Reflective Professional Development

A Critical Review of an Academic Article

Harma, Joanna. (2009) Can choice promote Education for All? Evidence from growth in private primary schooling in India, Compare, Vol. 39, No. 5, March 2009, 151-165

Aims and Scope

Harma’s aim is to demonstrate that existing claims about the capacity for ‘low-fee private’ (LFP) schools, in providing education for children of families at the poorer end of Indian society, have been overstated; that in fact a majority of India’s poor can not afford to send their children to LFP schools (2009). Consequently, they have no ‘choice’ but to send their children to government schools which are considered to be “failing” (PROBE Team, 1999 in Harma, 2009, p. 151).  

Moreover, she argues that choice, categorised conceptually as “voice and exit” (Hirschman, 1978 in Harma, 2009, p. 153), only applies to those who can afford it. Harma’s research is concerned with demonstrating a correlation between levels of poverty and school choice. She suggests that some studies, notably Tooley (2001) and Tooley and Dixon (2006) are not representative due to limitations, including the geographic location of the sample, sample size and the way in which previous studies have defined poverty (p. 152). The latter, forming an important feature of her research.

Harma has chosen the rural setting of Utter Pradesh in which to conduct her research, justified by claiming that rural areas represent “the living context of the majority of India’s poor” (p152). However, she provides limited evidence to support such a claim, representing one of the main shortcomings in her research: Harma’s use of existing research is, to some extent, manipulated to frame and tell the story she wishes to present. As will be discussed in more detail later, there is limited sense of critical voice at various stages of the article.

The scope of the research is appropriately large given the intended aims of the study however, there are concerns in relation to scalability. Certainly, the research can be considered to be applicable to similar rural settings however, this fact also established the limitations of the research in that the framing, and subsequent observations that took place provide a highly specified data set. It would be difficult to extrapolate from the research wider trends due to specific contextual factors not least the local population, number of schools and the social (caste/religious) make-up of a given area.

Moreover, she seeks to add reliability through qualitative data in the form of FGDs with parents, demonstrating their views about education in India. While this adds a more human dimension to the research, it is used in an limited way; adding little to the primary purpose of the survey. While the defining of poverty through land ownership and assets, arguably providing a more “stable and reliable picture” (p. 158), the FGDs, observations and interviews raise a number of questions that are never answered, not least: ‘Why are government schools failing?’

Research Strategy

The main research strategy is that of a detailed survey; both quantitative and qualitative data collected through interviews and observations. Despite potential biases on the part of researcher and participant, both methods provide targeted, contextualised data (Yin, 1994, p. 80) that can be standardised. Combining this with the initial ‘mapping’ exercise (p. 152) Harma was able to collate a detailed data set from which to develop a correlation between poverty and school choice.

Coen et al. define two types of factor analysis: exploratory and confirmatory (2007, p. 560). Harma categorises the stages of her research under these headings, however, given Harma’s knowledge of existing research, it could be argued that beyond the lack of known variables at the early stage of the study, the primary purpose of the study was arguably to prove an already established hypothesis about social stratification in India and its impact on school choice.

However, Harma’s explanation of the factor analysis is detailed and pragmatic. There is an emphasis on her desire to identify “broad trends” (p. 153) in keeping with the purpose of survey-based research to answer questions about an identified group or situation (Bell, 2005). Therefore, the establishing of a “indices” of poverty (p. 159) from a diverse range of variables is an effective and valid strategy.

While Harma’s description of the research strategy is to be commended, there are questions to be asked as to whether her chosen methodology is an effective indicator of the larger picture? (Coen et al, 2007, p. 207) Given the complex stratification of India’s poor as well as the geographic setting of the research, to what extent can the survey be applied in the wider context?

Reliability and Validity

As the research took place over a five month period, responses at different times could effect the reliability of the results. FGDs, are potentially unreliable due to the capacity for “response bias” (Yin, 1994, p. 80). Similar concerns can be placed on the interviews conducted, which are potentially subject to bias on the part of the researcher (Yin 1994, p. 80). As explained by Bell, reliability, particularly in surveys and interviews is reliant on transparency concerning the recording of information. (2005, p. 119) Harma, provides limited information about the questions asked during interviews and FGDs, arguably allowing her to present the material that suits her needs. This lack of transparency is a significant short coming of the article.

Bell, explains that for a sample to be representative it has to work beyond its specific context (2005, p. 120). The sampling range: 250 families (p. 152), certainly adds to the validity of the study large enough to be a representative sample. However there is limited justification for Harma’s choice of Utter Pradesh as a geographic location for the study. Additionally, spreading the research across 13 villages (p. 152) has to have garnered nuanced differences in response; especially given the caste/religious composition of the area. While these differences may not have affected the findings significantly, there is no acknowledgement of any outlying responses.

