Notes on Methods of Data Collection, Organization, and Presentation

Learning Objectives

• Identify the different methods of data organization and presentation

• Understand the criterion for the selection of a method to organize and present data

• Identify the different methods of data collection and criterion that we use to select a method of data collection

• Define a questionnaire, identify the different parts of a questionnaire and indicate the procedures to prepare a questionnaire

Introduction

• Before any statistical work can be done, data must be collected. Depending on the type of variable and the objective of the study, different data collection methods can be employed.

• Data collection techniques allow us to systematically collect data about our objects of study (people, objects, and phenomena) and about the setting in which they occur. In data collection, being systematic is essential; haphazard collection makes it hard to answer research questions conclusively.

Data Collection Methods

• Data collection techniques aim to collect data about objects and their setting in a systematic way.

• Common techniques include:
  1) Observation
  2) Interviews and self-administered questionnaires
  3) Use of documentary sources
  4) Other methods (implied by context, e.g., high-level measurements and equipment)

1. Observation

• Involves systematically selecting, watching, and recording behaviors or phenomena to obtain specified information.

• Ranges from simple visual observations to the use of sophisticated equipment (radiography, biochemical tools, x-ray, microscope, etc.).

• Advantages:

  • Gives relatively more accurate data on behavior and activities.

• Disadvantages:

  • Observer biases, prejudices, desires, etc.

  • Requires more resources and skilled personnel when high-level equipment is used.

2. Interviews and self-administered questionnaires

• Among the most commonly used data collection techniques.

• Preliminary design considerations (before designing):

  • What exactly do we want to know, based on objectives and variables?

  • Is questioning the right technique, or do we need additional techniques (observations or records analysis)?

  • Are informants literate? If illiterate, self-administered questionnaires are not an option.

  • How large is the sample to be interviewed?

Interview types

• Decision on structure: interviews may be unstructured or more standardized.

• Unstructured interview:

  • Flexible; content, wording, and order of questions vary across interviews.

  • Investigators know what they want to learn but don’t predefine exact questions/order.

• Standardized methods:

  • Wording and order of questions are decided in advance.

  • Forms include highly structured interviews or self-administered questionnaires (respondent reads questions and fills answers; interviewer may assist).

• Standardized methods are preferred in community medicine research for reproducibility.

• Less-structured interviews are useful in preliminary surveys, perceptions, attitudes, motivation, or when exploring topics qualitatively.

• Self-administered questionnaires advantages: simpler, cheaper, can reach many people simultaneously; can be mailed.

• Disadvantages: response depends on respondent education/skills; lower response rates among less literate populations.

Face-to-face vs. telephone interviews

• Face-to-face and telephone interviews can stimulate and maintain respondent interest, build rapport, and help address anxiety.

• Interviewers can repeat unclear questions, provide clarifications, or rephrase as needed.

• Follow-up or probing questions are easier to use in interviews than in self-administered formats.

• Generally, interviews are preferred over self-administered questionnaires, provided interviewers are skilled.

Mailed Questionnaires

• The investigator prepares a questionnaire and sends it by post with a covering letter explaining aims and requesting cooperation.

• Return postage is typically paid by the investigator to ensure quick responses.

• Main problems: relatively low response rates and under-representation of less literate subjects.

Use of Documentary Sources

• Involves personal or clinical records, death certificates, published mortality statistics, census publications, etc.

• Examples include:

  • Official publications of statistical authorities

  • Publications from health ministries

  • Newspapers and journals

  • International publications (WHO, World Bank, UNICEF)

  • Hospital or health institution records

• Caution: while data from documents can be less time-consuming and lower cost, pay attention to data quality and completeness; there may be differences in objectives between data authors and users.

