molla, 2014;Eriksson and Sigvant, 2019) and air pollution (UN Environment, 2018) are Ethiopia's causes of environmental degradation. Over the last 30 years, the urban environment impaired following population expansion, industrialization, and urbanization (Akalu et al., 2011;Eriksson and Sigvant, 2019;Worku and Giweta, 2018). The Ethiopian Environmental Protection Authority (1997) emphasized improving, sustaining, and keeping the environment. Besides, the Climate Resilient Green Economic strategy gives due attention to green cities, landfill gas, and wastes (FDRE, 2011). The transport policy also emphasizes wastes associated with the transport system (Ministry of Transport, 2011).
Even though materials on environmental awareness (MoFED, 2006) have been producing, the environment in Ethiopia faces multi-dimensional problems (Danyo et al., 4 2017). Therefore, examining factors affecting people's environmental awareness is incontestable. Despite the importance of the topic, empirical studies hardly examined environmental awareness in Addis Ababa. Existing studies focused on the human impact, urban rivers, watershed land use, surface water pollution, and flood vulnerability (Akalu et al., 2011;Asnake et al., 2021;Eriksson and Sigvant, 2019;Mohamed and Worku, 2020). Several studies conducted on solid waste (Beyene and Banerjee, 2011;Destaw et al., 2013;T. Getahun et al., 2012;Regassa et al., 2011), river and groundwater contamination (Awoke et al., 2016;Gebre and Rooijen, 2009;Gondo et al., 2010;Goshu et al., 2010;Mazhindu et al., 2010), and air pollution (Do et al., 2013). While, few studies examining the environmental awareness in the farming communities (Adem, 2017) and among students (Emiru and Waktola, 2018). The paper's organization follows the introduction, material, and method section, explain the analytical framework, study area, data collection tool, model specification, and variable characteristics. The result section explains the demographic characteristics, level of environmental awareness, and factors affecting awareness. The discussion section elaborates the key findings concerning the existing knowledge. Last, the conclusion section summarizes the main findings and forward recommendations. Figure 1 shows the study procedure from questionnaire development up to defining components of environmental awareness, from the list of pre-defined environmental items, identifying and screening out the reliability by Cronbach's alpha. The framework depicts environmental awareness as the components of knowledge, cognitive skill, and attitude.

5
Environmental awareness helps to assess people's consciousness in their activity (Partanen-Hertell et al., 1999). Awareness is being conscious of something. According to Rohrer (2002), it is the sum of all abilities which permit humans to respect fundamental rights. Hence, a high level of awareness correlates with the conscious choice of environmentally friendly practices (Partanen-Hertell et al., 1999).
The components of awareness such as knowledge, attitude, and cognitive skills are described differently by different scholars. Morreale et al. (2007) divide knowledge into content knowledge and procedural knowledge. Content knowledge is a literal understanding of the subject, words, or meanings. Procedural knowledge emphasizes practicing the content knowledge. So, knowledge is a process that developed and constantly grew (Watson and West, 2006). However, Williams (2002) expressed it as a combination of belief and fact.
Attitude is the association between an object and the evaluation of that object (Fazio, 1990). Motivation is the determining component of behavior and attitude (Fazio, 1990). At the same time, attitude and behavior have some relationship (Borba, 2004;Fazio, 1990). The perception of potential reward value determines motivation (Morreale et al., 2007). Similarly, individuals can build motivation (Partanen-Hertell et al., 1999), yet attitude enables them to decide (Crano and Prislin, 2008;Sanbonmatsu and Fazio, 1990). So, there is a correlation between attitude and behavior (Cialdini et al., 1981), yet they are too different.
Hence, Borba (2004)  6 Skills are also associated with the habit of using something (Partanen-Hertell et al., 1999). A person may have the motivation to act, due to lack of acting skill low performance might happen (Morreale et al., 2007). Thus, skills express the competency, ability, aptitude, capacity, and habit of doing something. Stopford (2009) expresses the distinction between skill and knowledge as the border between academic and professional. While Ingold (2000) looks at skills and knowledge together, the skill is the effective use of knowledge. female. The annual fertility rate was 2.1 (CSA, 2013). The organization of the city is by ten sub-cities and 118 districts (Abebe et al., 2018).

