I also did this test by making a new variable with just the hypothesized median value. Then run then the 2 related samples Sign test with extra and this new variable. I got the same output as your tweaked Binomial test (which I ran on my dataset)
thanks for the suggestion. That seems indeed to be an alternative approach i.e. to trick the two-sample sign test in performing a one-sample sign test :-) Would be nice if SPSS simply creates a one-sample sign test where you can enter the hypothesized median.
@@journeymantraveller3338 there is also another method, without creating an extra variable. You can use Data - Select Cases to select the cases that are not equal to the hypothesized median. Then set the measurement level to scale. If you then use the Nonparametric tests - One sample, you can set the cutpoint at the hypothesized median. It is not uncommon in SPSS to have multiple methods that can lead to the same result :-)
I also did this test by making a new variable with just the hypothesized median value. Then run then the 2 related samples Sign test with extra and this new variable. I got the same output as your tweaked Binomial test (which I ran on my dataset)
thanks for the suggestion. That seems indeed to be an alternative approach i.e. to trick the two-sample sign test in performing a one-sample sign test :-)
Would be nice if SPSS simply creates a one-sample sign test where you can enter the hypothesized median.
@@stikpet Todd Grande also uses the method.
@@journeymantraveller3338 there is also another method, without creating an extra variable. You can use Data - Select Cases to select the cases that are not equal to the hypothesized median. Then set the measurement level to scale. If you then use the Nonparametric tests - One sample, you can set the cutpoint at the hypothesized median.
It is not uncommon in SPSS to have multiple methods that can lead to the same result :-)