Problems in Gathering Data

• Language barriers

• Lack of adequate time

• Expense

• Inadequately trained and experienced staff

• Invasion of privacy

• Suspicion

• Bias (spatial, project, person, season, diplomatic, professional)

• Cultural norms

Choosing a Method of Data Collection

• Decision-makers need information that is relevant, timely, accurate, and usable. Data gathering can be costly, so methods should be chosen to balance timeliness, cost, accuracy, and scientific rigor.

• Data sources can be categorized as:
  1) Primary data
  2) Secondary data

1) Primary data

• Collected by the investigator for a specific study.

• Original data; generated by surveys or field work.

• Advantages:

  • More reliable and accurate because the investigator can obtain clarifications on difficult questions, remove doubts, and ensure correct interpretation.

  • Potentially higher response rates since questions are asked directly and on the spot.

• Disadvantages: time-consuming and potentially expensive.

2) Secondary data

• Data collected by others, which may be primary for the collector but secondary for the user.

• Sources: journals, reports, government publications, professionals, and research organizations.

• Advantages:

  • Less expensive in time and money.

  • May be of high quality if collected by trained personnel.

• Disadvantages: must be used with care; data may be biased toward different research purposes and may contain errors.

General considerations in method choice

• The choice is largely based on the accuracy (and relevance) of information yielded by the method.

• Practical considerations include:
  1) Need for personnel, skills, equipment, etc., given availability and urgency of results.
  2) Acceptability of procedures to subjects (no undue inconvenience or harm).
  3) Likelihood of good coverage of the population (if many people don’t know the answer, the question is inappropriate).

• Investigator familiarity with procedures can influence method selection (often researchers use techniques they are skilled in).

Types of Questions

Open-ended questions

• Permit free responses in the respondent’s own words.

• Useful for:

  • Facts the researcher isn’t familiar with

  • Opinions, attitudes, and suggestions

  • Sensitive issues

• Example prompts: e.g., describe what a traditional birth attendant did when labor started; reasons for high drop-out rates in a village health committee.

Closed questions

• Provide a list of options from which respondents choose.

• Design goals:

  • Options should be exhaustive and mutually exclusive

  • Keep the number of options small

• Examples:

  • What is your marital status? - single, married, live-in, separated, widow

  • Do you visit the local village health worker for treatment? - Yes, No

Using closed questions for opinions and scales

• Closed questions can also capture opinions using a rating scale, e.g.:

  • How useful have the Village Health Committee activities been?

  • Extremely useful, Very useful, Useful, Not very useful, Not useful at all

Requirements of Questions

• Must have face validity

• Must be clear and unambiguous

• Must not be offensive

• Questions should be fair

• Sensitive questions should be approached carefully and wisely

Steps in Designing a Questionnaire

• Designing a good questionnaire typically involves several drafts:

  • Draft 1: focus on content

  • Draft 2: critically examine formulation and sequencing

  • Draft 3: scrutinize formatting

  • Final: conduct a test run to see if it yields the required information and if respondents feel comfortable

  • Often further adaptations are necessary before actual data collection

Step 1: Content

• Start from objectives and variables; decide which questions measure or define the variables and help achieve objectives.

• Revisit variables as you draft; you may add, drop, or change objectives.

Step 2: Formulating Questions

• Formulate one or more questions to obtain information for each variable.

• Ensure questions are specific and precise to minimize differential interpretation among respondents.

• Example problem:

  • A general question like “Where do community members usually seek treatment when they are sick?” is too vague because respondents may think of different diseases or levels of care.

Step 2: Avoiding multiple ideas in one question

• Each question should measure one concept to avoid confusion.

• Example: instead of asking about preferred birth interval for both partners in one question, split into two questions to capture each partner’s opinion.

Step 2: Avoid leading questions

• Leading questions suggest a preferred answer. Examples and corrections:

  • Leading: "Do you think the district health team should visit each health center monthly?"

  • Non-leading: "How often should the district health team visit each health center? If you say yes, how often?"

• Revised examples targeting treatment or actions should avoid implying a specific choice.

