ABSTRACT

 

Dr. Peter A. Sam Jr., CEI,CTS,CRS,CES

Adjunct Professor, Chair of AERCG

Department Of Geography/Environmental Studies

May 2002

University of Kansas

E-mail Address: psam@ku.edu or aercgc31@aol.com or sam.peter@epa.gov

 

Developing countries such as Ghana present a unique problem when it comes to indoor air pollution.  Since most people in developing countries utilize wood, charcoal and kerosene as fuel for cooking, they are exposed daily to many air pollutants via inhalation.  This problem is exacerbated by residential housing design, land use patterns and the vast amount of time spent indoors. Different residential housing types, land use patterns and socio-economic strata of communities also impact the degrees of indoor air pollution.

Qualitative and quantitative analysis on selected indoor air pollutants (CO, CO2, PM10), oxygen levels (O2), residential housing types, and time activity patterns (Household Survey/Questionnaires) were collected for 96 households in eight residential socio-economic areas (where n =12 per community) in the Greater Accra-Tema Metropolitan Area (GAMA), Ghana, West Africa.

Comparisons of the different socio-economic residential communities with respect to their levels of indoor pollutants were tested with a multivariate analysis of variance followed by univariate and post hoc pairwise tests.  Results showed statistically significant differences between communities, with the poorer communities having higher levels of air pollutants and lower levels of oxygen readings.

These results supported the main hypothesis that the concentrations of indoor air pollutants differ for each community type as a function of land use  and socio-economic status.  Communities with poor infrastructures, poor amenities and poor community design appears to have higher indoor air pollution.

Three secondary hypotheses related to the hypothesis of poorer household air quality in low socio-economic communities were also tested. These hypotheses concerned differences in air quality between households: 1). relying on different types of cooking fuel (LP gas, Kerosene, Wood);  2).  using different types of lighting (Kerosene, Candle), and 3). in which other smoke producing products are used (Tobacco, Incense, Mosquito Coil). These hypotheses also were tested with multivariate analyses of variance, followed by univariate and post hoc pairwise comparisons where appropriate. Households using Wood for cooking, predominately those in lower income communities, had significantly higher levels of Particulate Matter than houses using LP gas, predominant used in homes in higher income communities. No significant differences were found between households using Kerosene for lighting versus those using Candles. And no significant differences were found between households that have other sources of smoke  (Tobacco, Incense, Mosquito Coil) versus those that do not use these products. 

 

These results generally suggest that low income high density residential communities, exhibit high levels of indoor air pollutants as compared to high income low density residential communities and that the levels of indoor air pollutants in Ghana vary significantly with the type of cooking fuel, land use design, and socio-economic status of communities.

 

Preliminary analyses were conducted to examine and describe the sample of households and to examine the properties of the measurement data collected.  These analyses consisted of frequency tabulations, means and standard deviations and other descriptive statistics (Appendix O, Tables I through XII).

Since residential communities are very homogeneous with respect to socio-economic status, I used Community as a Since residential communities proxy measure of socio-economic status. 

To test the null hypothesis of “No Socio-economic differences in household air quality” I used Multivariate Analysis of Variance, which tested for Community differences on all the air quality measures taken together.

 The result shows that there was a significant Community difference in air quality at the probability level of .001, less than my 0.10 criterion.  Wilk’s Lamda can be interpreted as a measure of variance NOT accounted for, and as you see, here it is a very low .075, meaning most of the air quality differences found can be attributed to inter-community differences.

Since the air quality measures are inter-related, it is not possible to precisely test for community differences on each measure in isolation, however I did conduct Univariate follow-up tests as a way to help characterize community differences.  Results, as you see, show that there were significant community differences on each of the air quality measures taken by itself.

Pairwise comparisons (with Bonferroni correction to protect against Type I error) showed that the community differences were between the low socio-economic communities and the high socio-economic communities.

 

   Discussion Box 1:

Hypothesis 1: No Socio-Economic Differences in Residential Air Quality.

