liveBook · Manning

Research

As explained in Effective Data Analysis

Research is the systematic investigation of a topic through study or experimentation to obtain new information.

Research is the investigation of a topic through a study or experiment to gain new information. In analytics, it also means finding and synthesizing existing information before designing that investigation. The two activities reinforce each other: prior knowledge informs a study, and its findings add to what is known. section 3.1.1

Start with the decision the investigation needs to support. Gather enough context to form a question and a testable hypothesis, then choose a way to evaluate it. Reading more is not always the answer: available data may limit what can be asked, and incomplete evidence may require a reasoned choice that others can revisit. section 3.0 section 3.1

Build that context from complementary sources:

  • Synthesize relevant published and public research to understand what is already known. section 3.4
  • Ask stakeholders for domain context and explore the organization’s available data. Check its patterns and limitations before committing to a study design. section 3.4 section 2.1.3
  • Document the sources, gaps, and reasoning behind the question and hypothesis. Give stakeholders enough background to understand why the investigation matters. section 2.3.2 section 3.1

Let the question determine the method. Descriptive analysis summarizes what the data contains; correlation examines a noncausal association; an experiment tests whether a change causes an outcome under controlled conditions. Research may also continue as a program of studies rather than end with one result. section 3.4 section 3.2

From a product question to an experiment

An e-commerce team wants to reduce abandoned shopping carts. It can gather background information, ask stakeholders what they expect from layout changes, and compare three layouts with a control group. Quantified null and alternative hypotheses make the expected outcome explicit before the experiment runs. section 3.1.4

The investigation must also fit its risks. A wearable-device study of potential predictors of hypertension or heart disease would collect sensitive health information. Its plan therefore needs to address informed consent, possible harm, protection against disclosure, and requests to withdraw data. section 8.1.3

How Effective Data Analysis covers it

How deep it goes
Defines two meanings of research and outlines evidence gathering and method choice; defers detailed statistical testing.
Examples
Uses a shopping-cart A/B test and a wearable-device health study to frame evidence gathering and research risks.
What it assumes
Assumes an analytics project and available organizational questions or data; defines research before explaining how to gather background context.
What you can do after
Practice/perform research by synthesizing background evidence, forming a quantified hypothesis, choosing a method, and documenting decisions.
How it connects
Connects research to hypotheses, exploratory data analysis, study design, correlation, experiments, and informed consent.

Also explained in

Compare how 5 books explain research →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Effective Data Analysis. The section links point to the parts of the book it draws on.