During this project, the authors also included a few questions pertaining to the general format and layout of the forecast. Results of these questions provided valuable feedback for the forecasters in this office, as well as providing useful information to some of the radio broadcasters in the Juneau local area.
One question asked respondents how they preferred to hear the temperature forecast. The respondents could choose between a verbal range, "a low in the lower 40s," a numerical point forecast, "a low near 42," or a numerical range, "a low of 41 to 44". Of those who responded, 58% wanted to hear a particular number given, either a numerical point forecast or a numerical range. However, no one particular answer among the three stood out as the clear choice among the participants.
When drafting the second questionnaire, How Do You Interpret Our Forecasts?, it was determined that a new question about temperatures should be used in order to get a better idea on how temperature forecasts should be worded. It was determined that a numerical point forecast would not be practical, especially in a zone forecast where the temperature often varies throughout the area. Therefore, in Questionnaire #2, the temperature forecast preference question narrowed the choices to two, a verbal range and a numerical range. Results from this question were much more meaningful; nearly 80% of the respondents chose the verbal range.
Another question was designed to determine when the respondent felt that a particular time group for a forecast began, specifically the "tonight" period of a forecast. Nearly three-quarters of the respondents correctly answered that the "tonight" period of the forecast began around 6:00 p.m. local time. Of the 26% of the respondents that answered incorrectly, 20% guessed 9:00 p.m. The perception of when "night" begins may vary somewhat from person to person. This may also be an issue dependent on the time of the year, since sunset times vary widely over the seasons at high latitudes. For instance, in Juneau, the sun sets after 9:00 p.m. during the summer months. Conversely, sunset in December is before 4:00 p.m., and 4% of the respondents actually chose 3:00 p.m. for the "tonight" period to begin.
The questionnaire also asked the respondents which expression of probability was preferable: verbal (ex. "rain likely"), numerical (ex. "chance of rain 60%"), or both. The most common response was "both," preferred by 59% of the respondents. Only 12% of the respondents chose the verbal answer. Considering that "both" includes numerical and verbal expressions, 72% want to see the POPs given at least verbally, but 87% want to see a numerical expression in the POPs. The National Weather Service currently gives the forecast both verbally and numerically, which is what the majority of respondents chose.
Another important aspect to forecast composition is length. Respondents were asked for their opinion of the length of the forecast given the following answer choices: too wordy or too detailed, too sparse or not enough detail, or just about right in length and detail. The percentage of respondents that felt the forecast was "about right" was 70%. As mentioned previously, it was found that 42% of the people who answered this questionnaire named the local radio station as their primary weather source, which was by far the most common answer. It is relevant to mention that some public radio stations have a tendency to edit or shorten the forecast in order to fit their time slots. Therefore, a significant number of people who felt that the forecast was "about right in length" were also radio listeners.
In the Juneau area, it was found that the forecasts are listened to by twice as many people as those who read it. This strongly suggests that a concise, reasonably short forecast best serves the Juneau public. First of all, a shorter forecast is more easily remembered. For instance, it is easier to remember "rain," than it is to remember "rain beginning this morning, then ending toward noon and changing to showers toward evening." Secondly, a shorter forecast has a greater possibility of being read verbatim by radio announcers, while a lengthy one may be edited for time restrictions, possibly leaving out important sections of the forecast.