In most cases, a higher temperature results in a faster reaction. The luminescence intensity should correlate in some fashion to the product turnover rate (concentration of products per unit time, say). Therefore a higher temperature should results in a more intense but shorter lasting luminescence. This is exactly what you see if you put commercial glowsticks in the freezer: they will last a lot longer, but will be a lot dimmer as well. I would expect the luminol reaction to behave in a similar fashion, although I haven't verified this.
Assuming the observation is correct, and that the luminol reaction behaves in the way described above, there are two possibilities for the discrepency:
(1) Error in measurement. There is a reason we use a spectrofluorimeter to measure luminescence in a laboratory setting instead of our eyes. Likewise, how are you measuring "duration"? And how are you controlling temperature? To really compare these conditions, you would need to do the experiment carefully, and measure the fluorescence kinetics using a calibrated photon detector.
(2) Statistical anomaly / error in method. You've only taken one measurement. Do it several times and see what the average result is. There might be errors in the starting material concentrations, for instance.