Data Collection and Questionnaire Design
Enumerators did data collection using a structured questionnaire from respondents in three sub-cities. Non-probability and probability sampling methods were adopted to collect primary data. We follow a three-stage sampling procedure to sample size selection. In the first stage, the researchers categorize ten sub-cities into three strata based on their population density. The official document shows that six sub-cities, such as Bole, Yeka, Gulele, Kolfe-Keraniyo, Nifas Silk, and Akaki-Kalit, had a population density lower than or equal to ten thousand.

7
In contrast, the Addis-Ketema sub-city alone has over twenty-five thousand. The remaining sub-cities, such as Arada, Lideta, and Kirkos, are between the two groups.
So, one sub-city was randomly selected from each stratum as a sample frame. In the second stage, three districts were selected from the three sub-cities by simple random sampling. Thus, 293 household members were designated as a proportional sample using the statistical formula developed by Yamane (1967), as Israel (1992) cited. The proportional samples are districts 3 from Addis Ketema sub-city; sample size equal to 88, districts 6 from Bole sub-city; sample size equal to 100, and districts 8 from Arada sub-city; sample size equal to 105. So, from the list of household heads (HH), one respondent was selected using the lottery method from the list of the first ten HH. Then, we used an interval technique to select respondents until the proportional sample meets.
A questionnaire was developed to investigate peoples' environmental awareness.
The tool contains questions on socio-economic, demographic, and environmental questions. Question on environmental knowledge addresses pollution, its cause, source, and the effect of solid, liquid, and gaseous wastes. The cognitive skills questions focused on the effect of consuming goods and services on the environment, for instance, using different types of energy sources, tree planting and cuttings, agricultural practice, livestock rearing, and solid and liquid waste disposals. Questions on environmental attitude also address practical actions by the respondents to reuse, recycling, reducing, and recovering of goods and services. Therefore, they responded on a five scale, '1= very low to 5 = very high'. Thus the questionnaire was translated from English to the Amharic language, and enumerators are recruited and trained. Then, data were collected on a face-to-face basis for three weeks, starting from the end of May 2019 to mid-June 2019. 8

Model Specification and data analysis
Descriptive and inferential statistics are used to analyze the data. Five Likert scales are prepared to see the level of environmental awareness through the environmental items under the three components: environmental knowledge, cognitive skills, and attitude.
Environmental things are expressed in negative and positive statements to avoid bias.
The response scale for each environmental item is1 to 5.Thus, the response to the negative report has a reverse value. The sum of the scale is represented the full scale.
The maximum total scale is 5*n, and the lowest possible scale is 1*n. Where 'n' is the total number of environmental items listed under the three components, each respondent's level of environmental awareness is computed by Eq. 1.
An econometric model is also used to examine factors affecting people's environmental awareness. Here, environmental awareness is a categorical dependent variable ordered as very high, high, medium, low, and very low. Although an unordered multinomial model can estimate such data, a much more economical and sensible model considers this ordering. Thus, the choice of the ordinal probit model fits more critically than the multinomial model to address the level of environmental awareness (Gujarati, 2004). Therefore, the starting point is an index model with a single latent variable, y* (Eq. 2).
Y is collapsing a version of y*, e.g., y* can take an infinite range of values which might be five orders of Y. As y* crosses a series of increasing unknown thresholds (Cut, αi), we move up the ordering of alternatives. For example, for y*< α1, awareness is very low, for y* > α1, awareness improved to the highest level. So, the observed variable 'Y' value depends on whether it crossed a particular threshold. Since there are five potential values for Y (Cameron and Trivedi, 2005;Greene, 2003), the respondent awareness probability is in one of the fifth levels (Eq. 4).
Where m is the level of awareness, α is a particular threshold (4 cuts) in which the value of the observed variable Y, Xi is an explanatory variable that affects the level of awareness, and β is the unknown estimated parameter. Therefore, factors affecting people's environmental awareness are analyzed by the Ordinal probit model expressed as Eq. 5 using STATA software version 15.
Where y* is the unmeasured latent variable whose values figure the observed ordinal environmental awareness, EA, Xi is an explanatory variable such as income group (I), family member (F), educational level (E), age (A), and sex (S). So, the probability of environmental awareness being in one of the five levels is computed as in Eq. 7-11. Higher education (First degree and above). The respondent's age is a continuous, categorical variable that is arranged into six groups (17-29, 30-39, 40-49, 50-59, 60-69, and 70-100). Gender is the biological classification of the respondent's sex. It is a dummy variable that is assigned 1 if the respondent is male, otherwise 0 for female respondents.

