What tools do you suggest to determine whether dividends are becoming popular or not?

The best indicators are along the lines of Zeelotes' old posts on building a speculative/defensive signal.
The essence is to look at a lot of metrics of style, to see which end of that spectrum is doing well.
In this thread I've mentioned dividends, but the Value Line "Safety Rank" is another indicator. There are many possible choices of "flavour".
His signal checked several such indicators to see which population was doing better lately, and build a consensus.
Here are some of his impressive posts on that subject
http://www.datahelper.com/mi/search.phtml?nofool=y...
http://www.datahelper.com/mi/search.phtml?nofool=y...
I think his starting point was to divide time into 3 states: speculative/defensive/normal based on the relative performance of different groups in the last couple of weeks. "Normal" times were about 70% of days.

For simplicity, I just look at dividends.
I look at three equally weighted portfolios of Value Line stocks, divided into three categories
* All stocks without any dividend
* Dividend payers, highest 50% of dividend yields
* Dividend payers, lowest 50% of dividend yields

Compare the relative recent performance of these three groups, and how each of the three groups does in the next month, and you'll see some very non random results.
(there are actually six different possible states based on the rank order of those three groups if you want to get dangerously overtuned)

Alas, this signal isn't easy to calculate without a backtester / stock database handy that lets you test various lookback time frames.
But you could probably learn something simply looking at the average "Total Return X-Week" field in the VL database for the three groups. 4 week, 13 week, or 26 week.
Or some sum/average combination of those.

Here is a simple example that any VL subscriber could do with Excel:
For each of the three groups, calculate the average 13-week and average 26-week returns for the group, and sum the two averages.
If the no-dividend group was the best performing of the three groups on that sum, go long the no-dividend group the next month.
In backtest this simple test is bullish only around 43% of the time. It seems better than random for identifying the minority of the time that speculative / high growth / momentum slate of equities do their best.
For most things you might go long, the CAGR during the 40% of months this is bullish seems to be about twice the CAGR in months it's bearish.
A long/cash switch won't outperform all-long-all-the-time of the same portfolio, but might beat S&P buy and hold, as the S&P returns during the bearish stretches aren't much.

Jim