Chapter 7
7. Data Processing and Analysis
7.1. Coding, editing and cleaning the data
7.2. Data analysis
7.3. Concept of Economic Modelling
7.4. Testing hypothesis: Falsification as a scientific approach
231
Data Processing and Analysis
• Numbers and fragmented notes are meaningless unless we can find the patterns that lie
beneath them.
• Data have to be cleaned, organized and analyzed in a meaningful way.
• The raw data need to be subjected to a number of procedures >>> data processing
• How the researcher prepares the data for inspection or interpretation will affect the
meaning that those data reveal.
• Therefore, every researcher should be able to provide a clear, logical rationale for the
procedures used to arrange and organize the data.
• These procedures are similar whether your study is quantitative or qualitative, but what
you do within each procedure is different.
• In quantitative research, we try to make better sense of the world through measurement
and numbers. The main emphasis in data analysis is to decide how you are going to
analyze information obtained in response to each question that you asked of your
respondents
• In qualitative research, the focus is the contents, discourse, narrative or event analysis of
the data, focusing on the main theme.
232
Data processing in quantitative studies
• Before using any statistical analysis, look closely at your data and
✓ Edit and clean the data for any errors, outliers and incompleteness.
✓ Consider potential productive ways of organizing them.
✓ Using an open mind and your imagination, look for patterns in the numbers.
• Errors are almost unavoidable. We may make errors during designing
questionnaires, conducting the interview, data capture, etc.
• These errors should be assessed and corrective measures taken if
necessary.
• To understand the pattern, identify errors and to organize and interpret the
data better, inspect the data using various ways including
✓ graphs and tables
✓Frequency distributions
233
Data processing in quantitative studies
A. Editing - consists of scrutinizing the completed research instruments to
identify and minimize errors, incompleteness, misclassification and gaps in
the information obtained from the respondents.
we may
• forget to ask a question;
• forget to record a response;
• wrongly classify a response;
• write only half a response;
• Write responses illegibly (less clearly)
• Recording error values e.g., recording 100 to input 10 or 0.001 instead of
0.1); mixing units (quintal/kg/g, hectare/gezim/kada, Birr/Dollar/Euro)
234
Data processing in quantitative studies
• Using software substantially eases data inspection and editing.
• Check for maximum and minimum values and make sure that it is
consistent with the criteria used to select samples.
• Check the distribution of the data to decide the method of data
analysis.
235
Data processing in quantitative studies
A. Editing –
• Hence, it is essential to carefully examine the data, and edit it, by:
i. Inferencing other question in the questionnaire - Certain questions in a research
instrument may be related to one another and it might be possible to find out the
answer to one question from the answer to another. E.g., income information may be
inferred from expense, saving, donation & gift. However, you must be careful about
making such inferences or you may introduce new errors into the data.
ii. By recall – If the data is collected by means of interviews, sometimes it might be
possible for the interviewer to recall a respondent’s answers. Again, you must be
extremely careful.
iii. By going back to the respondent – If the data has been collected by means of
interviews or the questionnaires contain some identifying information, it is possible to
visit or phone a respondent to confirm or ascertain an answer. This is, of course,
expensive and time consuming.
iv. Using statistical methods – usually we replace some percentage (depending on
sample size & level of errors, 1% to 5% from each side) of the lowest and highest
values by median or mean value. Replacing is better than trimming. We sometimes
also impute missing values. But, we have to be careful not to make errors.
236
Data processing in quantitative studies
B. Coding and content analysis
• Content analysis (descriptive info) - a process whereby you identify the main
themes that emerge from the descriptions given by respondents in answer to
questions.
➢ After identifying the main themes, either
1. Group the responses to either in support or contradiction of your arguments
2. Assign a code to each theme and count how frequently each has occurred
3. combine both methods to communicate your findings
Coding (for quantitative & categorical info) is simplified way of recording
participants’ responses into numerical data which you can statistically analyze.
237
Data processing
B. Coding and content analysis
• The method of coding is largely dictated by two considerations:
1. The way a variable has been measured (measurement scale) in the research
instruments (e.g., if a response to a question is descriptive, categorical or
quantitative);
2. The way we plan to communicate the findings about a variables.
• Steps in coding:
A. Developing a code book - provides a set of rules for assigning numerical values
to answers obtained from respondents. E.g., Female = 1, male = 0; married =1,
single =2, divorced =3, separated = 4, widowed = 5
B. Pre-testing the code book - selecting a few questions and actually coding the
responses to ascertain any problems in coding
C. Coding the data – code the raw data either on the questionnaire, separate
coding sheet or computer programs (e.g., STATA, SPSS)
D. Verifying the coded data for any discrepancies b/n the coded data and the raw
data
238
7.2. Data analysis
• Data analysis - a process of cleaning, transforming, and modeling data to
discover useful information for various purposes.
• One of the key activities in data analysis is to develop a frame of analysis –
• Although a framework of analysis needs to evolve continuously while
writing your report, it is desirable to broadly develop it before analyzing the
data
• A frame of analysis should specify:
➢ which variables you are planning to analyze;
➢ how they should be analyzed;
➢ what cross-tabulations you need to work out;
➢ which variables you need to combine to construct your major concepts or to
develop indices
➢ which variables are to be subjected to which statistical procedures
239
7.2. Data analysis
• Details the main findings
• Provides a summary explanation of results
• Accept or reject hypotheses if you have any
• Depending on various issues, you may analyze your data using various ways including:
✓ Frequency distributions
✓ Cross-tabulations - analyze two variables, usually independent and dependent or
attribute and dependent, to determine if there is a relationship between them.
