You are on page 1of 7

Our study was focused on a particular area of Sierra Leone, close to its capital Freetown.

As evident from the map below, two districts were object of our
research: Western Area Urban and Western Area Rural. Within each district, two wards were randomly chosen: Grafton, Kossoh Town, Portee and Shell;
moreover, 30 households were randomly selected for the interviews, for 121 respondents in total.

As preliminary tool, we compute –via a principal component analysis (PCA) methodology – two scores that can approximate quite efficiently two variables of
interest in our research: the good practices in handwashing and the wellbeing of a household.
The first index is computed considering key variables concerning the handwashing practice, such as using soap, rinsing, clean handwashing device. Its value is
normalized between zero and one, thus ranges from a very poor level of handwashing habits (close to zero), to a quite good adoption of handwashing best
practices.
Such index is strongly and significantly correlated with the respondent’s gender: in fact, male respondents have a -.55 lower score, denoting that females are
more attentive to a good handwashing.
To conclude, the index is found to be evenly distributed across the different wards, as per graph below.
The second index is a proxy for the household wellbeing and takes into account variables concerning consumer durables ownership, such as mobile phones,
electricity, radio, television. Similarly, to the previously computed index, also this one ranges from 0 to 1, indicating a range from poor household material
wellbeing (close to zero) to a good wellbeing situation (index close to one). The overall average value of this score is 0.71, indicating a quite positive situation.
The distribution of such score is quite even across wards, with the only exception of Grafton, where the score distribution is less skewed and more evenly spread
around a lower mean value of 0.41.
MODELS
Main source of water during the rainy season

A multinomial regression model is run to investigate the significant relationships between the choice of main source of drinking water during the dry season and some
demographic/educational variables. The model can be expressed in the following form:

Y=𝛽0 + 𝛽1 𝐴 + 𝛽2 𝐵 + 𝛽3 𝐶 + 𝛽4 𝐷 + 𝛽5 𝐸 + 𝜀

Where y is the chosen main drinking water source during the dry season, and the explanatory variables are: the respondent’s education (A), the respondent’s age (B), the
respondent’s gender (C), the household size (D), the ward (E).

The model performs quite well (Pseudo-R2=0.32) and provides some insights in the choice of changing the main source of drinking water, given as base category the most
common one (street tap/pump). People with higher education are more likely to switch from street taps to unprotected dwells and to protected springs; moreover, people who
come from Portee are less likely to switch from street taps to Guma pipes.

Still, when we consider the marginal effect of the explanatory variable (i.e. a quantitative percentage variation in the outcome of interest, given a marginal variation in the
explanatory variable), we find that none of them is significant, as per the table below.

Piped Piped Borehole Protected Unprotect Protected Plastic Packet Bag/tie- Water Guma
water into water into dug well ed/open spring bottle water tie bowser Pipe
dwelling yard well water

Education 2.8^e-7 9.56e-06 -.0010927 .0022832 .000588 .0000147 NA .00001 NA .003652 9.81e-
07 9 06
Age of respondent -5.1^ e-7 .0000135 -.0006003 -.0020421 -.0008183 1.10e-06 NA 4.04e- NA .013944 1.61e-
06 2 06
Gender of respondent 3.3^e-5 -.0000404 .0025875 .4115784 .0050276 -.00004 NA - NA .255093 -
.00003 7 .000030
36 3
Household size -9.51^e-5 -1.01e-06 .0000435 -.0010023 -.0001596 2.75e-07 NA -2.29e- NA - 2.00e-
07 .000772 06
2
Ward:
Grafton -.009 -.4059989 .047293 .599541 .0471677 .1766494 NA 1.76e- NA -2.27e- -
11 11 .236875
8
Kossoh -.009 -.3523141 .143964 .5676478 .008916 .013585 NA .03064 NA -3.60e- -
Town 85 11 .236875
8
Portee -.009 -.4059989 .0917228 .2224672 .0152661 4.91e-11 NA .08625 NA .063224 -
71 2 .187574
Shell

Message reception, understanding, application

Several multinomial and ordered logit models are run to understand if the messages actually reach their target effectively, and how this might be determined by some socio
demographic variables such as education level, household size, respondent’s gender, and ward.

