Since the 1940s, surveys have been asking questions about overall happiness--"In general, how happy would you say you are--very happy, fairly happy, or not very happy?" Of course, there are are all sorts of things that could make people happy or unhappy, so I've kept an eye out data on more specific feelings. There isn't much, but a few surveys have included a series of questions: "During the past few weeks did you ever feel....
1. Particularly excited or interested in something
2. So restless that you couldn't sit long in a chair
3. Proud because someone complimented you on something you had done
4. Very lonely or remote from other people
5. Pleased about having accomplished something
7. On top of the world
8. Depressed or very unhappy
9. That things were going your way
10. Upset because someone criticized you?"
Notice that the odd numbered ones are generally regarded as good feelings, and the even numbered ones as bad. You can make a total score by adding the percents for the five good feelings and subtracting the percents for the five bad ones. For example, in 1963 it's:
The numbers mean 52% said they'd been particularly excited or interested, 45% said they'd been restless, and so on. Here's how the total scores have changed over the years:
Nov 1963  +98
July 1965 +152
April 1981 +203
June 1990 +222
Sept 2001 +180
Jan 2002 +211
So it looks like people have been feeling better according to these questions. That's interesting, because happiness as measured by the general question hasn't really changed since the 1950s.
I also looked at trends for the individual question. Excited, proud, pleased, going my way, and bored have become more common. Restless and unhappy have become less common. There's no clear change in lonely, on top of the world, and upset. So four out of the five positive feelings have become more common, but so did boredom. Why? I don't know. That's the advantage of a blog over a research paper--you can just say "I don't know" and leave it at that.
Technical PS: Notice the dates--the first survey was taken just after the assassination of JFK, and the one in September 2001 just after the 9/11 attacks. It seems reasonable that both of those events produced some negative feelings. If you adjust for that, the upward trend in the total scores gets even more clear. In a linear regression including a time trend just a time trend, the t-ratio is 2.26; add a dummy for the Nov 1963 and Sept 2001 surveys, and the t-ratio for the trend rises to 4.41.