Observation, Description, and Prediction of Long-Term Learning on a Keyboarding Task by Mark L. McMulkin Thesis Submitted to the Faculty of the Virginia Polytechnic Institute and State University in Partial Fulfillment of the Requirements for the Degree of Masters of Science in Industrial and Systems Engineering APPROVED: —— AAlr / CHE. Kroginer, Chairman FT - NY \ i. % \ fo \ ‘ \ Hl ‘ \ oN vs | \ \ a ‘ “ye ' ( }\ wy pM? \ Te . SN SS \ Ry ¥ \ NX , MS “TRC. Wiliges | .C. Woldstad Blacksburg, Virginia January, 1992
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Observation, Description, and Prediction of Long-Term Learning on a Keyboarding Task
by
Mark L. McMulkin
Thesis Submitted to the Faculty of the Virginia Polytechnic Institute and State University
in Partial Fulfillment of the Requirements for the Degree of
Masters of Science
in
Industrial and Systems Engineering
APPROVED:
——
AAlr / CHE. Kroginer, Chairman
FT - NY
\ i. % \ fo \ ‘
\ Hl ‘ \ oN vs | \ \ a ‘
“ye ' ( }\ wy pM?
\ Te . SN SS \ Ry ¥ \ NX , MS
“TRC. Wiliges |
.C. Woldstad
Blacksburg, Virginia
January, 1992
Lv
SoS >
VAST
J 912
[226 cy
Observation, Description, and Prediction of Long-Term
Learning on a Keyboarding Task
by
Mark Lee McMulkin
Committee Chairman: Karl H. E. Kroemer
Industrial and Systems Engineering
(ABSTRACT)
Three major principles of learning a chord keyboarding task were
investigated. Five subjects were taught 18 characters on a chord keyboard,
then practiced improving their keying speed for about 60 hours.
The first objective of the study was to observe long-term learning on a
keyboarding task. The performance, in characters typed per minute, was
recorded over the entire range of the experiment. Typing skill improved quickly
in the beginning and then slowed, but performance had not reached a stable
peak by the end of the experiment.
The second objective of this study was to determine a function that
describes performance progress from initial training to a high keying speed.
Five functions were evaluated; a function which predicts the logarithm of the
dependent variable (characters per minute) from the logarithm of the regressor
variable provided a good fit to the actual data. The final form of the equation
was CPM; = ePo7 Br where CPM; = performance in characters per minute on
the i-th interval, T; = the i-th interval of practice, and By and B, are fitted
coefficients.
The second objective also considered the form that T; (from the above
equation) should take. Performance can be predicted from number of
repetitions such as trials, or from amount of practice such as hours. Both trials
and time were used as predictor variables and both provided equally accurate
predictions of typing speed. Both also provided excellent fits in conjunction with
the Log-Log equation. Thus, it appears the Log-Log function is fairly robust in
predicting performance from different variables.
The third objective was to investigate how many trials of performance are
needed before the entire learning function can be reasonably determined. In
this experiment, subjects practiced for an extended period of time (about 60
hours) so a fairly complete progression of performance could be gathered. Yet,
it would be more convenient to collect data for only a few hours and deduce the
ensuing performance of the subject. The coefficients of the Log-Log function
were determined using only the first 25, 50, 100, 150, and 200 of the initial
performance points (out of about 550 total actual data points). The mean
squared error (MSE) was calculated for each of these fits and compared to the
MSE of the fit using all points. It appears that at least 50 performance data
points are required to reduce the error to a reasonably acceptable level.
ACKNOWLEDGEMENTS
This work was possible due to the funds and facilities provided by the
Center for Innovative Technology (CIT) of Virginia, the Department of Industrial
and Systems Engineering, in particular, the Industrial Ergonomics Laboratory,
and Vatell Corporation.
| would like to extend my appreciation and thanks to my committee chair,
Dr. Karl Kroemer, for his encouragement, assistance, time, and support
throughout this project and my entire Masters Degree education. | would also
like to thank my other committee members, Dr. Jeff Woldstad and Dr. Robert
Williges, for their advice and guidance on this project which were instrumental
to the completion of this thesis.
Finally, | would like to thank my fellow graduate students for their help
during my thesis, particularly Chris Rockwell for his assistance in
troubleshooting the experimental set-up.
