Stack Overflow Developer Survey 2026
What makes work context hard to get by attitude toward ai
In the Very unfavorable group, 25% selected “AI tools use the wrong context”; in the Very favorable group, 14% did. Of the 15 answers tested, “AI tools use the wrong context” differs most by attitude toward ai, a negligible association: χ²(5, N = 10164) = 67.88, Holm-adjusted p < .001, V = .08. Respondents could select more than one answer, so the percentages add to more than 100%. These are unweighted respondents, not population estimates.
How results are checked
Named statistical review pending. Selected automated reference cases are published on the validation page.The table
The table
Question asked: Which of the following make it difficult to get the context you need for work? Select all that apply.
Respondents could select more than one answer. Shown by attitude toward ai, as the percentage of each group's respondents who selected each answer, with the number of respondents beneath. A respondent is counted in every row they selected, so a column adds to more than 100%.
| What makes work context hard to get | Very favorable (A) | Somewhat favorable (B) | Neutral (C) | Somewhat unfavorable (D) | Very unfavorable (E) | Not sure (F) | All |
|---|---|---|---|---|---|---|---|
| Information is incomplete | 60.6%1,999 | 66.4%A2,061 | 63.4%938 | 65.4%A974 | 65.2%464 | 61.4%51 | 63.8%6,487 |
| Important context lives in people’s heads | 58.6%1,931 | 64.2%A1,990 | 61.6%911 | 67.4%ACE1,003 | 59.8%426 | 60.2%50 | 62.1%6,311 |
| Information is outdated | 55.6%1,833 | 59.5%A1,847 | 56.1%830 | 60.4%A899 | 61.9%Af441 | 45.8%38 | 57.9%5,888 |
| Information is spread across too many tools | 54.5%1,799 | 57.4%1,779 | 53.7%795 | 57.8%861 | 53.4%380 | 54.2%45 | 55.7%5,659 |
| I do not know where to look | 29.2%964 | 34.1%A1,058 | 30.9%457 | 32.6%485 | 31.3%223 | 38.6%32 | 31.7%3,219 |
| Important decisions are hard to find later | 25.0%824 | 27.5%854 | 26.7%395 | 29.7%A442 | 27.2%194 | 42.2%ACbe35 | 27.0%2,744 |
| Different teams have different versions of the truth | 25.1%828 | 25.5%791 | 27.0%400 | 27.9%416 | 25.7%183 | 36.1%30 | 26.1%2,648 |
| Search does not return useful results | 19.2%633 | 23.7%A734 | 23.8%A352 | 28.5%ABC425 | 25.3%A180 | 24.1%20 | 23.1%2,344 |
| AI tools do not have enough context | 23.1%763 | 24.3%CD754 | 19.9%294 | 20.3%303 | 21.1%150 | 25.3%21 | 22.5%2,285 |
| I do not know which source is authoritative | 16.6%549 | 19.1%591 | 18.2%270 | 20.6%A306 | 21.1%a150 | 19.3%16 | 18.5%1,882 |
| AI tools use the wrong context | 14.3%472 | 19.2%A596 | 18.5%A274 | 21.8%A324 | 24.6%ABC175 | 20.5%17 | 18.3%1,858 |
| Access or permissions prevent me from seeing what I need | 15.8%520 | 16.4%509 | 15.8%234 | 17.1%254 | 15.4%110 | 20.5%17 | 16.2%1,644 |
| I usually have the context I need | 11.7%BCDE386 | 7.7%239 | 7.0%103 | 8.8%131 | 6.3%45 | 12.0%10 | 9.0%914 |
| I intentionally limit context because of privacy or security concerns | 5.9%194 | 6.8%212 | 7.3%108 | 7.5%112 | 5.9%42 | 3.6%3 | 6.6%671 |
| Not applicable | 2.0%67 | 1.8%55 | 1.6%24 | 1.7%25 | 1.8%13 | 6.0%bcd5 | 1.9%189 |
| Other, please specify | 1.2%39 | 0.7%23 | 1.2%18 | 1.5%23 | 2.4%B17 | 1.2%1 | 1.2%121 |
| Respondents | 3,298 | 3,102 | 1,480 | 1,489 | 712 | 83 | 10,164 |
A letter beside a percentage names the columns it is significantly higher than (uppercase 95%, lowercase 90%; pairwise column tests, Bonferroni-corrected). 20,739 of 30,903 respondents have no answer to one of the two questions (not asked or not answered) and are left out.
