The null hypothesis is always that the two groups identified in the CONTRAST statement have equal success probabilities:

H₀: Pr(TEST_variable = 1 | CLASS -1) = Pr(TEST_variable = 1 | CLASS 1) ,

where CLASS -1 means the CLASS variable level for which the CONTRAST coefficient is -1. See the MULTTEST Coefficients table in the displayed output to see the assignment of coefficients to levels. Note that if no CONTRAST statements are given, default contrasts comparing the first level (coded 1) with each of the other levels (coded -1) are automatically generated. With the UPPERTAILED option, the alternative hypothesis is

H_a: Pr(TEST_variable = 1 | CLASS -1) < Pr(TEST_variable = 1 | CLASS 1) ,

This essentially means that the probability of success is higher in the 1 level than in the -1 level. With the LOWERTAILED option, the alternative hypothesis is

H_a: Pr(TEST_variable = 1 | CLASS -1) > Pr(TEST_variable = 1 | CLASS 1) ,

This essentially means that the probability of success is lower in the 1 level than in the -1 level.