SEC: Political Leadership & Communication - Unit 4 & 5

On Psephlogy, Election Studies & Survey Methods

● Voting in elections is the most visible form of political participation in democratic

countries.
● The origin of the study and statistical analysis of elections can be located within the
discipline of political science which with the passage of time developed into a
subdiscipline called Psephology. Hence, Psephology is the study of elections based on
precinct data on voting, public opinion polls for gauging the mood of the voters,
information on campaign finance and other available statistical data on elections.
● Such election surveys are conducted by media and polling organizations all over the
world to measure the popular mood of the voter during the elections, to find out issues
that would be crucial in the elections, the voter’s choice of political parties and leaders,
and the voting intentions of the electorates. The opinion polls do an in-depth and holistic
measurement of voting behavior and opinion of the electorates that include gains and
losses for political parties or candidates in terms of vote share and their winnability
chances in the elections.
● The opinion and attitudes of the voters are gathered through empirical research and
are quantified or translated into figures to provide a macro-view generalization
about electoral trends and patterns in the elections.
● While one gets an impression that opinion polls are new in India, there is a long history
of conducting election analysis/election studies which dates back to Sirsikar’s study of
the Poona Lok Sabha constituency in the year 1967. This was followed by a large-
scale election survey of the electorate by the Centre for the Study of Developing
Societies (CSDS) in the same year, marking the beginning of measuring voting
behavior and attitudes in India. The historical overview of election surveys provides a
key insight and perspectives about the development and growth trajectory of opinion
polls in India.
● In fact, The National Election Study series based on post-poll surveys conducted by
the Lokniti, Programme for Comparative Democracy at the Centre for the Study of
Developing Societies, is the largest and the most comprehensive social scientific
survey of Indian General Elections and perhaps of any election in the world. The NES
series not only is a study of voting behavior of the Indian electorate but also gathers the
most robust information about how Indians voted in the various national elections. The
NES treats elections as a window to capture the most accurate snapshot of the
political behavior, attitudes, and opinion of the Indian voters. The last NES
conducted in 2009 also explored the awareness levels and opinion of the Indian
citizens on issues concerning the Indian economy, national security, democracy, and
diversity. The National Election Study is the most comprehensive information
 database of social and political change in India and has been used as a source for
international comparative studies. The purpose of conducting election surveys in
India is to measure voting patterns and trends of the electorate, their electoral
preferences and choices, issues that voters considered important, and the reasons
for their participation or nonparticipation in elections.
● The Election Commission of India houses the election results and data of all general
elections and state assembly elections held in India since independence. The data are
in the public domain and the electronic copies of it are available on their official website.
The kind of information available at ECI are: (a) the number of people who voted in
Parliamentary or Assembly elections that is termed as Voter Turnout; (b) the list of
candidates who contested the elections at the national/state level, the winner, the runner
up, and the percentage of votes polled by all the candidates; (c) the number of political
parties that contested the elections and the votes polled by them. It also provides the
number of candidates fielded by parties and the number of candidates that won; (d) data
on the gender break up of candidates who contested and won the elections. It also
provides information about the scheduled castes (SC) and scheduled tribe (ST) status of
candidates elected in a particular election and (e) it also archives the affidavits filed by
the candidates who contest elections containing particulars like judicial cases filed
against them, property held by them, and other personal information for public scrutiny.
● However, it does not provide any data about the voting behavior and attitudes of the
electorates like how people from different socioeconomic and caste communities voted in
the election, whom they voted for or did not vote for and why they voted for a particular
party or candidate. Did the young voter have different political preferences compared to
the old? Are most women voters guided by their husbands in taking their voting
decisions? Which election issue dominated the mind of the voters during a particular
election? There are numerous questions of such kind, which can’t be answered from the
results/data available from the ECI. The available information cannot answer questions
related with elections like: Do Muslims in India vote for the right wing political party
BJP or do they vote en bloc and strategically vote to defeat the party in elections?
Forward castes are considered to be traditional supporters of national parties like
Congress or BJP always vote for them or do they vote for regional parties also in some
states? What have been the voting patterns of voters from communities like Dalits and
