When Social Media Images and Messages Don’t Match: Attention to Text versus Imagery to Effectively Convey Safety Information on Social Media

Figures & data

Figure 1. Example of mismatched (A) and matched (B) social media posts on infant safe sleep recommendations*

*modified to de-identify the health agency name

Table 1. Characteristics of parents of young children (n = 150)

Sample characteristics	Percent (n)
Female	70.7 (106)
Mean age in years (SD)	32.8 (5.9)
Race White African American/Black Asian More than one race	52.0 (78) 37.3 (56) 2.7 (4) 6.7 (10)
Educational attainment ≤High school/GED Associates/some college ≥College degree	16.0 (24) 34.0 (51) 50.0 (75)
Household characteristics
Two parent household	74.0 (111)
Annual income ≤$60 k	48.0 (72)
Children’s age ranges Under 5 only Over 5 only Younger and older mix	40.7 (61) 18.0 (27) 41.3 (62)
Mean # of daily social media actions (SD) Facebook Pinterest Instagram Twitter	18.9 (25.3) 8.4 (14.5) 18.1 (38.6) 10.0 (6.5)
Health literacy Newest Vital Signs Limited literacy Possible limited literacy Adequate literacy	0.7 (1) 14.7 (22) 84.7 (127)

SD = Standard deviation

Table 2. Visual attention to safety-related social media posts on among parents (n = 150)

Aggregated safety-related posts, by matching status	Seconds (SD)	Mean % of viewing time*
Matched post Text Imagery	13.5 (4.6) 6.5 (2.8) 5.3 (1.9)	47.7% (10.0) 39.2% (8.3)
Mismatched post Text Imagery	12.3 (4.1) 5.5 (2.4) 3.3 (1.5)	43.1% (10.5) 27.0% (8.1)

*totals less than 100% due to attention to white space

SD = Standard deviation

Table 3. Associations between matched imagery and textual elements and visual attention and correct recognition of safety recommendations* (n = 150)

Variables	% visual attention on imagery	% correct recognition	Effect size^	p-value
Matched (vs. mismatched)	12.1 (0.71)	–		<0.001
Matched	–	3.2 (0.66)	0.13	<0.001
Mismatched	–	−1.1 (0.49)	0.02	0.02

*Adjusted for sex, health literacy score, and social media engagement

^Omega squared