Intermediateexam strategy
Chi-Square in Biology Exams: The Complete Worked Guide
Learning objectives
- Compute chi-square values step by step from observed and expected counts.
- Determine degrees of freedom and use the critical value table.
- State correct conclusions without overstating what the test proves.
- Recognize which exam questions call for chi-square.
When to Use Chi-Square
Chi-square tests whether observed categorical data differ significantly from expected values. In biology exams, the trigger is almost always genetics: 'Do these offspring counts fit the expected 9:3:3:1 ratio?' or 'Does this dihybrid cross show independent assortment?' It also appears in ecology (habitat preference counts vs expected distribution) and behavior (choice-chamber results vs random expectation). The recognition cue is a table of observed counts plus an expected ratio or distribution to test against. If the question gives means or continuous measurements instead of counts, chi-square is the wrong test — that confusion is a deliberate trap. Counts vs expectation: that's chi-square territory. Tabulate every chi-square calculation identically, every time; the fixed table format prevents skipped steps under pressure.
The Calculation, Step by Step
The formula: χ² = Σ((O − E)² / E), summed over all categories. Worked example: a dihybrid cross yields 90, 30, 28, 12 offspring (total 160), testing a 9:3:3:1 ratio. Step 1: expected counts = 90, 30, 30, 10 (160 × 9/16, 3/16, 3/16, 1/16). Step 2: compute each (O−E)²/E: (0)²/90 = 0; (0)²/30 = 0; (2)²/30 = 0.133; (2)²/10 = 0.4. Step 3: sum: χ² = 0.533. Step 4: degrees of freedom = categories − 1 = 3. Step 5: compare with the critical value at p = 0.05, df = 3, which is 7.81. Since 0.533 < 7.81, the difference is not significant — the data are consistent with 9:3:3:1. Show every step: tables of O, E, and (O−E)²/E earn method marks independently.
Expected counts → (O−E)²/E per category → sum → df = categories−1 → compare with critical value at p=0.05. Tabulate O, E, and each term.
Reading the Critical Value Table
Exams provide the critical-values table; your job is using it correctly. Find the row for your degrees of freedom and the column for p = 0.05 (the standard significance threshold in school biology). The decision rule: if calculated χ² exceeds the critical value, the difference is significant — reject the null hypothesis (the data do not fit the expected ratio). If calculated χ² is smaller, the difference is not significant — fail to reject the null (data are consistent with expectation). The most common table error is reading the wrong column (p = 0.01 or 0.10 instead of 0.05) or the wrong row. Circle the correct cell before comparing — under time pressure, table misreads are embarrassingly common.
Stating the Conclusion Correctly
The conclusion wording is itself marked, and overstating loses the mark. Correct: 'Since χ² (0.53) is less than the critical value (7.81) at p = 0.05 with 3 degrees of freedom, there is no significant difference between observed and expected; the data are consistent with a 9:3:3:1 ratio.' What not to write: 'This proves the ratio is 9:3:3:1' — chi-square never proves; it only fails to find significant deviation. Also avoid 'the hypothesis is correct' — write 'we fail to reject the null hypothesis' or 'the data are consistent with.' Include the numbers (calculated value, critical value, df, p-level) in the conclusion sentence; conclusions without numbers read as hand-waving and lose the interpretation mark. Memorize the critical values for df 1-4 at p = 0.05; instant recall of common values saves table-lookup time.
Chi-square can only show data are consistent (or inconsistent) with expectation. 'Proves the ratio' is the single most penalized phrase in chi-square conclusions.
Key points
- Use chi-square for observed counts vs expected ratios — genetics crosses are the classic trigger.
- Tabulate O, E, and (O−E)²/E per category; sum for χ²; df = categories − 1.
- Compare against the critical value at p = 0.05 in the correct row and column.
- Calculated > critical → significant difference, reject the null; smaller → consistent with expectation.
- Conclusions must include numbers and never claim proof — 'consistent with,' not 'proves.'
Common misconceptions
✓ Truth: It means no significant deviation was detected. Other ratios might fit too — the test only evaluates the one you specified.
✓ Truth: df = categories − 1. Using the raw category count reads the wrong table row and flips borderline decisions.
Frequently asked questions
What if expected counts aren't whole numbers?
That's normal — expected values like 37.5 are fine. Keep one decimal through the calculation; only the observed counts must be whole numbers.
Can chi-square be used with percentages?
No — convert percentages back to counts first using the total. The formula requires frequencies, and using percentages directly gives wrong values.
What does p = 0.05 actually mean here?
There's a 5% probability of seeing deviation this large by chance alone if the null hypothesis were true. Below that threshold, we consider the deviation significant rather than random.