Table 2.

Summary of results.

Outcome	General findings	Skill differences
		Less-skilled cohort	More-skilled cohort
Attitudes
	Participants hesitated to use online health data (20/22, 91%), in particular to make inferences	Rationale: the plethora of data cannot be managed (6/12, 50%)	Rationale: the quality of online data is low (8/10, 80%)
	Although data quality was a matter of concern...	...keywords may override distrust toward sites (6/12, 50%)	...once accessed, data from any site were used (4/10, 40%)
Technical skills
Search implementation	Reliance on search engine Google.de (22/22, 100%)	No differences observed	Few entered URLs directly (2/10, 20%)
	Use of search terms (20/22, 91%); often misspelled; rarely corrected	83% used a single search term (10/12); 67% used search term suggestions (8/12); 17% used natural language phrases (2/12)	100% used two or more search terms (10/10)
Site selection	Reliance on first 5 links on first search engine result page (20/22, 91%)	Selection based on keywords (often unrelated to original search) inferred from links/excerpts (7/12, 58%)	Selection based on data sources inferred from URL or links/excerpts (9/10, 90%)
Site navigation	Relevance of website contents was appraised	Text was read rather than scanned. In 23% of searches (14/61) links were followed up.	Text was scanned for keywords. In 6% of searches (3/49), links were followed up.
	Information was rarely cross-referenced	No cross-referencing or use of tabs (0/12, 0%).	Use of multiple tabs to compare results (2/10, 20%).
Cognitive strategies (based on inference scenarios only)
Search intentions	People searched for online contents related to personal, a priori opinions, knowledge, cues, or expert opinions.	Distribution of intentions: 70% a priori opinions (21/30); 10% cues (3/30); 10% knowledge (3/30); 10% expert opinions (3/30)	Distribution of intentions: 14% a priori opinions (4/28); 29% cues (8/28); 29% knowledge (8/28); 29% expert opinions (8/28)
Information evaluation	Information was trusted if consistent with search intentions—that is, if...	...a website confirmed a priori opinions (21/30, 70%) or yielded search contents (9/30, 30%)	...a website confirmed a priori opinions (4/28, 14%) or yielded search contents (24/28, 86%).
Stopping rule	Search was stopped once the first piece of online information satisfied search intentions	No participant further cross-referenced (0/12, 0%)	20% further cross-referenced information (2/10)
Inference rule	Participants were hesitant to make inferences based on online searches, except when they searched to confirm personal, a priori opinions	In 73% of the inference queries, inferences were made based on a priori opinions (22/30). In 27% no inferences were made (8/30).	In 14% of the inference queries, inferences were made based on a priori opinions (4/28); in 7% based on cues (2/28). In 79% no inferences were made (22/28).