RESULTS
This section describes the environmental awareness components, socio-demographic characteristics, and order logit model results. On average, the respondents have medium to high conceptual knowledge, high knowledge of the causes and effects, and medium knowledge of the source of pollution (Table 1). They also have higher knowledge of forest degradation as compared with surface and air pollution. In contrast, their knowledge of air pollution is the least of other types of environmental knowledge.
The questions on cognitive skills focus on energy sources, deforestation, planting trees, and Green House Gases. The mean of these measures ranges from medium to high, which has a high-reliability index. On average, the respondents have high cognitive skills on the negative contribution of waste disposal and deforestation. In contrast, they have medium cognitive skills on the cause of acid rain (Table 2).
Attitude questions focus on reducing glass bottles, plastic bottles, cans, and fossil fuels. As a result, the mean value ranges from medium to high attitudes with a high-reliability index. On average, the respondents show a high attitude towards reducing the consumption of cylinder gas, while they have a medium attitude towards reducing the use of glass bottles (Table 3).

Respondents' Characteristics and Environmental Awareness
The descriptive result in Table 4 shows the variation in the level of environmental awareness. There are variations among the income groups, family size, educational level, age groups, gender, and districts. The Chi-square value shows awareness varies significantly among income groups, education, age, and the gender of respondents.
Nevertheless, the levels of environmental awareness do not show substantial variation within the family member and among districts.
The level of environmental awareness differs across income groups at p < 0.01.
In all income groups' high level of environmental awareness is the dominant, except the income group 601-1650. It is high and very high for 91% of respondents in the highest income group, while the remaining 8.7% have a medium level of awareness.
Environmental awareness varies among educational groups at p <0.001. Most TVET 13 (55.6%) and higher education (29.6) score a high and very high level of awareness, respectively.
In contrast, the secondary academic level has a medium level of awareness (33%) compared to others. The primary education level has very low (1.9%) and low (14%) environmental awareness. It suggests that as the completed educational level increases, the level of environmental awareness shows improves.
The levels of environmental awareness also vary within the respondents' age group at p < 0.1. The level of awareness is highest with the age group of 40-49, while it is the lowest for 17-29, 50-59, and 70-100 years old. Gender variation also shows a difference in the level of environmental awareness. Most male (50%) and female (43%) respondents have a high level of environmental awareness, while 24.7% of males and 35% of females have medium awareness. The number of male respondents with a high and very high level of environmental awareness is more than female respondents. The income groups from 601 up to 5250 significantly affect environmental awareness at P<0.01. Those respondents with TVET and first degree and education levels positively affect it at P < 0.01 and P < 0.05, respectively. Similarly, being above 49 years old has lower environmental awareness than being below 30 years old.

Factors that affect Environmental Awareness
Likewise, age between 60 to 69 and 70 to 100 years old negatively affects their environmental awareness level at the P<0.05 level of significance. The marginal effect of TVET education shows that, the odds of respondents being in low and medium levels of environmental awareness decreased by 8% and 12%.
However, the chance of being in high and very high levels of environmental awareness increase by 9% and 12%, respectively. Similarly, the probability of low and medium levels decline by 7% and 10% for a minimum of first degree completed. The odds of high and very high environmental awareness are likely to increase by 8% and 9%, respectively.
The marginal effect proves that age determines the level of environmental awareness. Being 50-59 years old, the corresponding probability of low and medium levels of environmental awareness increases by 9% and 12%. In contrast, the chance of high and very high levels of environmental awareness decline by 8% and 13%, respectively. The odds of the medium and very high environmental awareness increase by 12% and decrease by 13%, respectively, for the age between 60-69 years old. The chance of being in the low and medium level of environmental awareness likely increase by 15% and 16%, while the probability of being in high and a very high level of environmental awareness decrease for the age of 70 to 100 years old.

