My #1 doubt always comes from what constitutes a statistically significant temperature change on a 100-150 yr scale (or however long we have had consistent measurements). What is the variance in 100 yr temperature differences over the last 100K to 1M years? Even if the current historical models are accurate do we really have accuracy at this resolution? 2 or 3 degrees C change over 100 yrs may be a common occurrence.
To me it like trying to judge whether a stock price movement over the course of 10 minutes is significant based on historical daily stock prices.
As far as I'm aware, there is (indirect) data on that.
In fact I believe part of the concern is that the data suggests that when the planet was in fact 2 or 3 degress off, it was a rather less hospitable planet.
Regarding your stock analogy: People make real money decisions based on ten minutes of stock price movement and historical daily stock prices.
To me it like trying to judge whether a stock price movement over the course of 10 minutes is significant based on historical daily stock prices.