Adivasis in India? Has the emergence of parties based on scheduled castes/tribal
identities led to the consolidation of scheduled castes/tribal votes in their favor? What is
the support base of political parties at the national level and in different states of India
and what does their profile of voters look like? Do electoral choices amongst voters, from
different background (educated/illiterates, rural/urban, and rich/poor) vary significantly?
The study of behavioral or contextual aspects of voting in Indian elections becomes
important and relevant as answers to these can be found in an evidenced manner by
conducting empirical election studies.
● Hence, election surveys are scientific methods of answering if not all, but to a great
extent many such questions. The overall purpose of these electoral studies is to find
out the voting patterns and trends based on the opinion and attitudes of the voters,
by probing key issues like why people vote or do not vote, how they arrive at voting
decisions and questions connected with voters’ engagement and participation in the
elections.
● While election surveys are more than merely predicting the likely outcome of elections,
in recent times election surveys are increasingly being used as a mere tool of forecasting
election results. Yet, for political analysts and pollsters it is more than just predicting
numbers, it is also about a deeper scientific analysis of elections like who voted for
whom, how did the vote swing from one party to another, was development an issue, did
anti-incumbency affect the ruling party or not?
○ The swing of votes helps in explaining the election verdict in two ways: (a) the
general measure of the degree of movement of popular electoral support in favor
of the main winning parties and (b) in systematically measuring the relationship
between a party’s gain or loss in popular votes and its gain or loss in the number
of seats won or lost.
● Hence, there is a proper scientific method involved in conducting election surveys
and forecasting elections. It is as scientific as social science could be with various
inherent limitations of analyzing human behavior, opinion, and attitudes. For
example, Yogendra Yadav, one of the leading political analysts in India states that
“psephology is not a discipline, a science like microbiology is. Psephology is nothing
more than election studies”
● This is further so given the additional challenges of first-past-the-post electoral
system, nature of coalition politics, and electoral competition that complicate
forecasting elections even with one of the most reliable survey data.
● This raises the question how voting behavior and attitudes are commonly measured and
the kind of scales that are used for measuring them in quantitative terms. There are
various research methods for studying the voting behavior of the electorates like sample
survey, case study, and participation observatory study. However, case studies and
observation studies do not use quantitative methods for studying elections, so voting
behavior cannot be measured in certain terms. Thus a sample survey of the electorate
remains the most commonly used method for measurement of voting behavior as it
generates quantitative data for analysis and drawing inferences.
● Surveys meld sampling, question design, and data collection methodologies.
● Hence, in order to conduct a reliable election survey, one needs to collect data from a
scientifically selected sample of voters which is representative of all the voters. Hence, it
is an important art to learn to collect electoral data from a representative sample derived
from several crore voters who may be geographically spread nationally or in one state.
● This often involves learning the art of data collection and using a carefully designed
questionnaire for the collection of information from the voters.
 ○ The questionnaire design involves a lot of deliberations, discussion, and pre-
testing; it is not merely putting down a few lines and reading out those lines by
way of asking questions to the voters.
○ Similarly, data collection by way of conducting field work is a challenge. The
challenge of data collection is due to enormous social, religious, linguistic, and
geographical diversity. Different situations demand flexibility in the method of
data collection without losing out on the basic method of data collection which is
standardization. Sensitive questions which people may not be willing to answer in
the open or in the presence of others need to be asked with utmost care, the voting
question being a good example.
● Even with a fairly reliable survey, one could make an error in converting vote estimates
into seats, which in common parlance is referred to as seat forecast. The survey gives us
an estimate of the vote share, which is converted into seats using statistical tools or
models. Different pollsters use different predictive models and every model has its own
strength and weakness. The difficulty of the workability of any model is compounded by
the first-past-the post electoral system where victory or defeat could be decided by one
vote or by a much bigger margin. Various predictive models which have been used for
converting seat forecasts fail to account for marginal victories or “surplus votes” or
“wasted votes”—in case of very big victories.
● The frequent splits and mergers of political parties and the changing nature of coalition
between two elections add to the problem of adjusting the vote share of the party in the
previous election complicating the method of seat forecasting.

