How to find a fitted statistical model to a series of data? There are several outputs of a numeric computations, I'm seeking a method to model them with a general statistical model (e.g., distribution function). 
For example:
...
0.044423
0.127713
0.010692
0.019980
0.013242
0.071450
0.054766
0.013618
0.067680
...

Each series has different length. Intuitively, Statistica has an option such as "Distribution Fitting" which works well but only on a series of data at a time. It provides a graph and model as follows for each series:
Exponential, Chi-Square test = 72.94, df = 12 (adjusted) , p = 0.00


There are at least 30 such a series. How to fit a general (best) model to each of them but finally select the best overall fitted distribution model? How to avoid doing the job manually?
A short comment about the above output is also appreciated. How can the information be interpreted?

Here is what I recently did. I just put together all histograms of 30 series (black dots) and then used a kernel-based smoothing algorithm to find a curve (red curve). Now it should be easy to fit a known distribution to that curve (because it is only one).
Any comments about this solution?

** please note the updated comments put on answers.
 A: I would first look at histograms, and it would be helpful if you would show them to us, and also give us an idea of sample sizes.
In the output you show, I expect that the data were binned and then the frequencies of observations in each bin were compared to the expected counts for an exponential distribution (where the rate parameter for the exponential was estimated from the data, as $1/\bar{x}$).
In choosing among possible distributions/models, it can be helpful to know more about where the data came from and what you are trying to learn.  Are you trying to make comparisons among the 30 groups, are you trying to learn more about the underlying process, or are you trying to estimate some other features, like tail probabilities?
Adding more, following clarifications in the comments...
You have a number of independent data sets, which you are assuming follow the same family of distributions (in other words, a model) but with different parameters, and want some automated procedure for selecting the appropriate family (or model).  
I must say that I don't like this.  I prefer to have some understanding of the process giving rise to the data, with perhaps some physical or other mechanism that might justify a particular distribution.  And in the end I'd want to look at a lot of pictures.
$\chi^2$ tests
Nevertheless, if you want to proceed: you can form a $\chi^2$ goodness-of-fit test for pretty much any kind of distribution; for continuous data you'll have to do some binning.  And for each series you'll have to estimate the unknown parameters.  But then you just sum the $\chi^2$ statistics across series to get a measure of the overall fit.  The degrees of freedom (no. bins $-$ 1 $-$ no. parameters estimated) add, too.  A problem with this is that you can't easily compare the $\chi^2$ statistics for models with different numbers of parameters.  You might want to compare p-values, but that's not right.
What I would do
I would first look at the histograms.  You say they look exponential; I'd consider a family of distributions that includes the exponential, such as the gamma distribution.  Gamma distributions are quite flexible, so they might work.  I'd then fit the gamma for each series, look at the quality of the fit in each (by eye, either with a qq-plot or by superposing the fitted distribution on the histogram), and also look at plots of the estimated parameters (and SEs or confidence intervals) across the 30 series.
A: You say that you want a "general statistical model"---does that mean that you want it to fit in a known distributional family? Have you thought about doing kernel density estimation? This would be a non-parametric solution to your problem, but it doesn't give you either a gamma, an exponential, a normal, etc back.
A: I this is time series Histograms provide little or no information/clues as to the underlying model. The answer to your questions is fairly simple of this is time series data , fit an ARIMA model taking into account Pulse, Level Shifts, Seasonal Pulses and any local time trends.
