Any reference exist on when non directional hypothesis can be used. I'm in a fix between two reaserchers. One says you have to use directional hypothesis other says as you are working on a new constructs relation with other existing construct, you can use non directional hypothesis. Kindly shed light.
The use of a directional or non-directional hypothesis depends entirely on theory. If you have a strong theoretical basis for the direction of your hypothesis, then it doesn't matter whether you are working with a new versus familiar construct.
A nondirectional hypothesis is of the theta not equal to theta-nought type giving rise to a two-tailed test. It is to be entertained if there is no directional comparison. If it is enough to be interested in the difference of two variables the corresponding hypothesis will be nondirectional. A directional comparison i.e. of the < or > type also calls for a corresponding directional hypothesis depending on the test statistic.
The use of a directional or non-directional hypothesis depends entirely on theory. If you have a strong theoretical basis for the direction of your hypothesis, then it doesn't matter whether you are working with a new versus familiar construct.
In general, the alternative hypothesis (one-sided versus two-sided) depends primarily on what you would like to prove - Two sets of results are simply different or a directional difference (e.g., owing to a factor or new process that is expected to produce a smaller or larger value for the parameter of interest).
- Badges
- Science topic