Shortening a survey and using alternative forms of prenotification: Impact on response rate and quality

Table 3 Logistic Regression Results, odds ratios comparing likelihood of response.

Variable	OR	95% CI		P-value
Main effects only
Notify: Postcard vs. Letter	1.04	0.77	1.39	0.80
Pages: 2 vs 4	0.81	0.60	1.09	0.16
Age: 1 year increase	1.03	1.01	1.04	0.0001
Sex: Female vs Male	1.86	1.38	2.50	<.0001
Race: White vs non-white*	3.35	1.77	6.35	0.0002
Race: White vs other/unknown	2.23	1.49	3.35	0.0001
Including interaction between survey length and age
Notify: Postcard vs. Letter	1.05	0.78	1.42	0.73
Pages: 2 vs 4 (for 25-year-olds)	0.34	0.19	0.62	0.0004
Pages: 2 vs 4 (for 35-year-olds)	0.53	0.36	0.79	0.0015
Pages: 2 vs 4 (for 45-year-olds)	0.84	0.62	1.13	0.25
Pages: 2 vs 4 (for 55-year-olds)	1.32	0.87	2.00	0.19
Sex: Female vs Male	1.86	1.38	2.50	<.0001
Race: White vs non-white*	3.40	1.79	6.47	0.0002
Race: White vs other/unknown	2.33	1.55	3.50	<.0001

Notes: Age was modeled continuously, therefore, the significant interaction in the second model was summarized by calculating the odds ratios at select points across the age range. P-value for survey length-by-age interaction = 0.001. There was no significant interaction found between notification method and survey length (p = 0.25).
* "Non-white" includes: American Indian/Alaskan Native, Black or African American, Native Hawaiian/Pacific Islander, and Asian.

ISSN: 1471-2288