Step 3: Sequencing of Questions

• Design the questionnaire to be consumer-friendly:

  • Logic should be intuitive for respondents and allow natural discussion, even in structured interviews.

  • Start with minimal background questions; place sensitive questions late in the interview if possible.

• Start with an interesting but non-controversial open question related to the subject to engage respondents and reduce suspicion.

• Place more sensitive questions (income, sexual behavior, stigmatized diseases) as late as possible.

• Use simple, everyday language; keep the questionnaire as short as possible; consider dividing into two parts if it would take more than an hour.

Step 4: Formatting the Questionnaire

• Final questionnaire should include: a heading, space for interview number, date, location, informant name (if needed); interviewer name for quality control.

• Layout should group related questions together visually; use subheadings if long.

• Provide adequate space for open-ended answers.

• Pre-categorized answer boxes should be aligned consistently across the questionnaire.

• Ensure the questionnaire is both consumer-friendly and user-friendly.

Step 5: Translation

• If interviews will be conducted in multiple local languages, translate the questionnaire to standardize wording.

• Translate back to the original language and compare versions to resolve difficult concepts before finalizing.

Methods of Data Organization and Presentation

• Data collected in a survey are raw data. Useful information is not immediately evident, so data must be organized to reveal patterns of variation.

• Different techniques of data organization and presentation include ordered arrays, tables, and diagrams.

• The primary objective is to condense information to reveal patterns and to decide precise analysis methods based on data characteristics.

Frequency Distribution and Its Forms

• A frequency distribution arranges data into a table or graphical form to show patterns.

• It is especially useful for voluminous data (e.g., health center records) to summarize information meaningfully.

Array (orderly array) and Simple Frequency Distribution

• An array orders numerical data in ascending or descending order.

• Useful when data size is small (usually n < 20).

• Example data: time spent watching full-length movies (example data with numbers such as 0, 1, 2, 3, 4, 5, 6, 7, 8) with corresponding frequencies and relative frequencies (example values: 0 → 72 occurrences, 1 → 106, 2 → 153, …, total 400).

• Relative frequency formula (example):
  $f_r = rac{f}{N} imes 100 ext{?}<br>$
  (Where f is the class frequency and N is the total number of observations. In the example, relative frequencies are given as percentages summing to 100%.)

Categorical Distribution

• Non-numerical information represented as frequencies. Example: plans after high school for 548 seniors (e.g., attend college, may attend college, vocational school, not attend).

Grouped Frequency Distribution

• For large data sets, group data into several class intervals.

• Example age distribution (large counts): Under 18, 18-24, 25-34, 35-44, 45-54, 55 and over (with corresponding totals).

• Advantages: provides an overall picture; concise summary of a large dataset.

• Limitation: grouped distribution may mask individual values (e.g., ages within a class cannot be identified).

Construction of Grouped Frequency Distributions: Four Steps

1) Choosing the classes (number and width)
2) Sorting/tallying data into classes
3) Counting items in each class
4) Displaying results as a chart or table

Choosing the number of classes (k)

• Rule of thumb (Sturges):
  $k = 1 + 3.322 \, ext{log}(n)$
  where n is the number of observations.

• Class width (w):
  $w = rac{ ext{Maximum value} - ext{Minimum value}}{k}$

• Practical notes:

  • Avoid open-ended classes; ensure every item falls into exactly one class (mutually exclusive).

  • Start points should place frequencies near the middle of class intervals for interpretability.

• Note: Sturges’ rule is a guideline, not a hard rule; adjust k for clarity.

Class limits and boundaries

• Class limits should be definite and clearly stated; avoid open-ended classes (e.g., "less than 10" or "greater than 65").

• Lower limit of the first class should be chosen to concentrate the frequency near the middle of the class interval.

• True class limits (class boundaries) ensure a continuous variable has no gaps between classes.