TABLE OF CONTENTS
ADSHACE cee eeessssecssssccsesnescsccvsnsecesssscsensecsssascesssesesnensucensesensnaseesnansetsoeeseseasenenanaeesenes il
ACKNOWlECQMENMS ..........:cccesecessceceteesseceeseesesecesseeestececseccssneeseaeeeeseecueneessaeeesseesseneceenens iv
List Of ADPONdICES ou... eee eesseseeesssecceecncessceeesetseseessensnsesneecaaeseseseeesesesessteneessneeeness vill
List Of T€ADIOS oo... cceccecssccessensessceessseeceeeseeesecesessseaecesaaeesneeesessaeesesaeesenenseesnnseenseenes ix
List Of FIQUIES oo... cetecseeeeseessecseesseeeessecsseeeseeeeesacseecseceeseeceseeaeeeaeeeetuaesnaesnaseagens X
2.3 Alternative Layouts of the Standard Keyboard ..............ccccsssssssseeeeees 14 2.3.1 StrONG(1956) ee ceccsecseeesssseseesseerseseseeceeesseeessuesseesseeessnees 16
2.4 Binary Chord Keyboard ou... ccecsesssesssesscssecsesenseseessesseeeecessecsaeceeeceaseas 16 2.4.1 Kl@MMEr (1958) oo. ccceesessseecscseeccsseceessesessesseeetsneeesseeseseeens 17 2.4.2 Lockhead and KlemmMer (1959) ou... ccc cesssssceeeseessseeseeeseeeeees 19 2.4.3 Ratz and Ritchie(1961) and Seibel (1962) 0... 20 2.4.4 Cornog, Hockman, and Craig (1963) ..........c ce ececesseseeeeeeseees 20 2.4.5 Conrad and Longman (1965) .0.........cccceceeseseeeeeesesssneecereeseees 22 2.4.6 Bowen and Guinness (1965) .........cccesssscccesssseecessssseeeeeeseees 24 2.4.7 SidOrsky (1974) ooo... ccceccecscsccesssecssseeesseeeessceceeeeeeeeneeseceneesssneenss 25 2.4.8 Bequaert and Rochester (1977) and Rochester,
Bequaert, and Sharp (1978) ...........cccececsessennececceeeeeceeeeeeseeees 26 2.4.9 Gopher and Eilam (1979) ..........ccccecssseseceeeeseneseteessssneeeeeesees 28 2.4.10 Gopher, Hilsernath, and Raij (1985); Gopher (1986);
and Gopher and Raij (1988) ....... cc eeeeeeceeeseeseessesneeees 29
2.4.11 Richardson, Telson, Koch, and Chrysler (1987) ..............
2.5 Ternary Chord Keyboard Experiments .........eeeceeeecesceeceseeeeeceeeeeaeens 2.5.1 Fathallah (1988) - TOK #1... cccccccsccesseccesssseeeeesseesesssneeeees 2.5.2 Kroemer (1990) - TOK #2 oe cecccscsscceceeenseeseceesneceeeenersennees 2.5.3 Kroemer (1990) - TCK #3 uc ccecssessesecesseeseesssessesaneseeeseens 2.5.4 McMulkin and Kroemer (1991) and
Lee and Kroemer (1991) oe eeeseescessessssssesseesseeseeseeeeenens
2.6 Length of Training S@SSION ........ eee eeecesecececeeeteeeeeeecesaeeeeseeeneseneeaneernans
2.7 Format of Text for TYPiIng ooo... cee ccsccecssecsseeeesteeseeeeeseeecssesessaeeseeenesseseeas
2.8 Sequences of Finger Movements that Yield Fast Typing Rates ....
2.9 Which Hand to Use on a One-Hand Keyboard... eeeecessesereeecees
2.10 Literature SUMMALY ........cccccesscccsssssseeeesesssceeeeeesssceeesesscseneetesesssseneass
3.2 APPArAatUS oo. ceecesccseccesscesececnseseecseesssescseeeeecesseeesaeesaessatessteeseesaeeees 3.2.1 Dimensions of the One-Hand TCK uuu... ccc cccseenecssesseseeenees 3.2.2 CHOP COIN ......eccecceccceseesseesseeseecnesseeeseceseeesessecseessaneanecseeses
3.3 PLOCECUIES o.oo... eecestcecceesssccesseecssecesseneesacecsancessesessecesseecesesessanesuaeeneneeteagens 3.3.1 Experimental De@SIQN ...........ccccsssccsssssseecsssseceecesssnseneessesseesenes 3.3.2 Schedule Of Practice 0... ee eeecssscsseeseeseeesetsceesnsssreeeecsseeeeeeas
3.3.2.1 Breakdown Of SESSIONS ou... eesessesecsseeseeeesenens 3.3.2 2 Training Program .........ccssesceessteessseeeesseetsneesseeeenees 3.3.2.3 Makeup Of Trials oo... cece eeseeeseeeeeneceeseeeteeeeenes
4.9 Analysis of Average Performance of All Subjects ....0000...00 ee. 111 4.9.1 Prediction of All Subjects’ Performance oo... eee 111 4.9.2 Prediction of Average Performance... cecceeeeceeeees 114
5.2 Using Trials to Predict Performance 0.0... ccccccccccceeseeeetssseeeeeeees 117 5.2.1 Coefficients of the Log-Log Equation ............ cece 118 5.2.2 Prediction of Peak Keying Speed oo... cece ccc cece sees 119 5.2.3 Combined and Average Performance of All the