# Survey Method

● Surveys are designed to produce statistics about a target population. The process by
which this is done rests on inferring the characteristics of the target population from
the answers provided by a sample of respondents.
● Most people are familiar with three uses of survey techniques: the measurement of public
opinion for newspaper and magazine articles, the measurement of political perceptions
and opinions to help political candidates in elections, and market research designed to
understand consumer preferences and interests. Each of these well-developed programs
of survey research is aimed primarily at tapping the subjective feelings of the public.
There are, in addition, numerous facts about the behaviors and situations of people that
can be obtained only by asking a sample of people about themselves. There is probably
no area of public policy to which survey research methodology has not been applied.
● In addition to meeting needs for data that are not available elsewhere, there are three
potential properties of data from a properly done survey that may make them preferable
to data from other sources:
○ Probability sampling enables one to have confidence that the sample is not a
biased one and to estimate how precise the data are likely to be. Data from a
 properly chosen sample are a great improvement over data from a sample of those
who attend meetings, speak loudest, write letters, or happen to be convenient to
poll.
○ Standardized measurement that is consistent across all respondents ensures that
comparable information is obtained about everyone who is described. Without
such measurement, meaningful statistics cannot be produced.
○ To meet analysis needs, a special-purpose survey may be the only way to ensure
that all the data needed for a given analysis are available and can be related. Even
if there is information about some set of events, it may not be paired with other
characteristics needed to carry out a desired analysis. For example, hospital
discharge records invariably lack information about income. Hence, a survey that
collects both income and hospitalization data about people is needed to study the
relationship between a person’s income and hospitalization experience.

● Components of a good survey: A sample survey brings together three different

methodologies: sampling, designing questions, and data collection.
○ Sampling: A major development in the process of making surveys useful was
learning how to sample: to select a small subset of a population representative of
the whole population. The keys to good sampling are finding a way to give all (or
nearly all) population members the same (or a known) chance of being selected
and using probability methods for choosing the sample. Early surveys and polls
often relied on samples of convenience or on sampling from lists that excluded
significant portions of the population. These did not provide reliable, credible
figures.
○ Question design: The initial survey efforts, representing extensions of journalism,
were not careful about the way that questions were posed. It soon became
apparent, however, that sending an interviewer out with a set of question
objectives without providing specific wording for the questions produced
important differences in the answers that were obtained. Thus, early in the 20th
century, researchers began to write standardized questions for measuring
subjective phenomena. The major advance in question design in the last 20 years
has been improved strategies for evaluating questions. More than before,
researchers now evaluate questions to find out if they are well understood and if
the answers are meaningful (see Presser et al., 2004; Madans, Miller, Maitland, &
Willis 2011). Pretests of surveys have become more systematic, using analyses of
tape-recorded interviews to identify problem questions. As a result, the choice of
question wording is becoming more objective and less a matter of research
judgment.
○ Interviewing: Although not all surveys involve interviewing, it certainly is
common to use an interviewer to ask questions and record answers. When
interviewers are used, it is important that they avoid influencing the answers
 respondents give, at the same time maximizing the accuracy with which questions
are answered. The first major step in increasing interviewer consistency was to
give them standardized questions. It subsequently was found that interviewers
also needed to be trained in how to administer a survey to avoid introducing
important biases in the answers they obtained.
○ With respect to sampling, critical issues include the following:
■ the choice of whether or not to use a probability sample
■ the sample frame (those people who actually have a chance to be sampled)
■ the size of the sample
■ the sample design (the particular strategy used for sampling people or
households)
■ the rate of response (the percentage of those sampled for whom data are
actually collected)

