5.4.A When a cross does not fit
A ratio that will not fit
- Mendel's ratios are a model. Many traits do not follow the ratios his laws predict.
- A deviation is found by counting: the observed phenotype ratio differs statistically from the predicted one.
- The test is chi-square. A value above the critical value means reject the null hypothesis: chance alone is unlikely to explain the difference.
- A rejected ratio is a clue. Something about the gene, or where it sits, is outside Mendel's model.
The detail
- Mendel's model assumes each trait is set by one gene with a dominant and a recessive allele, and that genes for different traits sit on different chromosomes.
- The null hypothesis is always the Mendelian one: the offspring fit the predicted ratio, and any difference is chance.
- Chi-square adds up (observed − expected)2 ÷ expected over every phenotype class. Degrees of freedom are the number of classes minus one.
- The critical values at p = 0.05 are on the AP equations and formulas sheet: 3.84 for 1 degree of freedom, 5.99 for 2 and 7.81 for 3.
Worked example
- Two heterozygous plants are crossed. The prediction is 3 tall : 1 short. Of 160 offspring, 100 are tall and 60 are short.
- Expected: 120 tall and 40 short.
- Chi-square = (100 − 120)2 ÷ 120 + (60 − 40)2 ÷ 40 = 3.33 + 10.00 = 13.33.
- With 1 degree of freedom the critical value is 3.84. 13.33 is above it: reject the null hypothesis. These plants do not fit 3 : 1.
Watch out
- Rejecting the null hypothesis does not say what the cause is. It says the Mendelian ratio is a poor fit, and sends you looking.
- A value below the critical value means fail to reject. It never proves the ratio is right.
In the last guide every cross came out the way Mendel's laws said: 3 : 1, 1 : 1, 9 : 3 : 3 : 1. Those ratios are a model, built on peas and on seven traits that happened to behave. Many traits do not follow them.
You find a deviation by counting. Cross the parents, sort the offspring by phenotype, and compare what you observed with what the model expected. No brood matches exactly, so the question is whether the gap is bigger than chance would give. That is what the chi-square test answers. The null hypothesis is that the offspring fit the Mendelian ratio. If chi-square is above the critical value, you reject it.
A rejected ratio is the start of the interesting part. Something about the gene, or about where it sits in the cell, is outside the model. This guide covers five such causes, then one more thing Mendel's model leaves out: the environment. Tap each card in the figure to see where the guide is going.
- Many traits do not follow Mendel's ratios.
- A deviation is identified in data: observed ratios differ statistically from predicted ones.
- Chi-square above the critical value: reject the null hypothesis.
A cross is predicted to give 3 purple : 1 white. Of 200 offspring, 135 are purple and 65 are white. What does a chi-square test show? (The critical value for 1 degree of freedom is 3.84.)
Expected: 150 purple and 50 white. Chi-square = (135 − 150)2 ÷ 150 + (65 − 50)2 ÷ 50 = 1.5 + 4.5 = 6.0. That is above 3.84, so reject the null hypothesis. Both classes go into the sum; stopping after the first gives the 1.5.
Tap a card to read how that cause changes a ratio.