Overall, the survey provides a snap shot of a particular sub-set of India’s poor as opposed to being representative of the whole of India’s poor. Both the rural setting and the complicated social stratification described in Section 5.1 (p. 158-160) suggest that the the findings may not be directly applicable beyond Utter Pradesh itself.

Quality of Analysis

Cohen et al., explain that effective analysis of quantitative data must be systematic in order to ensure accuracy (2007). In this Harma, has been successful, the pragmatic approach she takes, describing and cross-referencing the quantitative data demonstrates a clear correlation between poverty and school choice (fig 1-3, p. 161-162)

Section 4 (p. 155-158) describes the results of the school observations. The data is presented accurately with recognition of the main differences between government primary schools and LFPs. The data clearly implies that there is a significant difference in provision between the two schools, however, the research itself falls short in providing almost no discussion about the reasons behind such apparent differences.

Section 5.1 and 5.2 (p 158-163) provide a manly quantitative, statistical analysis of the data collected. Harma’s use of three indices of income (table 3, p. 159) provide an increased level of authenticity to her analysis. Comparing this with the number of children who attend LFP schools it is consistently apparent that there is an “inverse relationship between poverty and access” (p. 160). Moreover, the complexities are discussed with consideration given to social factors including caste and religious background.

The latter stages of the analysis are more problematic. In trying to add a more human dimension to the findings, Harma begins to discuss more qualitative information, from interviews and FGDs. The subsequent analysis generates more questions than answers, with limited critical depth. The justification for parents wanting to see improvements in government schools (p. 163) never fully explained.

Author Reflectivity

The main weakness with Harma’s research is the lack of a critical voice. What is offered is a rather descriptive reflection of the research, rather than a probing, exploratory one. Harma weaves a succinct, well structured narrative, establishing the shortcomings of previous research, selecting references and quotations that support her hypothesis, shaping the views of the reader. For example, in citing research conducted by the World Bank in 2004 (p. 153), she fails to question the reliability, validity and context with which the research was conducted.

In describing her methodology (section 2) there is no acknowledgement of potential limitations. Statements such as “structured interviews” (p. 152) and “visited unannounced” (p. 153) represent the limited scope of Harma’s effort to demonstrate reliability.

Moreover, in sections 4, 5.1 and 5.2 there is evidence of constructed bias, achieved through careful selection of evidence and data being ignored. As Evans explains it is difficult to tell whether distortion of this type is deliberate or not (2002). However, I would argue that the emphasis placed on quantitative data over qualitative responses implies that there is a positivist bias with the research. It could be argued that the limited use of qualitative responses from interviews and FGDs (p. 162-163) is representative of a reluctance to delve into more complicated questions about the schooling situation in rural India; not least the short comings of Government Schools described on pages 157-158.

Finally, in concluding the article, there is no consideration of future research; surprising given the sample size and location. To what extent is the research representative? Even more surprising, given the questions raised about the current state of Government funded schools and the parents preference to see them improved (p. 163).


Overall, the research is methodical, offering a detailed, pragmatic insight into the schooling situation within Utter Pradesh. The use of multiple indices to define poverty, provide a triangulated data set, suggesting a clear correlation between poverty and school choice. However, the limited reflectivity of the author and potential shortcomings in terms of scalability require questions to be asked about the validity of the study.


Bell, J. (2005) Doing Your Research Project: a guide for first-time researchers in education, health and social science, Maidenhead: Open University Press (5th ed.)

Cohen, Louis., Manion, Lawrence., Morrison, Keith. (2007) Research Methods in Education. Taylor & Francis.

Evans, L. (2002) Reflective Practice in Educational Research, London and New York: Continuum.

Harma, Joanna. (2009) Can choice promote Education for All? Evidence from growth in private primary schooling in India, Compare, Vol. 39, No. 5, 151-165

Hirschman, A. (1978) Exit, voice and the state, World Politics, Vol. 31, No. 1, 90-107

PROBE Team. (1999) Public report on basic education in India, New Delhi: Oxford University Press

Tooley, J. (2001) Serving the needs of the poor: The private sector in developing countries, in Can the market save our schools?, ed. C Hepburn, 167-84, Vancouver: The Frazer Institute

Tooley, J. and P.Dixon. (2006) ‘De facto’ privatisation of education and the poor: Implications of a study from sub-Saharan Africa and India. Compare, Vol 36, No. ?, 443-62

World Bank. (2004) World Development Report, Retrieved 02.06.12 from the World Wide Web:

Yin, R. (1994) Case study research: Design and methods (2nd ed.). Beverly Hills, CA: Sage Publishing