✓ Statistical analysis including,
A. Measures of central tendency
B. Measures of dispersions
C. Measures of Relationship - correlations and Analysis of variance (ANOVA),
D. Statistical tests
E. Regression analyses
240
7.2. Data analysis
Which measure to use?
241
7.2. Data analysis
Which measure to use?
242
7.2. Data analysis
Which measure to use?
243
7.2. Data analysis
Which measure to use?
244
7.2. Data analysis
Which measure to use?
245
Some of regression models
• Linear models, e.g., OLS >>> continuous dependent variable
• Binary Response Models: Probit/logit >>>> binary/dichotomous
variable
• Corner Solution Responses: e.g., Tobit, Hurdle Model >>>>> censored
dependent variable
• Count models: Poisson, Negative Binomial, model
• Duration Analysis
246
7.3. Concept of Economic Modelling
• A model is an attempt to describe observed outcomes.
• It is a physical, symbolic, visual, mathematical, or graphical representation of a concept
or a theory developed to clarify or make it simpler to understand
• An economic model is a simplified, often mathematical or graphical, framework
designed to illustrate complex processes.
• E.g. the most common abstract model used in economics is that of ‘maximizing
behavior’.
• Its main purpose is to present a complex process in a simplified version using variables in
a logical way.
• A modeler aspires to achieve three objectives:
1. to draw a clean line from cause to effect,
2. to describe observed outcomes with fidelity (faithfully), and
3. to do so across a panoply (array) of situations, places, and times so that it is
generalizable. 247
7.3. Concept of Economic Modelling
• Two general categories of economic models by purpose:
1. Pure or abstract models which are simplified representations of the underlying
logic of the theory being modelled,
2. Applied models which are explicit, simplified representations of more general
theories and which are designed to be applied to specific real-world problems or
situations.
• Abstract models are intended to be true for all cases for which they apply
but specific models may be true only for the case represented.
• Usually, there is an abstract model underlying every applied model.
• Economic processes are so complex where numerous factors interact,
affecting and being affected by decisions.
248
7.3. Concept of Economic Modelling
• Diversity of factors determine economic activity; these factors include:
• Individual and cooperative decision processes,
• Resource limitations,
• Environmental and geographical constraints,
• Institutional and legal requirements
• Geo-politics
• Purely random fluctuations
• Economists therefore must make a reasoned choice of
• Which variables and which relationships between these variables are relevant, and
• which ways of analyzing and presenting this information are useful.
249
7.3. Concept of Economic Modelling
• In addition to their professional academic interest, uses of models include:
• Forecasting economic activity in a way in which conclusions are logically related to
assumptions;
• Proposing economic policy to modify future economic activity;
• Presenting reasoned arguments to politically justify economic policy at the national
level,
• to explain and influence company strategy at the level of the firm,
• to provide intelligent advice for household economic decisions at the level of
households.
• Planning and allocation
• In finance, predictive models have been used since the 1980s for trading
250
7.3. Concept of Economic Modelling
• There are various classifications of economic models:
• General equilibrium model, a partial equilibrium model, or a non-equilibrium model;
• Stochastic or non-stochastic
• Discrete or continuous choice model
• rational agent models or representative agent models
• Qualitative or quantitative
251
7.4. Testing hypothesis: Falsification as a scientific approach
• Test the Falsifiability of each of your null hypotheses by choosing significance level.
• Falsifiability refers to the notion that a theory or statement can be found to be false.
• The Falsification Principle, proposed by Karl Popper, is a way of demarcating science
from non-science.
• It suggests that for a theory to be considered scientific it must be able to be tested
and conceivably proven false.
• All economic models or theories, if they are going to be given serious consideration
by practicing economists, must be shown to be testable where a successful test of a
theory is defined as a falsification of that theory.
• A testable theory is a falsifiable theory with existing technology.
• However, falsifiability does not mean verifiability. Verifiability requires that we check
all aspects and scenarios about the theory, which could be impractical.
• For instance, verifying a statement “all economics students are better in quantitative
skill than other social science students” requires testing and comparing all students.
252
7.4. Testing hypothesis: Falsification as a scientific approach
• The establishment of falsifiability as an operative demarcation criterion directly excludes
existential statements (expresses the existence of at least one object) and tautologies
(always true and thus can never be falsified).
• Falsification is appealing because it tells a simple and optimistic story of scientific
progress, that by steadily eliminating false theories we can eventually arrive at true
ones.
• A false theory or model must surely be rejected regardless of how ‘useful’ it may be. The
discovery of even one fact which contradicts one of a theory’s implications constitutes a
refutation of that theory.
• Different ways are used by statisticians to draw conclusions about hypotheses on the
basis of available evidence.
• Fisher, Neyman and Pearson proposed approaches that require no prior probabilities
(probabilities using current knowledge before new data are collected) on the hypotheses
that are being studied.
• In contrast, Bayesian inference emphasizes the importance of prior probabilities 253