| 英文摘要: | To the Editor —
Perceived deviations in daily local temperatures can alter individuals' views on global warming1, 2, 3, 4, 5, 6. Here, however, I use an experiment to show that prompting individuals to remember how the weather felt over the past year severs the relationship between perceptions of the daily temperature with estimates of last year's temperature deviations and eliminates the 'local warming effect'. The results demonstrate the limits of this effect and suggest ways to rhetorically counteract it.
Zaval et al.6 show that when individuals perceive the day's local temperature to be warmer than usual, they then overestimate the number of warm days through the year, which, in turn, leads them to increase their stated belief in and concern about global warming. Moreover, information about the distinction between weather and climate, or changes in terminology (for example, global warming versus climate change) do not counteract the impact of local temperatures on beliefs. The size of the local warming effect rivals the impact of age, race, and education on global warming attitudes1.
Evidence for the local warming effect, as it is often called5, 6, is in line with literature showing that people base survey responses on whatever information most quickly comes to mind7. The easy availability of today's temperature in memory triggers people to then remember other warm days in the past, overestimate the frequency of warm days through the year, and increase their belief in and concern about global warming. What is at work is attribute substitution, where individuals base beliefs on what is salient in the mind regardless of its 'objective' relevance8. But the effect may not be particularly robust: it depends on how the survey question is asked. Attribute substitution can be counteracted via individual motivation and/or the information environment9, 10. In the case of the local warming effect, one approach is to ensure the availability of other memories, such as temperature trends over the past year10.
To test this, I implemented an experiment with two randomly assigned conditions. The 'non-prompt' condition replicates study 4 of Zaval and colleagues6. Participants answered the same four questions. Is the local temperature today colder or warmer than usual for this time of year? (1, much colder; 2, somewhat colder; 3, about the same; 4, somewhat warmer, 5, much warmer). Over the past year, what percentage of days seemed to be warmer than usual for that time of year, compared with the historical average? How convinced are you that global warming is happening? (1, not at all convinced; 2, a little convinced; 3, somewhat convinced; 4, completely convinced). How personally worried are you about global warming? (1, not at all worried; 2, a little worried; 3, somewhat worried; 4, a great deal worried).
The second condition was identical, except the question asking respondents to estimate the percentage of warm days last year included the instruction: “When thinking about temperatures over the last year, remember that temperature patterns vary; indeed consider last winter compared to today. Thus think not only of the feeling today but also how you felt throughout the year”. This prompt is meant to make the sensations of temperatures over time available in memory, eliminating the reliance on the sensory availability of today's temperature11, 12, 13, 14. This prompt should sever the relationship between today's temperature and last year's estimate.
I conducted the study on a convenience sample (as did Zaval et al.6) in Evanston, Illinois on 29 September 2014. Participants were recruited in person by approaching each potential respondent and asking whether he/she would complete a brief paper survey with no identifying information. As mentioned, participants were randomly assigned to the non-prompt or prompt condition. A prerequisite for participation was that the individual had lived in the area during the previous winter. The temperature on the day of the study registered a relatively high 80 °F, compared with a normal high of 70 °F. The previous winter was the coldest in the last 30 years and led to a number of school closings, atypical for the area. The use of a single sample/location on a particular day has the advantage of ensuring control over actual temperatures, thereby offering a test for the conditions of the local warming effect. Future work, however, should explore the impact of different prompts with distinct samples and locations where the daily temperature is not clearly high.
I present the results in Table 1, with a column for each condition. The first row reveals that, not surprisingly given the warmth of the day, the average for both groups on the 'today's temperature' (TT) question was near 4 on the scale. No participants rated it as 1 and only six rated it as 2. The next two rows reveal differences in the percentage of warm days (PDW) last year, and more importantly, the correlation between PDW and TT. Today's temperature substantially correlates with past year's estimates for the non-prompt group (0.38) at a level similar to that reported by Zaval et al.6. This relationship does not exist in the prompt group. The next two rows reveal strong relationships between global warming belief and concern with PDW, with similar correlations for both groups. |