Many article database search engines require you to use search operators like AND (to require that all terms connected by AND show up in the article record) and OR (use to connect synonyms). Others have that option but also have an option to work more like Google - just type in all your keywords and it will look for any and all of them, without being too finicky. Use of operators will give you more control over your search. Make sure you know what's called for in the one you're searching!
Tip: There's often an example near the search box that will show you how to put your search together correctly.
Get in the habit of using truncation symbols to capture all the relevant articles with your search. Don't miss "Environmental Issues in India" because you searched for "India and environment." Truncation symbols will retrieve all the relevant words: environment, environmental, environmentalism, -alist, ...
Tip: The most common truncation symbol is the asterisk (*) and you can generally just go with that. A quick check in the Help section will tell you for sure.
An article about "population growth" is really different than an article about a country's aging population and its economic growth. Pay attention to phrases - you can usually force words to appear as a phrase with "quotation marks."
Adding a proximity requirement to your search is a good way to make your result more relevant and focused. Often this is done with a tilde (~) like this: "education fertility"~4 means those two words must be within four words of each other. Proximity searching is not as widely available as some other search features like truncation and phrase searching.
The way your search results are sorted is important -- don't waste your time looking over a set of results until they're in the most useful sort order. Typical defaults are relevance-sorted or date-sorted results. The Library Catalog will list some search results alphabetically, which is rarely useful for research purposes. Web of Science has a cool times-cited sort feature, showing you results in order by the number of times they've been cited by other researchers.