Nominal data, derived from the Latin word 'name,' is just that: a name. Republicans, Democrats, Independents, and so on. You have no idea who or why you're voting for, but you do have a party name, so that's something, right? To make this more quantitative, give each a random number of your choosing (usually small sequential numbers like 1, 2, or 3 but it could be anything). The kinds of statistical analysis you can do with nominal data are severely restricted since the value of the number has no relevance in your environment; it's simply a placeholder.
Ordinal data gives a numerical order to a prior category, such as 'Republican,' although that order does not have to signify anything. Let me explain: suppose you're questioned on a poll to rank your preferred candidate on a scale of 1 to 5. Obviously, a value of '5' is better than a value of '1,' but what does it mean? I'm glad you asked.
Interval data takes ordinal ideas and couples them with schemata that make sense. Using our prior 1-5 scale, for example, 1 degree Fahrenheit is cooler than 5 degrees. Interval data lacks a point of origin, which makes sense when discussing temperatures since 0 degrees Fahrenheit does not indicate a lack of temperature.
Ratio data combines the preceding three ideas and adds a point of origin, or '0' value. The greatest example I could think of was speed. We all realize that 5 mph is quicker than 1 mph, precisely since we all know that 0 mph implies we're not moving at all.
