no use reasoning with adolescents? a randomised controlled trial comparing persuasive messages
TRANSCRIPT
03.05.2023 1Twitter: @heinonmattiFacebook: Heinon Matti
SMS reminders to increase accelerometer
wear-timeA within-trial RCT comparing persuasive messages
Reg. no: DRKS00007721
03.05.2023 2Twitter: @heinonmattiFacebook: Heinon Matti
Background
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 3
• Physical activity recommendations based on self-reported activity levels → problems:• Remembering past activity• Reporting “what the researcher wants to hear”
• Solution: objective measurement devices• New problem: need to wear it for most of the study
period!‒ E.g. if you only wear the activity device when exercising,
looks like 100% of your day was spent working out!
The ”why”
?
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
The ”why”
• Let’s Move It ‒ A school-based multilevel intervention to
increase physical activity and decrease sitting among youth*
‒ Ongoing since 2012‒ RCT phase from 2015 to 2017
‒ Ca. 16–19 year-old vocational school students
‒ Waist-worn accelerometers used
*Hankonen, Heino et al. (in preparation). Randomised controlled feasibility study of a school-based multi-level intervention to increase physical activity and decrease sedentary behaviour among older adolescents.
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
The Pervasive Problem: non-wear in accelerometry
• Let’s Move It Feasibility study• Students fell short of accelerometer wear
targets (>10hrs of data on >4 days)‒ Qualitative work: non-wear attributed to
forgetting
How do we ensure adequate accelerometer wear times?
03.05.2023 6Twitter: @heinonmattiFacebook: Heinon Matti
Methods
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
The setup
• Within-trial RCT during internal pilot study of the main trial
• Participants wear the accelerometer for seven consecutive days during Let’s Move It baseline data collection
• Fight forgetting with (SMS) reminders• Next question: What kind of reminders?
Could an old copy machine
help here?
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Langer, Blank and Chanowitz (1978) Mixed success w/ replication: Folkes (1987);Key, Edlund, Sagarin and Bizer (2009)
”Harnessing the power of ’Because’”…?
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Message types: an example
”Reason”, day 3:”Succinct”, day 3:
“Hello! Because the study wouldn't succeed without your help, please remember to put on the motion measurement device again and wear it until you go to sleep (except in the shower etc.) - thanks!”
[emphasis added]
“Hello! Please remember to put on the motion measurement device again and wear it until you go to sleep (except in the shower etc.) - thanks!”
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Consenting in Let’s Move It accelerometryN=375
Opting in SMS n=276
Randomised to ”Reason” n=138
Randomised to ”Succinct” n=135
Opting out n=95
Participants:
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 11
• Does providing a reason via SMS influence accelerometer wear time:
‒ Total wear time‒ Number of days >10 hours of data accumulated (0-7)
Research questions
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 12
• ”Probability of observed (or rarer) data, if null hypothesis is true” • (also assumes randomisation, stopping rules etc…)
What’s a ”p-value” again?
When p is high (eg. p=0.32), no conclusions can be drawn!
(Dienes, 2014)
Reactions upon discovering this can vary.
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 13
• remember, p-value: Probability of data, given H0• A Bayes factor BF01:
A better question?”Which is more probable, null or alternative?
0 …∞1 31/3
Very roughly:
BF01:
Data favor alternative
Insufficient data
Data favor null
A great explanation: http://alexanderetz.com/2015/11/01/evidence-vs-conclusions/
When < BF < 10, evidence quite weak
03.05.2023 14Twitter: @heinonmattiFacebook: Heinon Matti
Results
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Total wear time
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 16
Total weartime differences
Bayesian ANOVA gives us BF01=29.03
1:1 (50%) prior odds become 1:29, p(effect)=3%
10:1 (91%) prior odds become 10:29, p(effect)=26%
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Valid wear day percentages
Χ2(7) = 7.893, p = 0.342
BF01 = 1.27BF01 = 7.09
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 18
Measurement days of >10 hours of valid data
Horizontal lines represent means, with shaded 95% Bayesian Highest Density Intervals (HDIs)
03.05.2023 19Twitter: @heinonmattiFacebook: Heinon Matti
Discussion
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Conclusions
• Why didn’t reasons help?• Hidden moderators blah blah?• SMS format hinders the effect? (Why?)• Wearing the device a question of capability, not
motivation?• No use reasoning with adolescents?
‒ E.g. university students more compliant
• A case of an undead theory? ‒ Ferguson, C. J., & Heene, M. (2012). A Vast Graveyard of Undead
Theories.
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Conclusions
• Why didn’t reminders help?• Operationalisation failure?
‒ Messages claimed to have been read but no objective log data
• Self selection? ‒ Unlikely, as opting in was mostly determined by recruitment
prompt
• Non-wear not due to remembering?‒ Although they said it was and thought the reminder helped
immensely!
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com 03.05.2023 22
The Lakatos principle:
Meehl, P. E. (1990). Appraising and amending theories: The strategy of Lakatosian defense and two principles that warrant it. Psychological Inquiry, 1(2), 108–141.
“Accepting the neo-Popperian view that it is inadvisable to persist in defending a theory against apparent falsifications […] the rationale for defending by non-ad hoc adjustments lies in the theory having accumulated credit by strong successes, having lots of money in the bank.“ – Paul Meehl
- Does the “power of because” lean on Monopoly money?
Twitter: @heinonmattiFacebook: Heinon Mattimattiheino.wordpress.com
Thank you!
@heinonmatti
mattiheino.wordpress.com
Questions?
Comments?
Ideas?
https://linkedin.com/in/heinonmatti