ISSN (Print) - 0012-9976 | ISSN (Online) - 2349-8846

A+| A| A-

A Reply

mortality rate, (7) Children (age 12-23 months) fully immunised, (8) Total voting population (+18 years of age), (9) BPL population, (10) Percentage of persons living in urban areas, (11) Under five mortality rate, (12) Households with electricity connection, (13) Habitations connected by pucca roads, (14) Households having access to safe drinking water, (15) Violent crimes (% of total crimes), and (16) Crime against women (% of total crimes).





A Reply


y article “Electoral Politics and the Manipulation of Statistics” had highlighted, based on a selective data check using publicly available secondary data sources, the problems with some of the data in circulation on the eve of the elections. In particular, it had analysed the data used by media houses, which, based on an exercise by Indicus Analytics, had compared the performance of various constituencies and states between two points of time, 2004 and 2008.

Apart from pointing out inconsistencies such as a sharp decline in per capita consumption expenditure alongside a sharp increase in incomes, the article also showed that Indicus data were not consistent with any of the secondary sources cited.

These problems were not only evident in constituency-level data, which required the use of statistical techniques for spatial disaggregation and aggregation, but also with state-level data, which were readily available from secondary sources. It also highlighted the irrelevance of using outdated and old (pre-2004) data for arriving at any conclusion on the performance of elected MPs and state governments for the period 2004-09.

The response by Laveesh Bhandari (LB) has failed to provide any evidence to counter the criticisms of the Indicus data. It has neither refuted any of the inconsistencies nor rejected the alternative data

Economic & Political Weekly

october 3, 2009

I presented in my article. On the other off the mark for individual years and also
hand, LB has mentioned the same sources in the opposite direction of trends suggest
in his defence (without giving any statis ed by the DLHS data. This is true not only
tical evidence), sources which I had for district-level estimates but also for
used to point out the problems with state-level data. Further, there is no
Indicus data. e xplanation for some of the obvious
While LB’s explanation of using small p roblems such as estimates of a sharp
area estimation technique is a welcome d ecline in per capita expenditure accom
insight into the estimation procedures of panied by a reduction in poverty. It is then
Indicus data, he has failed to satisfactorily unfair to harp on the small area technique
explain the details of the process. One when the very foundation of the database
would have liked greater clarity on the is wrong.
sources of data used, particularly for Finally, LB’s pontification on the effi
block-level indicators. But what surprises ciency or otherwise of public service deliv
us is LB’s claim that his methods differ ery is irrelevant here. My comment was
from state to state and district to district. not on why public service delivery is poor
While this explains some of what I can only but on the fictitious statistics used by Indi
describe as the weird and absurd results cus to analyse the performance of elected
arrived at by Indicus, it is obvious that representatives and governments. While a
such data cannot then be used for intra d ebate on why public service delivery and
state or inter-state comparisons, which is basic services do not work is welcome, it is
the main objective of the Indicus dataset. better done with credible statistics and
Further, it does not at all address the real statistical evidence.
problem with the dataset, which are that It is unfortunate that LB has not used
is generated by extrapolating from out the opportunity of the response and
dated data to derive supposed “evidence” EPW’s columns to be transparent on the
of the recent performance of state govern- Indicus data and to do so by presenting
ments and elected representatives. statistical evidence.
The lack of explanation on this count is a At the same time, Indicus’ claim of
major issue because the primary purpose of being the first to show certain trends
the exercise was inter-temporal compari reveals its own ignorance of contempo
son using recent data to evaluate the per rary academic work. For example, the
formance of constituencies. claim that it was the first to demonstrate
LB, in his response, has failed to men that Himachal Pradesh had caught up with
tion a single source of data which could Kerala in basic education is invalid. The
have been used to arrive at his district- Public Report on Basic Education (PROBE)
level estimates for so many indicators for report was the one that first showed the
2008. The most recent data that LB men r emarkable achievement of Himachal
tions are the District Level Household and Pradesh on basic education. For the record,
Facility Survey (DLHS) data – but these the PROBE report was published in 1999 –
actually show that Indicus data were way before I ndicus was established.
vol xliv no 40 63

Dear Reader,

To continue reading, become a subscriber.

Explore our attractive subscription offers.

Click here

Back to Top