STATISTICS
From Sampath’s Desk:
STUMBLING STATISTICS
JUNE 29
NATIONAL STATIST8VS DAY
They say there are three things – lies, white lies and statistics
Statuary Warning
(This is a dry subject, excuse)
One simple definition of statistics of Prof.
Horace Sacrist is, “By statistics we mean aggregates of facts (expressed in
numerical terms) and placed in relation to each other.”
There are two types of data, primary and
secondary. Primary data is the one collected in an initial survey. Thereafter,
when used by different persons and agencies, it is secondary data. Secondary data is cheap and less time
consuming. Careful, realistic and discreet selection of samples from a riot and
mélange of information from different strata of the target population may
increase data’s reliability and representativeness to be used and reused to
check different variables. However, it is not an unmixed blessing. A researcher
cannot check the validity and find a mechanism for a causation theory; he/she can
only draw patterns and correlations from the data. Statistical data being more
often secondary one, it can be easily misinterpreted, liable to be misused or
abused by the researcher by manipulation to impose his/her own views as conclusions and research findings.
Unlike physical laws
which are permanent and not likely to change at all, laws on abstract sciences
like economics may vary, as the world constantly transmutes itself,
of late, increasingly just by the turn of the day. The statistical parameters like mean, median, mode, deviations
including standard deviation, co-efficiency of correlation, probability factor,
etc. have different relevance in different contexts. While the average (mean) factor holds good in many fields as a
rule, when it comes to certain specifics, say like for crossing a river,
average depth (mean) makes no sense. And trying to cross a river
relying on its ‘average depth’ will be disastrous. Again, average temperature of a person suffering from
fever - arrived at based on readings taken many times a day or over a
period - means nothing. Average rainfall of a state, say Tamil Nadu, has
no relevance per se to its capital city Chennai.
All principles don’t work for all activities
and/or all parameters at all times. The GDP or per capita income, albeit
throwing some light on the economic status of a country, is not inclusively and
conclusively perfect. Even comfortable figures of GDP, per capita income and
percentage of growth rate don’t reflect the level of either poverty or
prosperity of individuals, groups, states and/or regions.
Sampling and processing of data are the key
functions in a research. Some research analyses and conclusions arrived at were
falsified on ground, because of respondents’ cavalier response during survey
not carrying truth. Lack of cooperation, casual responses, suppression/distortion
of truth, etc. will lead us nowhere. We therefore need more scientific models
and truth-extracting response formats based on realistic input paradigms with
in-built provisions for veracity cross-checking that need to be deployed to achieve
as closest accurate and credible output as possible while processing the sample
data.
The 2010-Barrack Obama period recession was
actually moderate and not all so alarming as was made out, and was certainly
not as bad as the 1930’s great recession though some analysts, based on statistics,
pressed the panic button.
Some criticisms of
statistics are – there are three types of lies, simple lies, damned lies and
statistics - statistics are not substitute for judgment - correlation is not
always causation - you cannot feed the hungry with statistics - it is the
science of producing unreliable facts from reliable figures - facts are
stubborn but statistics are more pliable, etc. Sometimes, statistics are ‘cooked
up figures’ not backed by reality. In that case, it would set you at naught.
George Canning asserted, “I can prove virtually
anything and everything by statistics except the truth.” To quote George
Bernard Shaw, “Statistics show that of those who contract the habit of eating,
very few survive.” Funny, isn’t?
R.SAMPATH
2/7/2020
PS: - ‘World Statistics Day’ is observed on the 20th October, once in 5 years. It started on 20.10.2010. So, we will have the day on every 20th October in 2015 (already over), 2020, 2025, and so on. India observed ‘National Statistics Day’ on June 29 2020 on the birth centenary day of Padma Vibhushan Prof Prashant Chandra Mahalanobis, remembered as ‘Father of Statistics’ in India. Thus, in India, we observe 'STATISTICS DAY' on June 29 every year.
Comments
Post a Comment