SUDJOCHS. 200. ee cece ecececesscceeecetseeesceceeessueseesenerseuereeeestnnaees 124
5.3 Predicting Performance from TIME o....... ccc ccc ccccceceeessceecessteentseeeesees 124
F Average Response Times for Each Chord by Trial .........eeceeeeeeee: 172
viii
List of Tables
Table Number Page
Table 1. Summary of Important Variables in Keyboarding Studies. .......0..... 48
Table 2. Average Accuracy Across Trials for Each Subject... 64
Table 3. Number of Trials Completed by Each Subject... eee 65
Table 4. Summary of R-Squared Values for All Functions Considered When Trial is the Regressor Variable. 0... ccc eeceeeeeeeeeeeereen 68
Table 5. The Fitted Log-Log Equation for Each Subject. oo... ee 78
Table 6. Statistical Comparison of Coefficients Bo and B1. 0.0... 80
Table 7. Average Length (and Standard Deviation) of 30, 40, 50, and 60 Minute Time Increments Used in the Regressions. ..............0..ccee 82
Table 8. Summary of R-Squared Values for All Functions Considered When Time Increment is the Regressor Variable. 30, 40, 50, and 60 Minute Increments and Sessions are Used. oo... 84
Table 9. Ratings Ranging From 0 to 9 Given by the Subjects Each Week Assessing Their Progress and Corresponding Performance. ......... 87
Table 10. Fastest Speed Attained on Randomized and Repetitive Trials. ..... 88
Table 11. Mean Squared Error for Log-Log Functions Fit Using Only Some of the Initial Performance Data Points. ........... eee 100
Table 12. Mean Squared Error for Four Equations Fit Using Only Some of the Initial Performance Data Points. For Subject 2 Only. ....0...... 109
Table 13. Chord Response Times (in ms) Averaged Over Trials 468 to 487, the Last 20 Trials Performed by All Subjects. «0.0.0... 110
Table 14. Summary of R-Squared Values for All Functions Considered for Average of the Subjects and All Subjects’ Performance. .......... 112
Table 15. Values of R@ and Peak Performance when Fitting the Equation
CPNj = Bo + B;/T; + Bo/T2 Without the First 100 Performance POINS. occ ccccccccccececcesseeeee sees e na nsseeeeetesecetsaeeeeeseeeesessieeeeseseeenesteiieeeeeas 123
List of Figures
Figure Number Page
Figure 1. View of Binary and Ternary KeyS.. ou... eseesceceesseeeceseeseeeeetneeeeeseeeneees 2
Figure 2. Learning Curve fora QWERTY Keyboard Found by Chapman (1919). eee eeecsesescseeeeecseesneeeeeeaecnaecsaesnseeseeseceseceeeseaecuseneesaseneeseeseseeanees 9
Figure 3. Learning Curve fora QWERTY Keyboard Found by Thurstone (1919) Predicted by Number of Pages Typed. ........cseeseneeeeeee 11
Figure 4. Learning Curve fora QWERTY Keyboard Found by Thurstone (1919) Predicted by Number of Weeks of Practice. oe 12
Figure 5. Median Speed for Four Semesters of Typewriting on a QWERTY Keyboard. Adapted from Green (1940). wu. 15
Figure 6. Typing Speed on a 10-key Chord Keyboard as a Function of Practice. Taken from Klemmer (1958). ou... ccsccssssecsssseesseeereesseeseens 18
Figure 7. Comparison of Typing Speeds for Chord and QWERTY Keyboards. Taken from Conrad and Longman (1965)... 23
Figure 8. View of a Chord Keyboard Developed by Rochester et al. (1978). Taken from Rochester et al. (1978). oo. eeeeseeeseeeseeneee 27
Figure 9. Typing Speed for Three Keyboards as a Function of Practice. Taken from Gopher and Raj (1988). oo. ceeeeeceteeeeeeeseeeeeeees 31
Figure 11. Ternary Chord Keyboard Model #2. oo. ee eeseseeceseeeesseeeeseaeees 36
Figure 12. Learning Curve for the Ternary Chord Keyboard #2. owe 38
Figure 13. Ternary Chord Keyboard Model #3. oo. eccceesesessesseneesereeennenes 39
Figure 14. Learning Curve for the Ternary Chord Keyboard #3. owe 40
Figure 15. Learning Curve for the One-Hand Ternary Chord Keyboard. ....... 43
Figure 16. Effect of Training Schedule on Typing Performance. Taken from Baddeley and Longman (1978). ...........cccssssscceeeesseesesnsnsesereeerees 45
Figure 17. Dimensions of the One-Hand Ternary Chord Keyboard. ............... 53
Figure Number
Figure 18.
