Values Coding 101 (With Examples)
Coding what people value, and when they contradict themselves
The 30-second summary
Values coding labels the values, attitudes and beliefs a participant expresses. It is the technique to reach for when your question is about what drives people, rather than what they did.
- Three things, not one: a value is what matters to someone, an attitude is how they feel about it, and a belief is what they hold to be true.
- Code the reasoning, not the topic. "I stayed because the team needed me" is a value; "team dynamics" is a topic.
- Expect contradictions, and treat them as findings. What people say they value and what they describe doing often differ, and that gap is usually the most interesting thing in the data.
- Watch your own values. This is the technique most exposed to the researcher reading their own priorities into someone else's words.
If your question is about what people did rather than what they cared about, you'll likely want to use process coding instead.
Navigate
Values coding is the technique for labeling what participants care about: the values, attitudes and beliefs sitting underneath what they tell you. If your research question is about motivation, priorities or worldview rather than behavior, this is the technique that will hold your findings together.
It is also the one I see students apply most loosely, because “values” gets used as a catch-all for anything that sounds meaningful. So this post covers what the technique really labels, how to tell its three components apart in a live transcript, what to do when a participant’s values and behavior disagree, how to keep your own assumptions out of the analysis, and when to use something else entirely.
What is values coding?
Values coding is an inductive technique that labels the values, attitudes and beliefs a participant expresses. It comes from Johnny Saldaña’s work on coding methods, and it sits alongside in vivo coding and process coding as one of the standard first-cycle approaches.
The distinguishing feature is that you are coding the reasoning behind what someone says, not the topic they are speaking about. A participant discussing their workload might produce a descriptive code of “workload” and a values code of “professional duty”, from the same sentence. The first tells you what the passage is about; the second tells you what is at stake for them.
Telling the three apart
The technique is named for values, but it labels three related things, and mixing them up is the most common way a values coding frame falls apart. The distinction is worth learning properly because it decides which code you apply.
| What it is | Sounds like | Coded as | |
|---|---|---|---|
| Value | What matters to the person, what they hold as important | “My family comes first, always.” | Family priority |
| Attitude | How they feel or think about something specific | “I find the paperwork demoralizing.” | Frustration with admin |
| Belief | What they hold to be true about the world | “Nobody gets promoted here without knowing someone.” | Promotion is political |
A rough test that works in practice: a value answers what is important to you, an attitude answers how you feel about this, and a belief answers how you think the world works. Values tend to be stable across a transcript; attitudes shift with the topic; beliefs are stated as fact even when they are not.
You do not have to code all three, and many studies deliberately code only values. What matters is that you decide which of the three you are labeling, define it in your codebook, and hold to it. A frame that drifts between the three produces categories nobody can defend.
What it looks like on the page
Here is an extract from an interview with a nurse about staying in a high-pressure unit:
| Interview extract | Code | Type |
|---|---|---|
| “I could earn more somewhere quieter, I know that.” | Money is not the driver | Value |
| “But these are the patients who most need someone.” | Duty to the vulnerable | Value |
| “Management have no idea what happens at 3am.” | Leadership is disconnected | Belief |
| “The rota system is exhausting and it never changes.” | Frustration with scheduling | Attitude |
| “You need the people beside you to be good, or it falls apart.” | Reliance on the team | Value |
Read the code column and you have this person’s reasoning for staying, which is what a question about retention needs. A descriptive pass on the same extract would give “pay”, “patient need”, “management”, “rota” and “teamwork”, which tells you the topics she covered and nothing about why she is still there. If you want to see how one extract looks under several techniques at once, there are worked coding examples side by side.
How to do values coding
Values coding runs in four steps.
- Decide what you are coding for. Values only, or all three? Write the definition into your codebook before you start, because this is the decision that keeps the frame consistent. George Washington University’s guide to the coding process sets out what a codebook entry should contain.
- Read once without coding. Get the shape of the whole dataset. Values are easier to spot on a second pass, once you know what a participant generally cares about.
- Code on a second reading, and record the extract. Always keep the passage alongside the code. Values codes are interpretive, and you will need the evidence when you write up or when an advisor asks how you got there.