● With respect to question design, the researcher must decide the extent to which previous
literature regarding the reliability and validity of questions will be drawn upon, the use of
consultants who are experts in question design, and the investment made in pretesting and
question evaluation. With respect to interviewers, researchers have choices to make about
the amount and kind of training and supervision to give.
● A design decision cutting across all these areas is the mode of data collection: whether
the researcher will collect data by telephone, by mail, by personal interview, over the
Internet, or in some other way. The decision about which mode of data collection to use
has important cost implications and affects the quality of the data that will be collected.
● These pieces, taken together, constitute what is called the total survey design.

● Two of the main goals of survey methodology are to minimize error in data collected by
surveys and to measure the error that necessarily is part of any survey.
○ One fundamental premise of the survey process is that by describing the sample
of people who actually respond, one can describe the target population. The hope
is that the characteristics the survey is designed to describe are present to the same
degree, and are distributed in the same way, in the sample responding as in the
target population as a whole. One goal of survey methodology is to minimize the
random differences between the sample and the population.
■ The way the sample is designed and selected can affect how closely the
sample is likely to mirror the characteristics of the population from which
it is drawn. This variation, the possible error that stems solely from the
fact that data are collected from a sample rather than from every single
member of the population, is called sampling error.
■ A second kind of error that affects the relationship between a sample of
respondents and of the population is bias. Bias means that in some
systematic way the people responding to a survey are different from the
 target population as a whole. There are three steps in the process of
collecting data from a sample, each of which could, potentially, introduce
bias into a sample:
1. The first step involves choosing the sample frame, those who
actually have a chance to be selected. If there are some people in
the target population who do not have any chance at all to be
selected for the sample, and if they are somehow consistently
different from those who do have a chance to be selected, the
resulting sample will be biased in those ways.
2. If somehow the process of selecting who is in the sample is not
random, the result could be a sample of respondents who are
different from the target population as a whole.
3. Finally, failure to collect answers from everyone selected to be
in the sample is a third potential source of bias.
● It is important to understand the distinction between the two kinds of
errors in data. Sampling error, discussed above, and random error. By
chance, sometimes there will be too many females in the sample,
sometimes too few, but on average, a series of properly drawn samples
will have very close to the same percentage of females as the population
as a whole.
○ A second fundamental premise of the survey research process is that the answers
people give can be used to accurately describe characteristics of the respondents.
The extent to which those answers are not accurate measures is the second
fundamental source of error in surveys.
■ In theory, one could divide what surveys try to measure into two
categories: objective facts and subjective states. Objective facts include a
person’s height, whether or not a person is employed at a job, and whether
or not a person voted in the last election. Subjective states include how
much of the time the person has felt tired and whether or not a person has
liberal or conservative political view.
■ Conceptually, the way we assess the answers to a question is to measure
how well they correspond to the “truth.”
■ Errors can be caused by all kinds of things: misunderstanding the question,
not having the information needed to answer, and distorting answers to
look good are only a few examples. The point is that to the extent that
answers are affected by factors other than the facts on which the answer
should be based, there is error in the answer
■ think of answers as consisting of two components: the true score, what a
perfect reporter with perfect knowledge would give as an answer, plus
some element of error
 x = t + e [x is the answer given by individual i, t is the true value for
individual, e is the error in the answer given by individual]
■ Validity is the term that psychologists use to describe the relationship
between an answer and some measure of the true score. Looking at the
equation, the goal of the psychometrician and the survey methodologist is
to make the error term (e) as small as possible so the answers mainly
reflect the true score.

● In sum, the inference that answers can be used to accurately describe a sample of
respondents and that we can accurately generalize from a sample of respondents to an
entire population, there are two analogous kinds of error: random variability around the
true values and systematic (biased) differences between the sample of respondents and
the whole population or between the answers that are given and the true values for those
who are answering.