Chapter 4: Empirical Methods and Fieldwork in Malawi
4.4 Data transformations and analysis and epistemological issues
After each round of household survey, the questionnaires were processed and entered in
SPSS. It is important at this point to stress the restricted character of the empirical fieldwork
research. In order to support the secondary data of this thesis, earlier described, the researcher
conducted a limited fieldwork exercise to gain a more definite feel for critical strengths and
weaknesses of the two programmes under discussion in this thesis. The researcher had no
resources to carry out a full-scale livelihoods and vulnerability analysis of a large sample of
households. For this reason, this thesis is mainly concerned with simple statistical methods for
describing observed food security effects of the two schemes under consideration in this
thesis. To test for differences in means, One-way Analysis of Variance (One way ANOVA)
comparing at least three sample groups or Independent-Samples T-Test comparing two groups
have been employed on the data. To test for differences in proportions between groups,
Mann-Whitney U test (for 2 independent samples) and Kruskal-Wallis H Test (for k
independent samples) have been conducted. In addition, simple regression (y=α+βt) is applied
to the different productivity time series (secondary) data in order to test for the existence of a
positive or negative trend. Simple correlation analysis using Pearson’s product-moment
correlation coefficient is also conducted to test for linear positive or negative relationships
between two variables. These are most appropriate tests to yield the evidence sought in this
research (Colman and Pulford 2006).
A number of transformations have been conducted to the original data in order to permit
comparisons between households. To allow comparisons of labour availability, ages of
household members were converted into adult labour equivalent using conversion factors in
Table 4.18. From the literature all persons above the age of 7 years provide productive labour
in one way or the other, especially on family farms in Africa (Johnson 1982, p.205). In doing
this calculation, there is no implication that ‘child labour’ is desirable. To the extent that the
labour contribution of persons aged below 19 years is considered child labour is a matter of
government policy, which currently lacks a proper framework.
43The government has so far
not outlawed or criminalized the participation of young people either on family farms or in
ganyu in neighbouring farms. The attention to date has focussed on child labour in
commercial tobacco and tea estates because of a previous history of using children as cheap
labour (Eldring 2003, Otanez et al. 2006). In fact, the ISP as a main vulnerability reduction
policy initiative in Malawi targets (amongst other categories) households where the head
43
The Malawi Constitution provides for the protection of children from economic exploitation. The Employment Act of 2000 prohibits the employment of persons below the age of 14 but allows the employment of persons aged 14-18 years as long as it does not harm child development (Eldring, 2003, p.12) labour (Eldring 2003, p.12). Government economic reports (e.g. by NSO) describe economically active age group in Malawi to be 15-64 years.
cannot provide active labour but there are dependents, including children, who can (Mwale
2009). In the Medium Term Plan for the Farm Inputs Subsidy Programme covering the period
2011-16, there is a stated policy provision to reach resource poor households headed by
children and orphans, in the same way as special consideration is to be given to households
headed by the elderly, HIV positive persons, females, disabled persons, or household heads
caring for the elderly, chronically ill or disabled persons (Government of Malawi 2010d,
p.16). During the fieldwork, children as young as five years were reported to have participated
in family farms and even in ganyu but for the purpose of this thesis, labour input of
individuals aged below the age of 7 years is considered to be zero.
Table 4.18: Labour conversion factors (adult units equivalents)
Age (Years)
Female
Male
Average*
Below 7
0.0
0.0
0.0
7 -14
0.4
0.4
0.4
15-64
0.8
1.0
0.9
65 and above
0.0
0.5
0.5
*
author
calculations to remove gender biasSource: Johnson (1982, p.205)
The fieldwork also monitored availability of food (maize) stocks at every survey round. The
stocks were collected in both grain and ufa (maize flour) in units reported by the households
and applicable to most rural Malawi. To permit comparisons to be made between households
in the sample, the quantities were ‘standardized’ into kilogrammes using conversion factors
developed by NSO and widely used as official measures for Malawi (see Table 4.19).