- Group into categories that answer your research question. Codes about autonomy, recognition and workload might categorize as “what keeps people here”; that category is only useful if it maps to something you actually asked.
When stated values and behavior disagree
This is the part most guides skip, and in my experience it is where values coding earns its place.
People are not reliable narrators of their own priorities. A participant will tell you that work-life balance matters more than anything, and then describe two weeks in which they worked every evening. That is not a participant lying to you, and it is not bad data.
It is a finding.
Two practical ways to handle it:
- Code both, and mark which is which. A column in your codebook for espoused or enacted costs nothing and turns a contradiction into an analyzable pattern.
- Look for the gap across your sample, not within one person. If eight of twelve participants state a value they later describe not acting on, you have a finding about the setting rather than about individuals, which is almost always the more defensible claim.
What you should not do is quietly resolve the contradiction by picking whichever statement fits your argument. That is where values coding most often goes wrong, and an examiner reading your appendix will see it.
Keeping your own values out of it
Values coding is the technique most exposed to the researcher’s own priorities, because you are inferring what matters to someone from what they said. In vivo coding constrains you to their words; process coding constrains you to their actions. This one asks you to interpret, which means your own frame is in the room.
Three habits that help, and they are a specific case of guarding against research bias more generally:
- Write down what you expect to find before you code. If you expect nurses to be motivated by vocation, note it. You will code differently knowing you had that expectation.
- Keep the extract beside the code, always. If you cannot point at the words that produced a values code, you interpreted rather than coded.
- Have someone else code a transcript. Even one, even informally. Where you disagree is where your assumptions live, and disagreement about a values code is usually more informative than agreement.
None of that is bureaucratic box-ticking. It is what lets you write a defensible reflexivity paragraph rather than a generic one.
Reading your own priorities into someone else’s words is a form of analysis bias, and this explainer covers it alongside the other kinds worth watching for.
When not to use values coding
Values coding is a poor default. It asks you to interpret, which means that when there is little reasoning in the data to interpret, it will produce a thin and strained set of codes rather than telling you so.
- When your question is about what people did. Actions and sequences are process coding territory. Coding a description of a workflow for values produces a thin, strained set of codes.
- When the language itself is the point. If you want participants’ own framing preserved, in vivo coding does that and values coding will paraphrase it away.
- When your data is thin on reasoning. Short survey responses and factual interviews often contain no expressed values at all. Forcing the technique onto them yields codes you inferred rather than found.
- When you are testing a values framework. Applying an existing model of values, such as Schwartz’s, means your codes come from the model rather than the data. That is deductive work and should be described as such; see inductive vs deductive coding.
In most of those cases the answer is not to drop values coding entirely. It is to use it on the parts of your data where people explain themselves, and something else on the rest. A study that codes its reasoning passages for values and its process passages for action will usually beat one that forces a single technique across everything.
Still have questions?
What is the difference between values coding and thematic analysis?
They operate at different stages. Values coding is a first-cycle coding technique that labels individual extracts; thematic analysis is the broader process of building themes from codes and interpreting them. You would typically use values coding to produce the codes and then run thematic analysis over them. They are not alternatives.
Can a single extract carry more than one values code?
Yes, and it often should. “I stayed because the team needed me, even though the money was better elsewhere” carries a value about duty and a value about money not being the driver. Splitting them gives you two codes that will each group with others; forcing one loses half the meaning.
How do I know a values code is not just my own interpretation?
You do not, entirely, which is why the evidence trail matters. The working test: could you show the extract to someone else and have them arrive at a similar code without you explaining it? If the code only makes sense with your commentary attached, you have interpreted too far.
Do values codes have to use the participant’s own words?
No. Using their exact words is in vivo coding, which is a different technique and can be combined with this one. Values codes are usually your own short labels: an interpretation of what the person conveyed, not a quotation of it.
What if a participant never states a value directly?
Most do not state them directly, and that is normal. Values usually surface in the reasons people give for choices, in what they compare things to, and in what they treat as not needing justification. If a whole transcript yields nothing, that is worth noting rather than forcing: not every interview is about what someone values.
Can’t find your answer here? Ask a Grad Coach directly – the initial chat is free.
Speak with a friendly coach →