• Mid-point (class mark) of an interval:
  $X<em>c = rac{L</em>{ ext{lower}} + U_{ ext{upper}}}{2}$

• Width of a class from true limits example: for a class with true limits 39.5 to 49.5, the width is:
  $w = 49.5 - 39.5 = 10$

Example: class-interval widths and true limits

• If weights are given to the nearest kilogram, a class might be 10-14; if given to the nearest tenth, 10.0-14.9; if to the nearest hundredth, 10.00-14.99, and so on.

Example: Constructed grouped frequency distribution

• Given data of time (hours) spent on leisure by 80 college students, with a range from 10 to 38 hours, the number of classes is approximately 7 using Sturges’ rule, and class width computed as $w = \frac{38 - 10}{7} \approx 4$.

• The table would include Time intervals (e.g., 10-14, 15-19, 20-24, 25-29, 30-34, 35-39, 40-44), tally, frequency, and cumulative frequency.

• Example cumulative frequency structure:
  Time spent (h) | Tally | Frequency | Cumulative Frequency
  10-14 | IIIII III | 8 | 8
  15-19 | IIIII IIIII … | 28 | 36
  20-24 | IIIII IIIII … | 27 | 63
  25-29 | IIIII IIIII … | 12 | 75
  30-34 | IIII | 4 | 79
  35-39 | I | 1 | 80

Small vs. Grouped (examples and notes)

• Small sample array: direct frequency counts with exact values and possibly midpoints used for a quick view.

• Grouped distribution: useful for large datasets; requires careful selection of class intervals and boundaries to avoid misinterpretation.

Cumulative and Relative Frequencies

• Cumulative frequencies sum frequencies up to a given value (Less than cumulative) or from above a value (More than cumulative).

• Relative frequencies express class frequencies as a percentage of the total:
  $f_r = \frac{f}{N} \times 100\%$

Mid-Point and Class Boundaries (True Limits)

• Mid-point (class mark) of an interval:
  $X<em>c = \frac{L</em>{ ext{lower}} + U_{ ext{upper}}}{2}$

• True class limits (boundaries) define a continuous interval without gaps between classes.

• Example: For a class with true limits 9.5-19.5, the mid-point is 14.5 and class width is 10.0.

Diagrams and Diagrammatic Representation of Data

• Diagrams provide quick, intuitive understanding and enable rapid comparisons.

• Types of diagrams:

  • Bar charts and pie charts: qualitative or discrete data

  • Histograms and frequency polygons: quantitative, continuous data

• General rules for graphs:
  1) Each graph should be self-explanatory and as simple as possible
  2) Titles should indicate what, where, when, and how classified; titles usually placed below the graph
  3) Include a legend or key if more than one variable is shown
  4) Axes should be labeled clearly; read from the left (vertical axis) and bottom (horizontal axis)
  5) Scaling units should be indicated clearly
  6) Frequency scales should start at zero or include a break in the axis

Bar Charts

• Simple bar chart: one-dimensional; each bar represents the frequency or magnitude; all bars have equal width and spacing.

• Multiple bar chart: compares more than one variable with adjacent bars for each category.

• Component (subdivided) bar diagrams: totals built from two or more components; can be either actual component bars or percentage component bars (each bar height represents the total, with segments showing proportions).

Pie Chart

• Circle divided into sectors with areas proportional to frequencies; suitable for qualitative or discrete data.

Histogram

• For quantitative continuous data.

• Construction principles:

  • Horizontal axis is a continuous scale (no label of the variable name and units on the axis itself)

  • For each class, draw a vertical rectangle with the base spanning from one class boundary to the next; no gaps between rectangles

  • Rectangle widths equal to class interval width; if class intervals are unequal, adjust the rectangle heights to reflect class width

• Example: Time spent on leisure activities by 80 college students can be displayed as a histogram using class intervals (e.g., 10-14, 15-19, etc.)

Frequency Polygon

• Constructed by joining the midpoints of the tops of the histogram rectangles with straight lines; the polygon can extend to the x-axis beyond the range of data.