Figure 19.
Figure 20.
Figure 21.
Figure 22.
Figure 23.
Figure 24.
Figure 25.
Figure 26.
Figure 27.
Figure 28.
Figure 29.
Figure 30.
Figure 31.
Figure 32.
Figure 33.
Figure 34.
Page
Chord Assignments for the Numeric One-Hand Ternary Chord KOYDOAIC. oo... ee ecesecsrseseeecnscereneeesenecsaaesenceesneetsvsessasesneeeesenecsaesenanersaeeesess 55
Scale Used by Subjects to Rate Their Performance Each WEEK. once. cccsccccsssecesscecssscecssscecescaeceesaeseesensecseacecsssaeeeeseeseesesseceesesaeessseesenees 61
Subject 2, Prediction of CPMi = Bo + B1*Trial and Actual PErfOrMaNnce. ou... ceecesscsssccesecescceesecssecsceesaeesessseeessensrseeaeecssssasecsasenseenaes 69
Subject 2, Prediction of CPMi = Bo + B1/Trial + B2/Trial*2 and Actual Performance. .0......cccccesssecsssssecceeceecceneessessssseeaeeeeaeseceeeeseeeceseesseees 70
Subject 2, Prediction of CPMi = Bo + B1i*Trial + B2/Trial + B3/Trial*2 and Actual Performance. oe. eecseessecesecseeeeceessesseees 71
Subject 2, Prediction of CPMi = Bo + B1*Ln(Trial) and Actual POrfOrMAN Ce. ce cecssccesssccsssseceesseeecesseecesseeeseeaeeeceneeeceeatecsusueesenueesseaeeeenees 72
Log-Log Prediction and Actual Performance, Subject 1. .............. 73
Log-Log Prediction and Actual Performance, Subject 2. .............. 74
Log-Log Prediction and Actual Performance, Subject 3. .............. 75
Log-Log Prediction and Actual Performance, Subject 4. ............... 76
Log-Log Prediction and Actual Performance, Subject 5. .......0....... 77
Log-Log Prediction for Each Subject. 0... eecssscecsssssecesessseeeceeseees 79
Log-Log Regressions Using Only the First 25, 50, 100, 150, and 200 Points, SUDJOCt 1. ou... eee esececesseceeseeseceeeseneeneecesseeeeeeseseeeees 89
MSE for Log-Log Regressions Using Only a Few Initial Points, SUDJOCE 1. oo. eeecessesssccvsseecseeceseesssecesceecssceecsneesseeecssesecceeseeeeessstecserseesees 91
Log-Log Regressions Using Only the First 25, 50, 100, 150, and 200 Points, SuDjeCt 2. oo... eee eceecceeeeeceeeeeeeeteeteeeeaeeneeesnaeaaaeeneess 92
MSE for Log-Log Regressions Using Only a Few Initial Points, SUDJOCH 2. cee cccccecscssscssescesscesecseecsaecseeenseseseeeesseceasensecseesseseesseesensseetseeeees 93
Log-Log Regressions Using Only the First 25, 50, 100, 150, ANd 200 POINTS, SUDJECt 3... eee scccsseeeeeseceessneeseseetenseeseseseeeeetsaneessnes 94
xi
Figure Number
Figure 35.
Figure 36.
Figure 37.
Figure 38.
Figure 39.
Figure 40.
Figure 41.
Figure 42.
Figure 43.
Figure 44.
Figure 45.
Figure 46.
Figure 47.
Figure 48.
Figure 49.
