My variable represents the outcome of a mock-election, between three different parties(so categorical). My hypothesis is that with the information presented to the subjects, a majority will vote for party A. A simple frequency table shows that indeed most subjects (75 percent) voted for party A. Does this confirm my hypothesis/reject the 0-Hypothesis, or is there a test that I should do to show if this majority is significant? I have the feeling that there is such a test, but can not seem to figure out which one it is and how to do it in spss. Can anyone here give me advice on that?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
2
You could think of this as a binomial test: let $p$ be the probability that a random subject votes for candidate $A$. Your null hypothesis is that $p \le \frac12$, and the alternative that $p > \frac12$. The linked Wikipedia article notes that this is available in SPSS.
-
$\begingroup$ Thank you Dougal. I have performed such a test, using a binomial version of my variables: 0 not voting for party A, 1 voting for party A. Test proportion 0.5. The output provides me with a two-tailed Sig. However, as you said, I am interested in the case that p>0.5, which is one-tailed. Is there a way of doing this in spss? $\endgroup$– StefanCommented May 6, 2017 at 11:58
-
$\begingroup$ I don't use SPSS, and don't know if it has a one-tail option. You can do it in Excel with the
Binom.Distfunction mentioned by the Wikipedia page. $\endgroup$– DanicaCommented May 6, 2017 at 13:07