Introduces systematic methods for analyzing recorded communication including text, images, audio, and video. Covers manifest and latent content analysis, developing coding schemes, assessing intercoder reliability, and choosing between quantitative and qualitative content analysis traditions.
Develop a coding scheme with operational definitions, code a sample of content, calculate reliability coefficients, refine scheme and retest.
Systematic content analysis is a way of turning text, images, audio, or video into data you can analyze rigorously. The core challenge connects directly to your prerequisite on measurement validity: communication is messy, contextual, and ambiguous, and your goal is to develop a procedure that extracts meaning consistently and without cherry-picking. The word "systematic" is doing real work in the name — it means you define your rules in advance, apply them uniformly to all units, and can defend your coding decisions to a skeptic who reads the same material differently.
The first major distinction is between manifest content and latent content. Manifest content is what is explicitly present — the literal words, visible symbols, stated claims. Latent content is the underlying meaning, tone, or implication that requires interpretive judgment. Counting how many times a newspaper uses the word "crime" is manifest coding; judging whether a story frames a neighborhood as dangerous or vibrant is latent coding. Neither is inherently superior, but they make different epistemological commitments. Manifest coding is easier to replicate and harder to dispute; latent coding captures richer meaning but introduces more interpretive variability. Most serious projects use both.
The backbone of systematic content analysis is the coding scheme: a structured set of categories with operational definitions clear enough that two different coders reading the same document would reach the same classification. Writing a good coding scheme is harder than it sounds. Definitions that seem obvious to you will produce inconsistent results when another coder applies them independently. This is why intercoder reliability testing is non-negotiable — you must check empirically that your categories are being applied consistently before trusting the data they produce. Common reliability statistics include Cohen's kappa (which corrects for chance agreement) and Krippendorff's alpha (which generalizes across multiple coders and measurement levels). High raw agreement (e.g., 85% match) looks reassuring but can be inflated when categories are rare or dominant; kappa corrects for this.
A critical design decision is whether your analysis will be primarily quantitative or qualitative. Quantitative content analysis counts and compares — how often does coverage of immigration emphasize crime versus economic contribution across different news outlets? Qualitative content analysis reads for patterns, tensions, and contextual meanings that resist simple enumeration. Your measurement validity training helps here: quantitative approaches prioritize reliability and generalizability; qualitative approaches prioritize depth and interpretive validity. The choice should follow from your research question, not from convenience. Many research designs benefit from combining both: systematic quantitative counts framed by qualitative reading of exemplary cases.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.