Page
MSE for Log-Log Regressions Using Only a Few Initial Points, SUDJSCE Oo icc cccccecesececenssececsseceseseeeesseeeecsateserseseseeseseseaesssieaeeeeses 95
Log-Log Regressions Using Only the First 25, 50, 100, 150, and 200 Points, Subject 4. occ rere teneteteeettnereetteeeees 96
MSE for Log-Log Regressions Using Only a Few Initial Points, SUDECE 4. cece ccc cecseccssccessecsesseecsseeecsesenseeeseseeecsseeeesesrtseeetseseneeees Q/
Log-Log Regressions Using Only the First 25, 50, 100, 150, and 200 Points, SUDjECt DO. ieee ce ceeceeeeeeeeeetsseettttnteeeseaeees 98
MSE for Log-Log Regressions Using Only a Few Initial Points, SUDJSCH ccc ccc ccccsecccnteceeecsaeceeaeseeseseesessseeeesegessssesesseeeesiiseeeecsas 99
Mean Squared Error as a Function of Number of Points Used in Regression for All SUBJECTS. ooo ce eect cee ce ects cece eeeereeeeseeneey 101
Subject 2, Bo+B1/Ti+B2/Ti*2 Regressions Using Only the First 25, 50, 100, 150, and 200 PointS. 2.0... cece eeeeeeeeeeetttneeaes 103
Subject 2, MSE for Bo+B1/Ti+B2/Ti*2 Regressions Using Only @ Few Initial PONS. ooo eeccccccsssseccessecesaseeesseeeeesssseesesseecesseseeees 104
Subject 2, Bo+B1*Ti+B2/Ti+B3/Ti42 Regressions Using Only the First 25, 50, 100, 150, and 200 Points. oer creee 105
Subject 2, MSE for Bo+B1*Ti+B2/Ti+B3/Ti*2 Regressions Using Only a Few Initial POInts. oo. ccecsecccceeeeeesseeeeeseseeeaees 106
Subject 2, Bo+B1*Ln(Ti) Regressions Using Only the First 25, 50, 100, 150, and 200 POINtS. ooo. cc ccccccccsteeeeesestseeeesentsteseeeeeas 107
Subject 2, MSE for Bo+B1*Ln(Ti) Regressions Using Only a Few Initial POINS. oo... ccc cc cceeceeeeeeeeceeceeeeceseeseeeeeeestttsssssseeeaeaees 108
Log-Log Prediction and Actual Performance, All Subjects’ Data. ioc ccc eccceececsceccssesenesececcssecesaseseessseseesesecsseseesseeensiveserteasereranens 113
Log-Log Prediction, Actual Average Performance, and + and - 1 Standard Deviation of the Average of All Subjects. 0... 115
Estimate of Peak Speed Using the Eqn CPM = B+B1/Ti+B2/TiA2, Without the First 100 Trials, Sub 3. ow... 121
xii
1. INTRODUCTION
1.1 The One-Hand Ternary Chord Keyboard.
The keyboard is currently the primary device which people use for data
input to, and interaction with, a computer. The accepted keyboard standard is
the QWERTY keyboard (named for the first six keys on the left side of the
second row). Chord keyboards have been proposed as an alternative device to
the QWERTY keyboard. A chord is defined as the simultaneous activation of
two or more keys to produce one input. Although many differences exist
between various chord and sequential keyboards, almost all use binary keys.
The binary key has only two states: on and off. An example of a binary key is
the type of key used on the standard QWERTY keyboard.
A new type of chord keyboard has been design by VATELL Corporation,
Christiansburg, Virginia (U.S. Patent 4,775,255 of October, 1988). This
keyboard, called the Ternary Chord Keyboard (TCK), uses keys which have
three states: two on and one off. Figure 1 shows a comparison of the binary and
ternary keys. The TCK uses the three-position ternary keys, whereas all
previous keyboards have used the two-position binary keys. The TCK,
discussed in detail in section 2.5, is intended to be used by two hands (eight
keys one for each finger) for data entry of the alphabet, numbers, punctuation,
symbols, and cursor keys. Each of 64 "simple" chords is generated by
activating one key with the left hand and one key with the right hand. Several
authors have analyzed the TCK for performance compared to the QWERTY
keyboard, learning times, and performance capabilities (Fathallah, 1988;
Thus, a given slope could be associated with top speed, the trial number
determined when this slope occurs, and finally that trial number could be used
to calculate maximal performance. For example, it may be assumed that when
the slope decreases to 0.025 CPM/Trial, a person has effectively reached peak
performance. For subject 3, a slope of 0.025 CPM/Trial occurs at trial 1684
(solving for T; when 0.025 = 9.6 «7, 9-891) at trial 1684, the Log-Log equation
119
predicts a speed of 212 characters per minute (CPMy¢94 = 48.231*16849-199 —
212). |
The third technique to predict top speed involves using the function CPM;
= Bo + B,/T; + Bo/T|2 as suggested by Bittner (1991). He assumed that the
equation Y; = Bo + By/T; + B2/T;2 can predict peak performance, where Y; is the i-
th trial performance, Tj is the i-th trial, and Bo, By, and Bo are fitted coefficients.
As the number of trials increases the magnitude of the last two terms decreases
very rapidly ( B,/T; and B2/T;2) and Bg estimates peak performance; it is the only
function of all functions considered that will reach an asymptote. However, this
formula does not predict the performance progress of the subjects very well (all
R2 are less than 0.5 as shown in Table 4); thus, Bp can not be considered an
accurate prediction of the fastest keying speed.
The function appears not to predict final performance well because of the
steepness of the initial learning curve; thus, it underestimates peak
performance, see Figure 21 for subject 2. If several of the early points are
deleted from the regression, this function might predict a more accurate
asymptote. As an example, the formula CPM; = Bo + B;/T; + Bo/T;2 was fit to the
data of subject 3 ignoring the first 100 data points. It appears that most of the
steep portion of the learning curve is eliminated by ignoring the first 100 points,
see Figure 49. The fit of the remaining 536 points yields the equation CPM; =
195.7 - 16,199(1/T;) + 919,625(1/T;2) which has an R2 of 0.863 and predicts a
peak performance of 195.7 characters per minute.