DISCUSSIONS
There is high knowledge about river deterioration, air pollution, and forest degradation in Addis Ababa. River pollution is common in most developing countries (Capps, Bentsen, & Ramírez, 2016). Poor sewer and inadequate infrastructure could aggravate the river and stream pollution (Colombani et al., 2018;Liu et al., 2015). Similarly, the quality of air and tree cover reduces following the expansion of industries (Ejaz et al., 2010;Li and Lin, 2015), urbanization (Gasimli et al., 2019;Kleppel, 2002;Li and Lin, 2015), and the population (Li and Lin, 2015). Besides, the respondents have high cognitive skills on the negative contribution of waste disposal and deforestation to wildlife disturbance and soil erosion.
There are high cognitive skills on the effect of wastes on the environment. The cognitive skills on the influence of deforestation on wildlife and soil erosion are also high. Nevertheless, respondents have medium cognitive skills on the cause of acidic rain. There are high and medium attitudes to reduce the consumption of cylinder gas and glass bottles, respectively. It means environmental knowledge, cognitive skill, and attitude vary between respondents because of heterogeneity in their socio-economic status.
Descriptive and ordered logit result shows variation in the level of environmental awareness within the income groups (Duroy, 2005;Ito and Kawazoe, 2017;Strieder et al., 2017). This finding is in line with Xun et al. (2017), Strieder Philippsen et al. (2017) and Altin et al. (2014), yet against Üstün and Celep (2007). This means, the higher the income, the more access to knowledge, cognitive skills, and attitude change. Thus, higher-income led to a high level of environmentally friendly actions (Xu et al., 2019;Zhang et al., 2015) and is likely to push to demand a better residential environment (White et al., 2007).
Respondents between the age of 17-29 years old have a high and very high level of environmental awareness. The marginal effect also shows the chance of high and very high levels of environmental awareness decline for the age greater or equal to 50 years old. It is against Ziadat (2010), while in line with Aminrad et al. (2013) and Karytsas and Theodoropoulou (2014). Young peoples have better environmental awareness than the elderly. The reasons are as follows, first; they have had better access to information on the environmental damage in Addis Ababa for the last thirty years; second, they passed through the revised educational curriculum, which incorporates environmental items. Third, they are more popular with climate change and global warming in the last thirty years.
Education could influence the level of environmental awareness (Aminrad et al., 2011;, which is against Üstün and Celep (2007). Over secondary education, enhance the level of environmental awareness. Our finding agrees with Karytsas and Theodoropoulou (2014)

CONCLUSIONS
This article provides an insight into the measurement of environmental awareness through environmental knowledge, cognitive skills, and attitude. Besides, to investigate factors that affect environmental awareness, we used an ordered logit model. The questionnaire survey data was applied to conduct our study in Addis Ababa, Ethiopia. This article has some limitations. First, using a quantitative approach is one limitation. Future research may benefit from a mixed approach. Second, the study area is in Addis Ababa. Hence, to get a better image, it would be more pragmatic to include regional cities. It would be sound for future work to use an in-depth interview and ethnographic study.

Consent for publication
Authors agree to conditions of submission, Springer Open's copyright and license agreement and article-processing charge.

Availability of data and materials
Data would be available on request.

Competing interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding
Not applicable.
Abrham Seyoum: Supervision, conceptualization, Validation.     Table 6 Marginal fixed effect for the levels of environmental awareness Source: Own sketch by using ArcGIS 10.5 adopting shape-file from Google search (2020)