Table 4.19: Maize conversion factors for Malawi
Measure of maize
Kilograms
50 kg bag
46.7
90 kg bag
84.0
Pail (small)
8.7
Pail (large)
20.0
No. 10 plate
0.2
No. 12 plate
0.8
Basket (dengu) shelled
34.6
Basket (dengu) unshelled
13.3
Oxcart (unshelled)
32.3
Oxcart (shelled)
350.0
The ufa was reported in the form of ufa woyela (refined meal), mgaiwa (whole meal) or
mixed. To permit comparative analysis of households, the ufa was converted into maize grain
equivalents using factors derived by leading researchers on food security and nutrition in
eastern and southern Africa (Jayne et al. 1996). These were consistent with conversion factors
earlier provided by nutritionists in the Department of Home Economics and Human Nutrition
at Bunda College of Agriculture of the University of Malawi
44. According to Jayne, et al.,
(1996, p.5), the extraction rate for mgaiwa is 96-99 per cent while the rate for ufa woyera is
65 per cent. To account for situations where the ufa in the sample was mixed, an average rate
was derived for this thesis as follows: the average of extraction rate for ufa woyera plus the
average extraction rate for mgaiwa ((60 + (96+99)÷2) ÷2). This gave a working rate of 78.8
per cent with which the ufa was converted back into maize equivalents. In any case, less than
10 per cent of the sample households reported food in ufa form at every survey visit.
To permit comparisons of the duration (days) the stocks would take to deplete, the stocks
were ‘standardized’ into household maize calorie equivalents by applying a conversion factor
(see Table 4.20) derived from two separate studies conducted on Malawi. In 1992, FAO
estimated that 468.8 grams of maize in Malawi provided 1,422 calories equal to 3,033.3
calories for each kilogram of maize (FAO 1992, Table 24). In 2010, Ecker and Qaim (2010,
p.5) estimated that 381.7 grams of maize provided 1,332 calories equal to 3,489.7 calories for
each kilogram of maize.
45These two sources provide an average figure of 3,261.5 calories for
each kilogram of maize, and this is the figure utilised for nutritional conversions in the rest of
the thesis.
The household maize calorie requirements were computed by applying conversion factors that
were provided by the government in the 1998 Profile of Poverty in Malawi that drew on IHS-
1 (see Government of Malawi 2000b, p.109). The factors are adjusted for age, sex and scale
of activity a person performs but the author derived age-based averages to remove ‘gender
biases’. On the basis of these conversion factors, the average per capita maize requirement is
0.43 kg per day or 158.4 kgs per year. To estimate number of days the food stocks would take
44
Mr Kingsley Masamba, Lecturer in Food Science at Bunda College, provided an extraction rate of 90 per cent for Mgaiwa and 60 per cent for ufa woyera as conversion factors used by Department of Home Economics and Human Nutrition at the College.
45
Mr Neil Orchardson (Technical Advisor for Food Security and Nutrition) of the Ministry of Agriculture and Food Security informed the researcher in October 2010 that the Ministry has since adopted this as the ‘official’ conversion factor.
to deplete, total household maize calorie requirements per day were derived by summing up
daily maize calorie requirements of different individuals in the household. This was then used
as a denominator with which to divide the total maize calorie equivalents:
Days food would last = Available food stocks (maize calorie equivalents)
Daily household maize calorie requirement
Table 4.20: Per capita calorie and equivalent maize requirements (kgs)
Age(Years
Mchinji sample Equiv group (yrs)
Daily calorie requirement Daily maize calorie Daily maize requirements (kgs) Annual maize requirements (kgs) 46 <1 0-0.9 820 597.0 0.18 65.7 1-2 1-1.9 1,150 837.2 0.26 94.9 2-3 2-2.9 1,350 982.8 0.30 109.5 3-5 3-4.9 1,550 1,128.4 0.35 127.8 5-7 5-6.9 1,800 1,310.4 0.40 146.0 7-10 7-9.9 1,950 1,419.6 0.44 160.6 10-12 10-11.9 2,075 1,510.6 0.46 167.9 12-14 12-13.9 2,250 1,638.0 0.50 182.5 14-16 14-15.9 2,400 1,747.2 0.54 197.1 16-18 16-17.9 2,500 1,820.0 0.56 204.4 18-30 18-29.9 2,600 1,892.8 0.58 211.7 30-60 30-59.9 2,567 1,868.5 0.57 208.1 60+ 60+ 2,225 1,619.8 0.50 182.5 Average 1,941 1,413.3 0.43 158.4
Source: as explained in the main text above.