The fourth technique used to predict peak performance employs only the
final trials. These trials were composed of a string of six characters repeated 84
times (see Section 3.3.3). It was felt that if keying the string became a highly
120
'E ANS ‘SIEUL
OOL ISUl4
SUL INOYWM
‘ZvIL/ZAtLL/ +a
= Wd
uby ay}
Guisq peeds
yead jo
ayewisy “Gy
aunb4 JOQUINN
jeu
069 009
OSS 00S
OSr O0r
ose o00€
ose 002
Ost ool
og 0
| I
} |
{ |
| |
| |
| |
| I
I {
| {
t 1
1 I
] I
| i
0
+ OV
I 09
+ O01
+ Oct
ainulyy Jad suajoeseuy
ry +
OVI Ph
TY
Nf yr
oF Wi
+
O91
+ O8t
— 00¢
ovil/ed+ W/lG@+@
--------- enjoy
121
automated motor program (with minimal response times), the performance on
these trials would represent peak keying speed. However, two of the subjects
typed faster in the random trials than in the repeated character string trial, as
shown previously in Table 10. In addition, the means of the fastest speeds
attained on randomized and repetitive trials were not significantly different
which indicates that the repetitive trials, in fact, were not a good indication of
peak keying performance.
Of the four techniques, using the equation CPM; = Bo + By/T; + Bo/Ti2
(without initial trials) might provide the best method for predicting peak
performance. This equation provides a much better predictor of actual
performance if initial performance points are ignored; the R2 values and
predicted peak performance for each subject are shown in Table 15. This
equation, without the 100 initial trials, predicts peak (asymptotic) performance
which was not attained by any of the subjects on the randomized trials
(comparing Table 15 with previously shown Table 10). The major difficulty with
this approach is to determine how many of the original points should be left out
of the regression.
Assuming a large amount of practice time, assuming a minimal slope, or
using the repeated string trials do not seem to be suitable methods to use to
predict peak performance. By what criteria could the number of trials be chosen
and considered the upper limit of practice; should it be 2000, 5000, 10000, or
more trials? The same dilemma arises when trying to assume a minimal slope
as the stopping point of improvement.
122
Table 15. Values of R2 and Peak Performance when Fitting the Equation CPM;
= Bo + By/T; + Bo/Ti2 Without the First 100 Performance Points.
Subject R2 Predicted Peak (in CPM)
1 0.940 196.8
2 0.854 197.7
3 0.863 195.7
4 0.870 158.8
5 0.759 161.1
where CPM j is the characters per minute typed on the i-th trial
(dependent variable) and Tjis the i-th trial of practice
123
5.2.3 Combin nd Aver Performan f All th
The Log-Log equation fit the combined and average performance of the
subjects best. It was expected to fit best because it was the best at predicting
each individual’s performance. The equation fit to the combined subjects’ data
probably had a high R@ (0.843) because there were not more than five subjects’
data. Kroemer (1990) and McMulkin and Kroemer (1991) did not find an
equation which accurately fit combined subjects’ data. This could be due either
to the larger number of subjects (10 to 12) or the shorter training periods (3 to
12 hours).
5.3 Predicting Performance from Time
The average characters per minute entered by the subjects were
calculated for 30, 40, 50, and 60 minute time periods and for sessions.
Because the time periods were composed of entire trials, the average lengths of
the 30, 40, 50, and 60 minute time periods were closer to 32, 42, 52, and 62
minutes respectively, shown previously in Table 7. It is not so important that the
length of the time periods are factors of 10, but that the average length of the
time increments is close to equal for all subjects. Inspection of Table 7 reveals
that for all four time periods each subject had approximately the same average
amount of practice.
After establishing the performance of the subjects over the time intervals,
the same five equations used to predict performance from trials were fit to the
time period data. The equations CPM; = Bo + ByP; + Bo/P; + B3/Pi2 and CPM; =
eBo p,B1 (a Log-Log equation) provided the best fits. P; is equal to 1, 2, 3,.... not
124
30 minutes, 60 minutes, and so on. The general form of the Log-Log equation
(CPM; = eBo P;B1) provided good fits for both repetition and time data.
It does not appear that changing the length of time increments has much
effect on the ability to predict performance. Five different time increments were
used as regressor variables: 30 minutes, 40 minutes, 50 minutes, 60 minutes,
and total sessions (1.5 hours, not all this time was spent practicing). For the
Log-Log function, the largest difference in R2 values was only 0.008 (Table 8).