A                             Manova of CO2, CO, PM10,O2 by COMMUNITY

1                          Significant difference , Wilk’s Lamda = .075,  F(28,268) = 8.69, p < .001

B                              Univariate analyses of each measure by Community.  All significant.

                      1.            CO2  ---- F(7,67) = 5.79, p <.001

2.             CO    ---- F(7,67) = 5.60, p <.001

3.             O2     ---- F(7,67) = 19.23, p <.001

4.             PM10 ---- F(7,67) = 9.59,  p <.001.

 

 

 

 

 

Next, I tested the hypothesis that – No cooking fuel differences in residential air quality.  Once again, I used Manova to test for differences in air quality, this time between homes where Wood is the primary cooking fuel versus Kerosene and LP Gas. Once again, I reject the null hypothesis of No Difference in air quality due to cooking fuel used.  Wilk’s Lamda  is again very low, meaning that much of the variance is accounted for by cooking fuel used.

Univariate follow-up tests showed that air quality differences center on cooking fuel effects on residential O2 and PM, that is, there were significant cooking fuel differences in levels of O2 and PM in these residences, but not in CO or CO2.

Pairwise Comparisons showed that Wood burning homes had significantly Lower O2 and significantly Higher PM that homes that burned LP Gas, that is, Wood-burning homes had poorer air quality.

Inspection of means showed that Kerosene users were somewhere in between LP and Wood, but not significantly different from either, perhaps due lack of statistical power in this study (i.e., insufficient Kerosene users in this sample).

Once again, since LP Gas and appliances are so much more expensive than wood and wood stoves, and only homes in the upper socio-economic communities use LP Gas, differences in air quality actually reflect socio-economic differences.

 

  Discussion Box 2:

 
 
II.Hypothesis 2: No Cooking Fuel Differences in Residential Air

                               Quality.

 

A                             Manova of CO2, CO, PM10,O2 by Cooking Fuel (LP Gas versusWood versus Kerosene)

1                          Significant difference , Wilk’s Lamda = .073,  F(8,108) = 2.32, p < .02

B                              Univariate analyses of each measure by Cooking Fuel.  Two significant.

1.             O2 ---- F(2,57) = 4.54, p <.02.

                                    PM10 ----- F(2,57) = 4.32, p <.02.

 

C                              Pairwise Comparisons

1                          Wood-burning homes have significantly lower O2 than LP gas-burning homes.

2.             Wood-burning homes have significantly higher levels of PM10

 

 

 

 

 

I conducted Multivariate Analysis of Variance to test whether air quality differed in homes that used Candles for lighting versus those that used Kerosene.  The omnibus test, taking all 4 air quality measures together found no significant differences between candle burners and kerosene burners, and Wilk’s Lamda shows that very little variance was accounted for by group differences.

Since no significant differences were found on the four measures taken together, no follow-up tests were conducted. I also ran Manova to test for effects of Use of Other Smoke-Producing products on air quality.  As with Lighting fuel, above, I found no significant difference in air quality, taking all 4 measures together, between households that reported use of cigarettes, mosquito coils or incense by residents and those households that reported no use.  As above, since no significant differences were found, follow-up tests were not conducted.

 

         Discussion Box 3:

III.Hypothesis 3: No Lighting Fuel Differences in 

                            Residential Air Quality.

A                             Manova of CO2, CO, PM10,O2 by Lighting Fuel.

1                          No Significant difference , Wilk’s Lamda = .87,  F(4,70) = 2.51, p = .05.

 

 

 

 

           

 

 

 

 

 

 

 

 

          Discussion Box 4:

IV.Hypothesis 4: No Other Smoke Producing Products

                             Differences in Residential Air Quality.

 

A. Manova of CO2, CO, PM10,O2 by Use of Other Smoke-Producing Products.

1                          No Significant difference , Wilk’s Lamda = .97,  F(4,70) = .467, p = .76.