Which conditions are required to use the sign test?

Which conditions are required to use the sign test?

• A. Data are nominal, design is repeated measures, and you look for differences
• B. Data are ordinal, design is between-subjects, differences
• C. Data are interval, repeated measures, correlations
• D. Data are nominal, between-subjects, mean differences

Answer: (A)

The sign test works by looking at the direction of difference for each pair of related observations and ignoring how big the difference is. Because it uses only whether one condition yields a higher value than the other for each subject, you need paired data from the same participants (a repeated-measures or matched-pairs design). If the outcome under one condition can be categorized in a way that lets you see which condition “wins” for each person, you can record a plus or minus for every pair and test whether pluses and minuses are balanced. That makes nominal data workable in a dichotomous sense, since you’re counting how many pairs show one condition greater than the other. So this option correctly captures the requirement of a repeated-measures design and data that can be ordered in a directional sense. The other descriptions miss this pairing requirement or rely on magnitude, correlations, or between-subjects designs that aren’t compatible with the sign test.

The sign test works by looking at the direction of difference for each pair of related observations and ignoring how big the difference is. Because it uses only whether one condition yields a higher value than the other for each subject, you need paired data from the same participants (a repeated-measures or matched-pairs design). If the outcome under one condition can be categorized in a way that lets you see which condition “wins” for each person, you can record a plus or minus for every pair and test whether pluses and minuses are balanced. That makes nominal data workable in a dichotomous sense, since you’re counting how many pairs show one condition greater than the other. So this option correctly captures the requirement of a repeated-measures design and data that can be ordered in a directional sense. The other descriptions miss this pairing requirement or rely on magnitude, correlations, or between-subjects designs that aren’t compatible with the sign test.