5.4 Subjective Ratings.
The subjective ratings could not be used in any analysis to predict
performance of the subjects. The subjects probably found it difficult to rate their
performance early in the experiment because they did not know what their final
performance would be. At the beginning of the experiment, subjects might have
simply chosen a middle rating and then gradually increased it until the end of
the experiment. For this reason, the ratings that the subjects provided are not of
an interval nature, but are instead at best an ordinal measure. The data of
subjects four and five illustrate some of the problems with the subjective rating
procedure, shown previously in Table 9. Subject four rated performance at a
nine by day 30 but was clearly still improving over the following two weeks.
Subject five rated performance at a four on day 5, then a five on day 10, and
then back to a four on day 15. This subject probably felt the ratings of the initial
two weeks were too high and then readjusted the scale for week three and
started the rating at four again. Overall, little information could be extracted from
the results of the subjective ratings.
125
5.5 Evaluation of Criteria to Predict Performance.
Part of the second goal of the experiment was to determine which
criterion of the following three criteria could be used to predict keying speed:
number of trials/repetitions, hours (or other time increments) of practice, and
subjective opinion. The Log-Log equation, CPMj = eBo T;B1, provided the most
accurate prediction of performance when considering two regressor variables,
time and repetition. The average R2 values using the Log-Log function for
these two criteria were as follows:
Trials (504 repetitions per trial) R2 = 0.956
Time of Practice (average for all increments) R2 = 0.982
This gives further credence to the assertion that the Log-Log function describes
the underlying progress of performance. It does not appear to matter much
whether repetition or time of practice is used to predict performance. Although
the equation CPM; = Bo + ByPj + Bo/P; + B3/P2 is effective when time increments
constitute the regressor variable, it did not predict performance as well as the
Log-Log equation when trials constitute the predictor variable. This indicates
that the equation to describe performance remains the same when the predictor
variable changes from trials to time. These findings contrast those of Thurstone
(1919); he asserted that the function changed when time instead of repetition
was used to describe performance progress. The Log-Log function continued
to show flexibility with different levels of the time predictor. The fits of the Log-
Log equation had fairly constant R@ values across the five time increments
considered.
126
Subjective opinion data, as elicited in this study, could not be used to
predict performance due to the ordinal nature of the ratings. See Section 5.4 for
further discussion.
5.6 The Concept of Peak Performance.
It is apparent from Figures 24 to 28 that the actual performance of the
subjects did not approach an asymptote which could be considered the peak
performance. The Log-Log equation, which provided a very good description of
the actual performance does not predict an asymptote; it presumes continual
improvement in performance which, however, becomes increasingly slow and
gradual.
Both the actual performance observed and the prediction of the Log-Log
equation seem to contradict the traditional presumption of learning - a slow
increase in performance until gradually a upper limit is reached that can not be
exceeded. It might be that there is no limit to performance improvement.
The response times of the keyboarding task of this study can be broken
down into their components. First, the stimulus must be sensed by the subject,
for example the number 8 is to be typed. Second, the sensed information must
be transmitted through the (afferent) peripheral nervous system to the brain.
Third, a decision must be made in the brain as to what response to make to the
incoming stimulus. Fourth, the response signal must be transmitted along the
(efferent) peripheral nervous system to muscles. Fifth, the ensuing movement of
the fingers determines the "response time" and a character is typed. The third
and fifth steps of the response are probably becoming faster with practice with
most of the improvement likely to occur in the third step, the decision making
127
process. There may be a limitation to how quickly response decisions can
become. However, that minimal decision time may never actually be reached.
5.7 Prediction of an Entire Curve from the Initial Data Points.
From a practical standpoint, to predict the performance of subjects, it is
costly to gather over 500 data points. It is, therefore, of great interest to find out
how many initial performance points are needed before an accurate learning
curve can be predicted.
For subject 2, four different equations were fit to the actual data when
only 25, 50, 100, 150, and 200 of the initial performance points were used in the
regression. The mean squared error between the actual and predicted
performance for all 20 cases (4 equations and 5 quantities of initial number of
points used in the regression) was shown previously in Table 12. By perusing
the information in Table 12, it becomes clear that the only function capable of
accurately predicting the entire learning curve from the initial data points is a
Log-Log function.
The remaining functions do not closely approximate learning or final
performance. The Log-Log equation has a mean squared error which is at least
an order of magnitude smaller that the other three functions when only 25, 50,
100, 150, or 200 initial data points are used in the regressions. Inspection of
previously shown Figures 32, 41, 43 and 45 clearly indicates that only the Log-
Log function predicts final performance effectively from regressions including
only a few initial performance points. The other functions either predict too high
a final performance (CPMj = Bo + By,Tj + Bo/T; + B3/Ti2) or too low a final
performance (CPM, = Bo + By/T; + Bo/Tj2 and CPM; = Bo + ByLn[TjJ).