The model can be expressed in the following form:

Y=𝛽0 + 𝛽1 𝐴 + 𝛽2 𝐵 + 𝛽3 𝐶 + 𝛽4 𝐷 + 𝛽5 𝐸 + 𝛽6 𝐹 + 𝜀

Where y is the a Likert variable expressing if the hygiene-related message was received/understood/adopted, and the explanatory variables are: the respondent’s education (A),
the household size (B), the respondent’s ability to read (C), the ward (D), the respondent’s gender (E), the respondent’s age (F).

Unfortunately, no explanatory variable is found to be significantly correlated with any of the variables of interest.

Receiving information about hygiene

The majority of interviewed people answered that they would prefer to receive hygiene related messages from community health workers. Still, some demographic and socio-
economic factors can change their behavior. By estimating a multinomial model of the form:

Y=𝛽0 + 𝛽1 𝐴 + 𝛽2 𝐵 + 𝛽3 𝐶 + 𝛽4 𝐷 + 𝛽5 𝐸 + 𝛽6 𝐹 + 𝜀

where the explanatory variables are: highest level of education achieved (A), respondent’s age (B), household size (C), ward (D), respondent’s gender (E), household wellbeing
(F); we find some interesting aspects about the favorite information providers. More educated and more aged people will be significantly less likely to prefer info from a clinic/PUH
staff to community health workers. In addition, people that are more educated will be less likely to switch to community committee (e.g. WASH) or radio, when it comes to their
favorite hygiene info providers.

Garbage disposal
A multinomial model is run to explore how the garbage disposal varies according to a set of explanatory variables, both demographic and socio-economic.
The model can be expressed in the following form:

Y=𝛽0 + 𝛽1 𝐴 + 𝛽2 𝐵 + 𝛽3 𝐶 + 𝛽4 𝐷 + 𝛽5 𝐸 + 𝛽6 𝐹 + 𝜀

Where y is the a categorical variable indicating the way of garbage disposal, and the explanatory variables are: the respondent’s education (A), the ward (B), household’s size (C),
respondent’s gender (D), household wellbeing score (E).

We find that, given the base (most common category) of government garbage collection, the likelihood of switching to a private garbage collection company decreases for people
with some secondary education and for those who come from Grafton. Moreover, it is less likely to switch from government collection service to burning the garbage inside the
compound if the household wellbeing increases; in addition to this, bigger households are less likely to use dirty boxes for garbage disposals, instead of benefitting from the
government’s service. We also find that having some secondary education decreases the likelihood to pass to a community designated disposal area.

Remembering to treat drinking water

An ordered logit model is run to explore the factors that influence the difficulty of remembering to treat drinking water; the explanatory variables are the household wellbeing,
the household handwashing score (as proxy for the attention paid to hygiene), household size, highest education level achieved, ward and respondent’s gender.
Unfortunately, no regressor is found to be significantly correlated with such crucial behavioral barrier.

Main challenge in house sanitation

Coming to the main challenges that are faced by households in sanitation, the most common answer is the safe storage and proper treatment of water. Still,
some variables affect such opinion. A multinomial logit model of the form:

Y=𝛽0 + 𝛽1 𝐴 + 𝛽2 𝐵 + 𝛽3 𝐶 + 𝛽4 𝐷 + 𝛽5 𝐸 + 𝛽6 𝐹 + 𝜀

Where y is the a categorical variable indicating the main house sanitation challenge in the respondent’s opinion, and the explanatory variables are: household wellbeing (A),
household size (B), highest education achieved (C), ward (D), respondent’s gender (E).

We find that larger households where the highest education level is primary school or some secondary school and where the respondent is male, are more likely to change
opinion in favor to handwashing with soap; conversely, if the education level is just some primary school or tertiary school, the likelihood of considering handwashing as more
challenging will decrease; this is due to opposite reasons: in the first case, the ignorance will likely make people underestimate the problem, while in the case of highly educated
people, handwashing won’t represent an extremely pressing challenge.

Moreover, people who have attended and/or finished tertiary school, who live in Portee and are male will probably consider more urging the issue or drinking water treatment,
with respect to water safe storage and handling. People with similar characteristics – yet not living in Portee – will also probably consider more pressing the issue of latrine’s
usage.
To conclude, if the household wellbeing and the respondent is male, we will have a higher likelihood of considering proper waste disposal as more important; such issue will
rather decrease its relevance for people that both have either a low educational level (some primary school) or a high one (some or completed tertiary education).

You might also like