Analysis
Results
Chi-square test of “AI tools use the wrong context”
χ² = 67.88
df = 5, Holm-adjusted p < .001
Effect size
Cramér's V = 0.082
a negligible association
The strongest answer, APA-style: χ²(5, N = 10164) = 67.88, Holm-adjusted p < .001, V = .08. Each answer is tested on its own, as the respondents who selected it against those who did not, across the 6 groups (N = 10,164). The p-values are Holm-adjusted for the 15 answers tested; 9 of them differ by attitude toward ai at the 0.05 level. 1 answer is not tested: too few respondents selected it for the test to be reliable (an expected count below 1, or more than 20% of the expected counts below 5).
| Answer | Selected | χ² (df = 5) | Adjusted p | Cramér's V |
|---|---|---|---|---|
| Information is incomplete | 63.8% | 26.45 | < .001 | 0.051 |
| Important context lives in people’s heads | 62.1% | 42.57 | < .001 | 0.065 |
| Information is outdated | 57.9% | 26.24 | < .001 | 0.051 |
| Information is spread across too many tools | 55.7% | 11.91 | .180 | 0.034 |
| I do not know where to look | 31.7% | 20.43 | .008 | 0.045 |
| Important decisions are hard to find later | 27.0% | 22.47 | .004 | 0.047 |
| Different teams have different versions of the truth | 26.1% | 9.94 | .308 | 0.031 |
| Search does not return useful results | 23.1% | 56.11 | < .001 | 0.074 |
| AI tools do not have enough context | 22.5% | 17.63 | .024 | 0.042 |
| I do not know which source is authoritative | 18.5% | 15.49 | .051 | 0.039 |
| AI tools use the wrong context | 18.3% | 67.88 | < .001 | 0.082 |
| Access or permissions prevent me from seeing what I need | 16.2% | 2.94 | .709 | 0.017 |
| I usually have the context I need | 9.0% | 50.62 | < .001 | 0.071 |
| I intentionally limit context because of privacy or security concerns | 6.6% | 8.02 | .311 | 0.028 |
| Not applicable | 1.9% | 9.28 | .308 | 0.030 |
| Other, please specify | 1.2% | Not tested: expected counts too small | ||
How to read it
No single chi-square test covers this table. A respondent appears in every row they selected, so the rows are not separate groups of people, and a test of the whole table would count the same person several times. Each answer is tested on its own instead: did the share who selected it differ between the groups? Testing 15 answers gives 15 chances of a false alarm, so the p-values are Holm-adjusted to keep the chance of any false alarm in this table at 0.05.
For “AI tools use the wrong context” the test rejects independence at the 0.05 level (Holm-adjusted p < .001): differences this large between the groups are unlikely if selecting it did not depend on attitude toward ai. Cramér's V of 0.082 is in the negligible range, which is trivially weak in practical terms, even where a sample this large can detect it.
With 10,164 respondents, even small differences reach statistical significance. Read the percentages and the effect size before the p-value.
Limits
- The survey has no weights. These tables are unweighted counts of the people who answered, so they describe the respondents, not developers in general.
- The test treats respondents as a simple random sample. They are people who chose to answer a survey Stack Overflow promoted, not a random sample of developers, so a result holds for these respondents and may not hold beyond them.
- The adjustment covers the answers of this question by this group only, not the other tables in the library. The headline answer is the strongest of those tested, so its effect size is larger than a typical answer's.
- A difference between groups is not evidence that group membership causes the answer.
Next steps
Work with this table
Open the test of “AI tools use the wrong context” in the calculator
The link fills in the respondents who selected this answer and those who did not, by group. The whole table does not open there: its rows overlap, and the calculator's test needs rows that are separate people.
Open this question in the full file
The workspace opens all 30,903 respondents in your browser with this question selected. The survey has no weights, so the table there is unweighted too. Change the group, add a filter or export from there.
Open in the workspace →Run this on your own data — free, no signup
Upload a CSV, Excel, SPSS, Stata, SAS, R, Parquet or JSON file. Everything runs in your browser; your file never leaves your device.
Open the workspace →The same question by other groups
Other questions that differ by attitude toward ai
- First actions when trying to solve a problem by attitude toward ai
- Effect of AI on satisfaction: Implementation by attitude toward ai
- What prevents taking part in technical communities by attitude toward ai
- Usefulness of AI: Implementation by attitude toward ai
- AI in the workflow: Do not use AI for this and do not plan to by attitude toward ai
- Tasks delegated to AI agents by attitude toward ai
Source
Stack Overflow. 2026 Developer Survey, https://survey.stackoverflow.co/2026, licensed under the ODbL 1.0. Tabulated by crosstabs.com. Variables ContextProblems (What makes work context hard to get) and AISentiment. Stack Overflow licenses the survey data under the Open Database License (ODbL) 1.0 and its contents under the Database Contents License (DbCL) 1.0: credit Stack Overflow, keep the licence notice, and share any database made from the data under the ODbL. These tables and their downloads are offered on those terms.