liveBook · Manning

Validity

As explained in Effective Data Analysis

Validity is the degree to which a measure accurately represents or supports conclusions about the construct it is intended to measure.

Validity asks whether a measure accurately captures the concept or process it is meant to represent. A metric can be collected consistently yet still miss what stakeholders need to know. Check reliability, then examine validity before using results to support decisions. section 6.3.2 section 6.4

Three checks address different questions:

  • Content or face validity: Does the measure cover the relevant facets of the concept? Review its definition against existing evidence and discuss it with subject matter experts, often including stakeholders. section 6.3.2
  • Construct validity: Does the measure assess the intended theoretical construct? section 6.4
  • Criterion validity: Is the measure associated with other measures of similar constructs? section 6.4

Measuring job satisfaction

A job-satisfaction questionnaire should reflect more than a vague sense of whether work is going well. Relevant facets might include work tasks, environment, manager, colleagues, and compensation. Reviewing research and checking the definition with people familiar with the workplace helps assess whether the questions cover the concept. section 6.3.2

Table 7.2 Many metrics are colloquially discussed as concepts and thus require stakeholder collaboration to create specific operational definitions.
High-level concept
Specific metric
The engineering team is productive
Percentage of tasks completed on time
Customers are supported in their needs
Percentage of customer problems and questions resolved on the first contact (first call resolution, or FCR)
Creating an easy-to-use application
Median time to complete the setup flow for new users

Compare each broad organizational goal with its specific metric. The translation makes a concept measurable, but it also creates a validity question: does that particular metric represent what stakeholders mean by the goal? section 7.2.1

Assess validity during metric design, not only after results arrive. Agree on a clear operational definition with stakeholders, and choose checks suited to the measure: a self-report questionnaire and a behavioral measure may need different validation steps. section 7.2.1 section 6.3.3

How Effective Data Analysis covers it

How deep it goes
Defines validity and distinguishes three assessment types, with qualitative guidance for content or face validity.
Examples
Uses job-satisfaction facets, employee well-being measures, and organizational goals translated into specific metrics.
What it assumes
Introduces validity from its basic definition while assuming familiarity with measures and the concepts they aim to capture.
What you can do after
Evaluate/understand whether proposed measures adequately represent their intended concepts and warrant use in stakeholder decisions.
How it connects
Connects validity to reliability, operational definitions, self-report measures, and behavioral measures.

Also explained in

Compare how 2 books explain validity →

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.