Table A1.
Significant predictors of screener response propensity in the final discrete time logit model for call-level data from the eight most recent quarters, after applying backward selection (n = 119,981 calls; Nagelkerke pseudo R-squared = 0.09; AUC = 0.66).
| Predictor | Coefficient | Standard Error |
|---|---|---|
| Intercept | −2.56 | 0.32 |
| Mail Delivery Point Type: Missing | 0.08 | 0.03 |
| Mail Delivery Point Type: A | 0.03 | 0.02 |
| Mail Delivery Point Type: B | −0.04 | 0.03 |
| Mail Delivery Point Type: C | −0.09 | 0.03 |
| Interviewer-Judged Eligibility: Missing | 2.46 | 0.10 |
| Interviewer-Judged Eligibility: No | 0.63 | 0.07 |
| Segment Listed: Car Alone | 0.03 | 0.02 |
| PSU Type: Non Self-Representing | 0.06 | 0.03 |
| PSU Type: Self-Representing (Not Largest 3 MSAs) | 0.03 | 0.03 |
| Previous Call: Contact | 3.97 | 0.28 |
| Previous Call: Different Window | −0.12 | 0.02 |
| Previous Call: Building Ever Locked | 0.32 | 0.05 |
| Previous Call: Building Locked | 2.16 | 0.14 |
| Previous Call: Strong Concerns Expressed | 0.26 | 0.04 |
| Previous Call: No Contact | 2.26 | 0.13 |
| Previous Call: Other Contact, No Concerns Expressed | −1.35 | 0.25 |
| Previous Call: Concerns Expressed | −1.58 | 0.26 |
| Previous Call: Soft Appointment | −1.03 | 0.30 |
| Previous Call: Call Window Sun.-Thurs. 6pm-10pm | 0.07 | 0.03 |
| Previous Call: Call Window Fri.-Sat. 6pm-10pm | 0.08 | 0.02 |
| No Access Problems in Segment | −0.05 | 0.02 |
| Evidence of Other Languages (not Spanish) | −0.09 | 0.03 |
| Census Division: G | −0.14 | 0.03 |
| Census Division: B | −0.32 | 0.03 |
| Census Division: D | −0.22 | 0.03 |
| Census Division: H | −0.24 | 0.03 |
| Census Division: C | −0.20 | 0.03 |
| Census Division: F | −0.27 | 0.04 |
| Census Division: E | −0.20 | 0.03 |
| Census Division: A | −0.19 | 0.04 |
| Contacts: None | −0.68 | 0.24 |
| Contacts: 1 | −0.54 | 0.22 |
| Contacts: 2 to 4 | −0.42 | 0.19 |
| Segment Domain: <10% Black, <10% Hispanic | −0.04 | 0.02 |
| Segment Domain: >10% Black, <10% Hispanic | −0.04 | 0.02 |
| Segment Domain: <10% Black, >10% Hispanic | 0.01 | 0.03 |
| Percentage of Segment Non-Eligible (Census Data) | −0.01 | <0.01 |
| Interviewer-Estimated Segment Eligibility Rate | −0.55 | 0.12 |
| Interviewer-Estimated Household Eligible | −0.09 | 0.02 |
| Segment Type: All Residential | 0.04 | 0.02 |
| Log(Number of Calls Made) | −0.60 | 0.03 |
| Log(Number of Calls Made) x No. Prev. Contacts | −0.04 | 0.01 |
| CML* HoH Age: 35–64 | −0.12 | 0.02 |
| CML Adult Count: Missing | −0.13 | 0.04 |
| CML Adult Count: 1 | −0.09 | 0.03 |
| CML Adult Count: 2 | 0.01 | 0.03 |
| CML Asian in HH: Missing | 0.21 | 0.04 |
| CML Asian in HH: No | 0.20 | 0.05 |
| CML HoH Gender: Missing | −0.03 | 0.02 |
| CML HoH Gender: Female | −0.01 | 0.02 |
| CML HoH Income: $35k-$70k | 0.12 | 0.02 |
| CML HoH Income: less than $35k | 0.14 | 0.02 |
| CML HH Own/Rent: Missing | −0.06 | 0.03 |
| CML HH Own/Rent: Owned | −0.02 | 0.02 |
| CML Age of 2nd Person: Missing | −0.13 | 0.03 |
| CML Age of 2nd Person: 18–44 | −0.15 | 0.03 |
| No Respondent Comments | 0.08 | 0.04 |
| Non-Contacts: None | −0.51 | 0.08 |
| Non-Contacts: 1 | −0.25 | 0.05 |
| Non-Contacts: 2–4 | −0.03 | 0.03 |
| Occupancy Rate of PSU | −0.26 | 0.10 |
| Respondent Other Concerns | 0.18 | 0.06 |
| Physical Impediment to Housing Unit: Locked | −0.35 | 0.03 |
| Day of Quarter | 0.01 | <0.01 |
| Respondent Concerns Expressed: None | −1.25 | 0.15 |
| Respondent Concerns Expressed: Once | 0.15 | 0.09 |
| Single Family Home / Townhome | −0.22 | 0.03 |
| Structure with 2–9 Units | −0.29 | 0.04 |
| Structure with 10+ Units | −0.21 | 0.04 |
| Respondent Concern: Survey Voluntary? | −0.46 | 0.15 |
| Respondent Concern: Too Old | 0.60 | 0.15 |
*
CML denotes that the variable came from a commercial data source.