Table C11.
Self-isolation intentions: multinomial logit estimates, ordering effects.
| (1) | (2) | (3) | (4) | (5) | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Sample: | Round 3 | Round 3: first observed scenario | Round 3 | Round 3 | Round 3: first observed scenario | ||||||
| Option: | Increase compliance | Decrease compliance | Increase compliance | Decrease compliance | Increase compliance | Decrease compliance | Increase compliance | Decrease compliance | Increase compliance | Decrease compliance | |
| Scenario: | A few months | 1.553** | 1.161 | 1.987* | 1.297 | 1.300** | 1.759*** | 1.388** | 1.320** | 2.095* | 1.551 |
| (0.295) | (0.201) | (0.819) | (0.496) | (0.133) | (0.167) | (0.206) | (0.160) | (0.831) | (0.511) | ||
| Until needed | 1.525 | 0.886 | 1.415 | 0.987 | 1.108 | 1.674*** | 1.192 | 1.216 | 1.864 | 1.839* | |
| (0.503) | (0.214) | (0.636) | (0.446) | (0.135) | (0.157) | (0.238) | (0.157) | (0.730) | (0.606) | ||
| Length of hypothesized extension, relative to expectations (five categories): | Much shorter | 1.475 | 1.201 | 0.789 | 1.954 | ||||||
| (0.708) | (0.483) | (0.812) | (1.902) | ||||||||
| Shorter | 0.767 | 1.012 | 0.406 | 0.297** | |||||||
| (0.271) | (0.323) | (0.316) | (0.171) | ||||||||
| Longer | 0.944 | 1.755*** | 0.472* | 0.819 | |||||||
| (0.180) | (0.339) | (0.211) | (0.329) | ||||||||
| Much longer | 0.679 | 2.656*** | 0.754 | 2.185* | |||||||
| (0.246) | (0.780) | (0.328) | (0.991) | ||||||||
| Length of hypothesized extension, relative to expectations (three categories): | Much shorter/Shorter | 0.967 | 1.084 | 0.461 | 0.487 | ||||||
| (0.330) | (0.323) | (0.320) | (0.269) | ||||||||
| Much longer/Longer | 0.870 | 1.964*** | 0.567 | 1.196 | |||||||
| (0.202) | (0.393) | (0.221) | (0.417) | ||||||||
| Ordering: | "Few months" question first | 1.238 | 0.962 | ||||||||
| (0.391) | (0.226) | ||||||||||
| "Indefinitely" question first | 1.254 | 1.208 | |||||||||
| (0.388) | (0.279) | ||||||||||
| Observations | 2646 | 882 | 2646 | 2646 | 882 | ||||||
| F-statistic | 3.045 | 2.080 | 2.570 | 2.540 | 1.750 | ||||||
| Respondents | 882 | 882 | 882 | 882 | 882 | ||||||
Notes: In all regressions, the baseline option is “maintain current behavior”, and the omitted lockdown length scenario is “a few weeks”. In specifications (1) through (5), the omitted relative expectations length is “scenario matches with expectations”. We report the estimates as relative risk ratios. Each respondent selected one option from each of three questions, therefore there are up to three observations per individual in the regression sample. The control variables include gender, age, education, whether the respondent lives alone or with her/his parents, marital status, whether the respondent has children, whether s/he is currently employed, whether the respondents’ household is currently facing economic difficulties, and whether they live in one of the “Red Zone” regions. Model (1) include all observations from the third round, for which we have information on ordering effects. Model (2) limits the observations of the third round to one per respondent, the one relative to the first hypothetical scenario they considered. In Model (3) we include again all responses in the third round, and include among the regressors indicators for what was the extension scenario that each respondent saw first. Models (4) and (5) include the same samples as models (1) and (2), respectively, but the indicators for the different scenario-expectations (mis)matches are more aggregated to account for the smaller sample size. Linearized errors are in parenthesis. *** p<0.01, ** p<0.05, * p<0.1.