Rain vs. Snow

Because the Juneau survey was conducted during the winter months, the authors wanted to use snow in some of the questions. The authors chose to avoid using the word "precipitation" in the questionnaires, because the questions were meant to be casual and conversational rather than sounding scientific or institutional. We were concerned, though, about the possibility that people might answer questions pertaining to snow differently than they would if rain were used in the same format. If any variability between public reaction to rain and snow forecasts did exist, it would surely influence the results of our POP questions. This possibility was also relevant to the determination of the overall public perception of accuracy of WSFO Juneau forecast products.

To detect a possible rain vs. snow bias among respondents, several of the questions which were related to perception of accuracy or interpretation of probability qualifiers had two forms, one which used rain in the question and the other which used snow. In fact, questionnaire #3, How Accurate Do You Think We Are?, had two separate versions designed to cover rain and snow in relation to accuracy independently. Both versions had an even mix of rain and snow based questions, and one version simply used snow where the other used rain.

One question which did appear to exhibit a marked difference in response to rain vs. snow oriented scenarios was,

On the morning weather forecast, the meteorologist says, "rain (snow) likely today ." You understand this to mean:

While the primary focus of this question was to determine how the public interprets the word likely, by using snow in half of the questions and rain in the others, we were attempting to confirm whether or not the form of precipitation would be an influencing factor in the responses. If precipitation type had no effect, then responses should be comparable within an acceptable standard deviation. For this question, some discrepancies did show up as 60% of the people chose the response, "Snow is likely somewhere in the forecast area during the day," and only 40% chose the equivalent response using rain as the precipitation type. Also, only 7% of the respondents chose, "Snow is likely to occur most of the day," while 20% of those given the rain scenario chose this response.

The above question was paired with another which was worded identically except used "a 60% chance of rain (snow)" in place of "rain (snow) likely." Results from this couplet shed more light on the rain/snow comparison. Here, 86% of the respondents chose the answer, "There is a 60% chance rain will occur somewhere in the forecast area during the day," and 84% chose the identical rain response. This result suggests that the inconsistencies encountered with the aforementioned question had more to do with the ambiguity of the word likely than with a difference of opinion between rain and snow forecasts.

The questionnaire had another set of questions designed for similar comparison purposes which asked,

The forecaster said, "chance of rain (snow) today," but it did not rain (snow) at your house. What is your opinion of the forecast?

Many versions of this question were asked, using a mix of both rain and snow and including the word chance alone paired with a 30% POP, and the word likely alone paired with a question using a 70% POP. Again, the greatest discrepancy was found when comparing the answers for the questions using likely and 70% regardless of whether they listed rain or snow as the precipitation type. People were an average of 12% less critical of the likely forecast than the 70% forecast. On these couplets, variation between rain and snow responses was slight (>10%) but did exhibit a trend suggesting some people were more critical of the forecast when snow was in it. This conclusion is consistent with the findings of Murphy et al. (1980) concerning the public's higher sensitivity to winter forecasts.

The questions discussed thus far were used completely independent of each other in separate questionnaires so that a single respondent only saw one version of the question. For the sake of comparison, we also chose a question for use with both rain and snow, and placed them one after the other. This way we could determine whether or not a person, when confronted with considering the same situation in rain and in snow, would answer exactly the same. The question read:

At what percent chance of rain (snow) would you consider altering your plans, dress or schedule to accommodate for rain (snow)?

Results from this couplet did indicate a slightly higher sensitivity to a snow forecast. The median response for the rain question was a 70% chance, and for snow, the median was 60%. However, the mean shows a much more narrow gap between rain and snow, 65% vs. 63%. Like many of the other questions asked in this project, this one had other objectives, so another version was included in which the numerical probability was replaced with an equivalent qualifying word. The results were very close in this scenario. For both rain and snow, 34% of the respondents said they would consider altering their plans, dress or schedule upon hearing the word likely applied to the chance of rain (snow). This was particularly interesting given that other areas of the questionnaire project indicated that there is little consensus as to what the word likely actually means in a weather forecast.

Based on these results, the authors concluded that there may be a slightly higher sensitivity to snow forecasts in Juneau, but it does not appear to cause a significant bias in the perception of forecast accuracy. These questionnaires were completed during the winter months, and it is possible that the results may have been influenced by the fact that Juneau was already in the middle of snow season.

Back to the introduction
Next section