Why is ma reaction time slower on a phone than on a computer?

Thir are arithmetic examples, no published corrections for phone-versus-desktop. A stimulus micht wait for a refresh, but its full display delay depends an renderin an the screen itsel. Input registerin gangs by a separate route. Measurements across online platforms shaw that accuracy an precision vary across combinations o equipment an software. Anwyl-Irvine et al., 2020 Bridges et al., 2020

Whit does the time shown include?

The recorded interval sterts an ends at software events. It is meant tae approximate the time atween seein a cue an giein a response, but the physical events can happen efter the software timestamps suggest. A laboratory study shaws the size o this difference: Woods an colleagues reported a measured simple reaction time o 231 ms, an 213 ms efter correctin for their equipment. Their 18 ms correction applies tae that apparatus, no tae yer phone. Woods et al., 2015

Tappin a screen, clickin a moose an pressin a key need different movements an aa. Changin the response changes the task, even gin the page looks the same. A choice-reaction experiment adds response selection, sae ye cannae compare it directly wi a simple tap-on-green task. Woods et al., 2015

Why a steadier average isnae aye a mair accurate average

Accuracy means bein close tae the true physical interval; precision means bein consistent. A device can be consistently delayed, giein a steady but offset score. Repeatin the test helps estimate variation, but disnae tak awa a fixed hardware offset. This difference is central tae the experimental timing comparisons. Anwyl-Irvine et al., 2020 Bridges et al., 2020

Whit else changes when ye change devices?

Screen size, posture, finger position an familiarity can change alang wi the hardware. Age differences reported in a controlled study cannae be uised tae infer that a hardware-caused score difference reflects biological ageing. Comparin yersel wi somebody else's different device brings the same uncertainty. Woods et al., 2015 Woods et al., 2015

Reaction-time distributions include unusually slow responses an aa. Ae lapse can shift the mean a guid bit in a short test. Sae outlier rules, an whether ye uise the mean, median or fastest response, matter; choosin whichever summary looks best efter the run weakens the comparison. Ratcliff, 1993

A useful way tae compare yer ain runs

Keep a separate record for ilk device an input method. Compare like wi like, includin the number o attempts an the summary measure. Repeatin a set protocol is easier tae interpret than switchin devices back an forrit an keepin the lowest score. Experimental work on trial numbers shaw that the richt amount o data depends on whit is bein estimated. Miller, 2023

Practice can improve performance on repeated cognitive tasks an aa. It is reasonable tae expect familiarity tae matter, but a general retest meta-analysis disnae tell us the exact practice gain for this browser gemme. Nor does a muckle difference atween twa runs prove a reliable personal trait: the reliability o cognitive tasks maun be measured, no just assumed. Scharfen et al., 2018 Hedge et al., 2017

Sae there is nae evidence-backed rule like takkin the same delay aff ilka phone result. Calibration needs the actual display an input system tae be measured. A refresh-rate specification alane isnae enough. Anwyl-Irvine et al., 2020 Bridges et al., 2020