4.4.2
Epistemological position in this thesis
Research in the field of social sciences entails purposive and rigorous investigation that aims
to generate new knowledge (Sarantakos 2005, p.4) based on facts that are not just given but
also produced (Mukherjee and Wuyts 1998, p.243). Different research works adopt different
epistemological stances regarding views on ‘generation of new knowledge’ but this research
adopts view that allows for the comparison of different ideas on a relative basis (Proctor
1998b). This epistemological position can be referred to as critical realist strands of thought
which approaches issues of knowledge as constituting both realism and practical activity; that
reality exists but it has to be interpreted within a given context by interpreting the observed
interactions between powers, institutions, actors or forces. In the field of social sciences, the
46
As a rule of thumb, the government encourages Malawians to keep two and a half 50-kg bag of maize per person per year. But working figures vary considerably. In the BNB surveys described in earlier sections of this Chapter, the methodology has adopted a working maize requirement of two 50-kg bags per month for a family of six persons, translating into 200 kgs per person per year. The average of the two sources yields 162.5 kgs per person per year, which is very close to 158.4 kgs derived for this study – a difference of 4.1 kgs may be acceptable for all practical purposes.
purpose is to uncover deeper issues around causes, patterns, outcomes and effects around
themes of research interest; the interaction between ‘powers’ in issues around poverty and
vulnerability and their most recent policy responses emphasizing social transfers/agriculture
in the case of this thesis. This interpretive process involves aspects of subjectivity as set of
empirical methods are explained and elaborated. The results of such analytical process in this
thesis can provide the basis for ‘policy’ change. In this research, examples that have received
critical analysis include power relations (e.g. role of private sector in ISP), ethical (e.g.
leakages of coupons or social transfers), economic (e.g. cost effectiveness of social transfers)
or political (e.g. politics of social transfer targeting and reporting of the outcomes that have
created unexplained gaps with reality on the ground (Bhaskar 1989, Proctor 1998a, Carter and
New 2004). This study combines social sciences including economics (e.g. optimal use of
resources such as inputs) and politics (how political factors can override evidence-based
policy making) with agriculture (e.g. maize production), geography and environment. These
disciplines fall into either or both of the two main paradigms of positivist or post-positivist
perspectives on knowledge generation. Importantly for this study, a number of different
fieldwork methods were found appropriate, and a combined qualitative and quantitative
approach was followed (Booth et al. 1998, Ellis 2000, Kanbur 2003).
The field work was conducted with the assistances of six research assistants (RAs) who
helped with field data collection variably at different periods but three worked on more or less
‘permanent basis’ while the other three were brought in periodically to confirm the data
collected. The original plan was to recruit research assistants resident in each of the three
study VDCs. It was apparent from initial consultations that the CSCTCs did not favour the
idea because it would kill the spirit of volunteerism since members of the CSCTCs were not
receiving any formal remuneration for their time, resources and services.
47On the basis of
these reservations, two assistants were recruited from communities neighbouring VDCs not
participating in the study. Agricultural staff at district and Mikundi EPA helped in the
identification and recruitment process. One RA was recruited from the district headquarters
(Mchinji Boma) and worked with the researcher throughout the study from initial sample and
community selection to data entry. All the three RAs commuted to the study sites on daily
basis using bicycles.
47
It is discussed in Chapter 6 that members of the CSCTCs are not volunteers in strictest sense because they draw a monthly allowance, now at MK1500 per month in addition to daily allowances every time they participate in project activities, mostly outside their communities.