{"items": [{"author": "Aryeh", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114655475368663", "anchor": "fb-114655475368663", "service": "fb", "text": "I think this one worked for me!", "timestamp": "1355615124"}, {"author": "Arthur", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114656578701886", "anchor": "fb-114656578701886", "service": "fb", "text": "What do you expect the effect size to me? Given your numbers I'd estimate it at one person per google-advertised dance, which is gonna be pretty darn difficult to suss out of door stats.", "timestamp": "1355615370"}, {"author": "David", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114657265368484", "anchor": "fb-114657265368484", "service": "fb", "text": "What is the standard deviation of your attendances?  (I've never studied requirements for sample sizes, but the standard deviation combined with the click rate should suggest some sort of ballpark figure.)", "timestamp": "1355615560"}, {"author": "Phillip", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114671355367075", "anchor": "fb-114671355367075", "service": "fb", "text": "You should not think of advertising as transactional, I see an ad&lt; I come. People don't work that way. There is seeing the ad multiple times, getting comfortable with the idea, figuring out what contra is, etc. I would love to see more advertising to all of the colleges. I think at that age lots of people try things they otherwise wouldn't (it's when I did).", "timestamp": "1355619384"}, {"author": "Lex", "source_link": "https://plus.google.com/111102660583646544610", "anchor": "gp-1355620099948", "service": "gp", "text": "A great question. After all, how do you know where to focus your efforts, if you don't know what is working and what is not?\n<br>\n<br>\nAre you asking newcomers where they heard of you? If not, that's an old and simple trick you could use. As newcomers come in the door, ask them where they heard of you. Add tick marks to one of 5-6 categories you have set up as they tell you.\n<br>\n<br>\nAnother trick is to give out discounts, or even free passes, via your advertising networks. Have the pass include some sort of clue as to which kind of ad it came from.\n<br>\n<br>\nFinally, note that in advertising, it's not so simple as an ad working or not. Ads can also raise awareness, which is very valuable to you, but is harder to measure.", "timestamp": 1355620099}, {"author": "Jeff&nbsp;Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114716012029276", "anchor": "fb-114716012029276", "service": "fb", "text": "@David: over the past 18 dances our average has been 108 and our standard deviation has been 19.", "timestamp": "1355632494"}, {"author": "David", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=114747835359427", "anchor": "fb-114747835359427", "service": "fb", "text": "That's very noisy... I think you're better off doing a survey or something.<br><br>One-standard-error bars for the mean attendance after N dances will be like +/- 19/sqrt(N).  So if you think the ads are bringing in 4 people per dance, which looks pretty optimistic for 10 clicks, then you'll need something like N=22 (that's N=22 dances with ads, and N=22 dances without ads, so about two years' worth) before the error bars for the \"with ads\" mean and the \"without ads\" mean don't overlap.<br><br>[I'm not a proper stats person, but this ball-park figure makes sense to me.]<br><br>If you think the ads are bringing in only 2 people per dance, then forget about doing an RCT using only attendances.", "timestamp": "1355642409"}, {"author": "John", "source_link": "https://plus.google.com/100633726263571579985", "anchor": "gp-1355673535506", "service": "gp", "text": "In addition to \n@Lex\n's thoughts it is also interesting though more \u00a0difficult to try to find out from your frequent dancers how they decide to come or not: \u00a0schedule conflicts, band/caller, notices, other friends, etc. I think a large portion of a dance goers are \"regulars\" and knowing what causes them to not come could be valuable. Also attempting to get feedback from newcomers before they leave to see what they thought. As someone once pointed out if you get 10 new people at each dance and one of them becomes a regular at twice a month that would be 24 regulars in a year.\u00a0", "timestamp": 1355673535}, {"author": "Jeff&nbsp;Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1355684644403", "service": "gp", "text": "@John\n\u00a0\"10 new people at each dance and one of them becomes a regular at twice a month that would be 24 regulars in a year.\"\n<br>\n<br>\nRight, though there's attrition too.", "timestamp": 1355684644}, {"author": "Peter", "source_link": "https://www.facebook.com/jefftk/posts/114654955368715?comment_id=115109055323305", "anchor": "fb-115109055323305", "service": "fb", "text": "Wouldn't there be any way to ask attendees if they were referred by the ad, to help reduce the noise?  (I don't know anything about contra dances.)", "timestamp": "1355709101"}]}