But the focus of the test is sexism towards to women. It would be pretty complex, and a little pointless to assess sexism to both men and women within one test.
That criticism is akin to saying that a test that measures depression doesn't think that anxiety exists and is biased, because it has no anxiety items. Although that would actually be more fair as at least anxiety and depression are often co-morbid.
If you read out a brief description of sexism such as the one you provided, and told people to read it and rate the extent to which they are sexist, do you honestly believe that all those who do hold sexist beliefs will say "yeah I hold my hand up, I am very sexist".
'Course not, for many reasons including social desirability, simply not realising it, or thinking that any beliefs which could be considered sexist are justified in some way and therefore not sexist but fact.
In other words, it would be a pretty **** measure of attitudes. And not only that, it would also completely ignore the entire rational behind their theory that sexism is multi-dimensional. The creation of this test was purely to asses dimensions within sexism.
I agree that the test could be more specific in regards to the use of "usually, always etc". But I think you're overthinking it a little, for example when I read a statement such as "women seek to gain power", I take that to mean usually not always.
And besides, I'm certain that the internal consistency (reliability)
for this test is high. Otherwise it wouldn't be so persistently used in research, and would have struggled to have faced publication in the first place.
This test is one of the most commonly used measures for sexism in psychology. For example one study found that the measure of hostile sexism positively correlated with rape myth acceptance (
http://link.springer.com/article/10.1007/s11199-007-9196-2)
I just think people are being too quick to brush of this test as useless when they don't know the research surrounding it.