• An alternative to a histogram for visualizing a frequency distribution.

Ogive (Cumulative Frequency Curve)

• Construct by plotting cumulative frequency against the true upper class limits (class boundaries) on the x-axis.

• The curve starts at the lowest class with cumulative frequency of 0 and ends at the maximum with total frequency N.

Line Diagram (Line Graph)

• Useful for studying variables over time; time is on the horizontal axis, the measured quantity on the vertical axis; points connected by lines show trends over consecutive periods (weeks, months, years).

Examples and Practical Details of Diagrams

• Diagram examples illustrate: bar charts for discrete categories, multiple/bar charts for comparing groups, component bar diagrams for parts of totals, pie charts for proportionate shares, histograms for continuous distributions, frequency polygons, ogives, and line diagrams for trends over time.

• Practical notes:

  • Ensure graphs are not overloaded with too many series; use legends when necessary.

  • When presenting grouped or stack diagrams, clearly indicate which category corresponds to which color or pattern.

Illustrative Examples (Summary of Given Data in the Transcript)

• Simple frequency example: distribution of the number of movies watched in a week among 400 people with frequencies for 0–8 movies and a reported relative frequency for each class.

• Categorical distribution example: plans after high school for 548 seniors with counts for college, may attend college, vocational school, and not attending.

• Age distribution example: grouped age intervals with counts (Under 18, 18-24, 25-34, 35-44, 45-54, 55 and over) totaling 10,392.

• Weights example: grouped weights with class limits and midpoints, illustrating width and true limits, and the need for consistent class intervals for continuous data.

• Immunization data examples: simple one-way table and two-way table showing immunization status by marital status with denominators (row totals) and the use of relative frequencies.

Practical Formula References from the Transcript

• Number of classes (Sturges rule):
  $k = 1 + 3.322 \log(n)$

• Class width (for grouped data):
  $w = \frac{\text{Maximum value} - \text{Minimum value}}{K}$

• Mid-point of a class interval (class mark):
  $X<em>c = \frac{L</em>{ ext{lower}} + U_{ ext{upper}}}{2}$

• Relative frequency (as a percentage):
  $f_r = \frac{f}{N} \times 100\%$

• Cumulative frequency concepts:

  • Less than cumulative frequency: sum of frequencies up to a given upper bound

  • More than cumulative frequency: sum of frequencies from a given lower bound upward

• True class boundaries and class limits: ensure no gaps between classes and that each item belongs to one class; open-ended classes should be avoided for calculation purposes

Notes on Data Quality and Application

• The choice of data collection and presentation methods should consider:

  • The objective and required precision

  • The cost and practicality of data collection efforts

  • The need for timely and actionable information

  • The potential biases and limitations of each method (e.g., biases in observation, response biases in surveys, data quality in documentary sources)

• When presenting data, balance clarity with completeness: provide enough detail to support conclusions, but avoid overwhelming the reader with unnecessary minutiae.

Summary of Key Concepts

• Systematic data collection is essential for credible research.

• Data collection methods include observation, interviews, self-administered questionnaires, postal questionnaires, and documentary sources.

• Questionnaire design is a structured, multi-step process focusing on content, wording, sequencing, formatting, translation, and testing.

• Data organization and presentation transform raw data into meaningful patterns via frequency distributions, class intervals, and various diagrammatic forms.

• Statistical tables and graphical representations (bars, pies, histograms, ogives, line diagrams) each have appropriate uses and limitations.

• Practical considerations (cost, accuracy, coverage, respondent acceptability) influence method choice.

• Formulas to remember: $k = 1 + 3.322 \log(n)$, $w = \frac{\text{Max} - \text{Min}}{K}$, $X<em>c = \frac{L</em>{\text{lower}} + U<em>{\text{upper}}}{2}$, and $f</em>r = \frac{f}{N} \times 100\%.$