3-Factor?

Moving from 7-Factor to 3-Factor Model

Executive Summary

Exploratory factor analysis of pilot data (N=65) suggests the digital health enablement scale comprises three factors that are theoretically coherent. The following sets out the possible options for proceeding with a 3-factor scale.

Key Recommendation

Reduce the scale from 45 items across 7 factors to 22-26 items across 3 factors, maintaining comprehensive construct coverage while reducing respondent burden.

Three-Factor Model Overview

Factor 1. Digital Health Motivation & Acceptance

Central Question: “Do I want to use digital health technologies?”
Content: Attitudes, trust, outcome expectations, behavioral intentions
Current Items: 23 items → Target: 8-10 items

Factor 2: Technical Skills & Self-Efficacy

Central Question: “Can I technically perform the required tasks?”
Content: Concrete skills, troubleshooting confidence, technical competence
Current Items: 13 items → Target: 6-8 items

Factor 3: Digital Health Context & Support

Central Question: “Is my environment conducive to success?”
Content: Social support, environmental barriers, external pressures
Current Items: 8 items → Target: 8-10 items

Current Status Assessment

Factor	Current Items	Status	Key Issues	Target Items
Factor 1 Motivation & Acceptance	23	⚠️ Too many	Redundancy, weak items	8-10
Factor 3 Technical Skills	13	✅ Good foundation	Minor tweaks needed	6-8
Factor 2 Context & Support	8	⚠️ Needs strengthening	Few strong items, poor wording	8-10
Total	44			22-28

Factor-Specific Recommendations

Factor 1: Digital Health Motivation & Acceptance

This factor currently contains too many items (23) with too much redundancy. The strongest items have high factor loadings (>0.80) and clear theoretical alignment.

Items to Retain

Strong performers with factor loadings >0.80:

“I would use a fitness app because I am motivated to improve my health” (loading: 0.914)
“I would enjoy using a fitness app to improve my health” (loading: 0.889)
“I believe that using a fitness app would be good for me” (loading: 0.849)
“I can’t see why I should bother using a fitness app” (loading: 0.840) [reverse-scored]
“I believe that using a fitness app would improve my physical health” (loading: 0.826)
“I don’t think a fitness app would work for me” (loading: 0.819) [reverse-scored]

Possible items to remove

Weak loadings (loadings <0.40):

Item 18: “I would expect health technologies to have measures in place to keep my information safe”
Item 12: “Those close to me would help me use a new health technology if needed”
Items 14, 16: Conceptually unclear and weak loadings

Redundant items:

Items 46, 47: Too similar to Item 44 (“would be good for me”)

Questionable construct fit:

Trust items (17, 19, 21): Consider whether these belong in motivation or should form separate trust subscale.

Is Trust a Separate Factor?

The trust items (17, 19, 21) load on Factor 1 but may conceptually belong elsewhere. Consider:

Keep in Factor 1 if trust is viewed as prerequisite for motivation
Move to separate trust subscale if distinct construct measurement needed
Remove entirely if outside scope of digital readiness

Factor 2: Technical Skills & Self-Efficacy

This factor has the strongest foundation with several high-loading items. Primary needs are minor refinements and removal of cross-loading items.

Items to Retain

Strong performers (loadings >0.70):

“I can troubleshoot basic issues with digital technology” (loading: 0.926)
“I can usually use digital technology” (loading: 0.827)
“I am confident I could use the advanced features of a fitness app if I wanted to” (loading: 0.772)

Moderate performers with good content validity:

“I often struggle to use digital technology” (loading: 0.636) [reverse-scored]
“I can usually adjust the settings on digital technology to suit my needs” (loading: 0.578)
“I can usually download apps” (loading: 0.575)

Items Requiring Attention

Items 2, 3 (health information evaluation): Weak loadings, consider removal
Item 25 (cross-loads): Rewrite to focus purely on technical confidence
Item 26 (“confident I could use a fitness app regularly”): Cross-loads with Factor 2, move to Factor 3 or remove

Factor 3: Digital Health Context & Support

This factor requires the most substantial revision. Current items are few in number, often negatively worded, and don’t adequately cover the construct breadth.

Current Issues

Limited coverage: Only 8 items for complex construct
Negative wording dominance: Creates response bias risk
Weak factor loadings: Several items <0.40
Conceptual gaps: Missing key support dimensions

Recommendations for Current Items

Strengthen existing items by converting from negative to positive framing:

Current Item	Possible revision
“I don’t have anyone in my life that could help me use a health technology”	“I have people who could help me if I struggled with a health technology”
“Few people I know use health technologies”	“People in my social circle regularly use health technologies”

Revise problematic items:

Item 32: “I wouldn’t use a fitness app because I find technology frustrating” → “I find the thought of using health technologies overwhelming”

Reintroduce items?

To strengthen construct coverage consider reintroducing items that were previously tested during cognitive interviews. E.g.:

“Those close to me would support my use of a health app”
“Using health apps is normal in my social groups”
“Others like me already use health apps”
“Those close to me would encourage me to use a health app”
“Most people I know use health apps”
“Those close to me would help me getting going with a health app”
“Those close to me would encourage me to keep going with a health app”

N.B. Need to avoid overusing the wording ‘Those close to me…’ to avoid method effects. Consider instead, ‘People I know…’, ‘People in my life…’, etc.

Cross-Cutting Recommendations

Reduce “Fitness App” Over-Reliance

Current issue: Excessive reliance on “fitness app” terminology creates method effects.

Method Effects

A method effect (also called method variance or common method bias) occurs when the way you measure something artificially inflates or creates correlations between items, rather than those correlations reflecting true relationships between the underlying constructs.

Solution:

Mix specific (“fitness app”) with general (“health technology”) items
Improves generalisability and reduces semantic similarity bias
Maintains concrete examples while broadening scope

Address Cross-Loading Items

Items requiring resolution:

Item 26: “confident I could use a fitness app regularly” - clarify factor placement
Item 4: “understand health information from health technologies” - decide between Factor 2 vs Factor 3

Balance Response Patterns

Ensure adequate distribution of:

Positively vs negatively worded items across factors
Specific vs general technology references
Current capability vs future confidence items

Maintain Quality Control Features

Attention check items (22-23, 48-49) performed well in pilot - retain in final version.

Proposed Final Item Distribution

Proposed Item Distribution Changes
Factor	Current Items	Keep	Remove	Add New	Final Target
Factor 1: Motivation & Acceptance	23	6	17	0	8-10
Factor 2: Technical Skills	13	6	7	5-7	6-8
Factor 3: Context & Support	8	3	5	0	8-10
Attention Checks	4	4	0	0	4
TOTAL	48	19	29	7-13	26-32

Next steps

Consider piloting revise scale with 30-50 online participants to assess whether three-factor model replicates.

Pilot data can be folded into main study if there are no further changes to the items.

Safeguards

Conservative wording for new items using established language
Consider over-sampling (N=600, e.g. 450 online, 150 offline) to enable split sampling.
Early monitoring of first 200 responses for issues in main study

Contingency Planning

Split-sample buffer: Use development sample to refine before cross-validation
Item removal protocol: Predetermined criteria for dropping poor performers
Alternative models: 2-factor solution ready if 3-factor doesn’t replicate