It depends on your research question. A directional hypothesis (positive or negative) may be appropriate in select circumstances when only one direction for an association is clinically important or biologically meaningful (e.g. drug x is more likely to cause toxicity than placebo). Also, a directional hypothesis may be appropriate when there is strong evidence from prior studies that association is likely to occur in one direction (e.g. studies measuring association between smoking and different types of cancer since prior literature indicates that smoking is not likely to be protective). However, it is safer to go with a non-directional hypothesis. In most instances both sides of the alternative hypothesis are of interest and worth publishing.
In either case, this should be decided prior to analyzing the data to avoid choosing a directional hypothesis to simply reduce the p value. Also most grant and manuscript reviewers expect non-directional (two sides) hypotheses.
Browner WS, Newman TB, Hulley SB. Getting Ready to Estimate Sample Size: Hypotheses and Underlying Principles. In: Designing Clinical Research, 3rd edition. Hulley SB, Cummings SR, Browner WS et al. (eds). 2007
Dear Tea, thank you so much far rapid replying and offering the reference. Based on Popper's falsifiability, is negative hypothesis better than positive?
Positive and negative hypotheses refer to the direction of the effect. That is, a positive hypothesis assumes there is positive correlation between the exposure (independent variable) and outcome (dependent variable) and negative hypothesis assumes there is negative correlation between the exposure and outcome. This will depend on the natural relationship between the variables you are studying and so neither can be better.
All hypothesis testing is somewhat related to Popper's falsifiability. That is, we can never prove a hypothesis. In the case of directional hypothesis, we set a null hypothesis and then predict an alternative hypothesis in either the positive or negative direction. From here we can only reject or fail to reject our null hypothesis. When we reject the null hypothesis, it is logical for the alternative hypothesis to be true (regardless of the direction). However, we can never be certain and therefore never "accept" the alternative hypothesis.
A better explanation can be found in Quinn and Keough's Experimental Design and Data Analysis for Biologists, Ch 3.1.1 (see attached)