128
It is difficult from a small sample size of five subjects to draw a definite
conclusion about how many points are sufficient to accurately predict the entire
performance curve using the Log-Log equation. The regression using all points
is assumed to have the lowest mean squared error. Thus, the MSE of
regressions using less points should be compared to the MSE of the all points
regressions to determine its relative accuracy. For four of the subjects (2
through 5), the MSE is drastically reduced when the number of points used
reaches 50, see Figure 40 and Table 11 shown previously. The MSE resulting
from including more than 50 points is a fairly flat function for these four subjects.
In contrast, subject 1 exhibited a much more modest decrease in MSE as the
number of points included in the regression increased.
Overall, the Log-Log function proved to be the only equation able to
predict the entire learning curve from a limited number of initial data points.
Due to the small number of subjects, it is difficult to make assertions about the
number of points needed with the Log-Log equation to capture the entire
performance progress with a reasonably small mean squared error: as few as
50 initial points were needed for four of the five subjects (out of about 550 total
points). However, when applying these findings, the amount of uncertainty that
is acceptable needs to be determined which then points to the number of initial
performance points needed.
5.8 Chord Analyses.
The average response times for each chord were calculated for the last
20 trials including all subjects, i.e., trials 468 to 487. Table 13, shown
previously, lists the chords’ average response times and relative ranking. The
129
major result of the chord analysis is that after over 400 trials of practice there are
still chords that are significantly faster than others. However, significant
differences found could be due to the large power of the tests. Each chord
average response time consisted of 2800 observations (28 chords per trial X 20
trials X 5 subjects).
The chord analysis was based on practice due to repetition not amount of
time because response times from trials 468 to 487 were used. For subject 1,
these trials were the last ones completed, while for subjects 2 to 5 they were
completed before the end of the experiment. The number of final trials to
include in the analysis was limited to 20 for two reasons. First, chord
performance was of interest after a large amount of practice. It was of interest to
find if differential chord performance existed after a large amount of practice.
Second, the last 20 trials include enough observations to obtain a
representative average and standard deviation. It is possible that fewer number
of final trials could be used for the analysis.
5.9 Further Research.
The results of this study point to several areas for future research and
analysis. First, the trials were held constant at 504 characters. It would be
interesting to investigate the effect of different trial lengths on the ability of the
Log-Log function to predict performance. In the same way that variations of time
increments were used with the Log-Log equation, number of repetitions per trial
could be varied and accuracy of the Log-Log function analyzed.
Second, the performance points seem to have some autocorrelation.
Performances of the subjects “bounced” about the Log-Log curve. Apparently,
130
series of points alternated between being higher and lower than the previous
point. It might be possible to include an autocorrelation factor in the model of
performance and then account for more of the variation.
Third, as mentioned in the literature review, there appears to be a
relationship between number of characters that are learned and the steepness
of performance progress. Research is needed on how the values of the
coefficients in the Log-Log equation change with character set size. For
example, faster keying speeds might have been obtained more quickly if this
study had used 10 characters instead of 18.
Fourth, the subjective opinions were that were obtained in this study
could not be correlated with performance. Perhaps other techniques of rating
could be used to find a relationship between performance and subjective
opinion about that performance.
131
6. REFERENCES
Alden, D.G., Daniels, R.W., and Kanarick, A.F. (1972). Keyboard design and operation: a review of the major issues. Human Factors, 14(4), 275-293.
Baddeley, A.D, and Longman, D.J.A. (1978). The influence of length and frequency of training session on the rate of learning to type. Ergonomics, 21(8), 627-635.
Bartram, D., and Feggou, O. (1985). An evaluation of four different keyboard designs for foreign destination coding desks. Report ERG/Z6545/85/4c. Hull: University of Hull, Ergonomics Research Group.
Bequaert, F.C., and Rochester, N. (1977). Teaching typing on a chord keyboard. IBM Report TR 00.2918. Cambridge, MA: SPD - Cambridge Systems Group.
Bittner, A.C. (1991). Analysis of learning-curve asymptotes: ergonomic applications. In W. Karwowski and J.W. Yates (eds.), Advances in industrial ergonomics and safety Ill. London, England: Taylor and Francis Ltd.
Bowen, H.M., and Guinness, G.V. (1965). Preliminary experiments on keyboard design for semiautomatic mail sorting. Journal of Applied Psychology, 49(3), 194-198.
Callaghan, T.F. (1989). The utility of a technique for testing the difference in ease of chords on the ternary chord keyboard. Unpublished masters thesis, Virginia Polytechnic Institute and State University, Blacksburg, VA.
Chapman, J.C. (1919). The learning curve in typewriting. Journal of Applied Psychology, 3, 252-268.
Conrad R., and Longman, D.J.A. (1965). Standard typewriter versus chord keyboard - an experimental comparison. Ergonomics, 8, 77-88.
Cornog, J.R., Hockman, J.F., and Craig, J.C. (1963). Address encoding - a study of the double-binary keyboard as a link in the machine-sorting of mail (ASME Report Number 63-WA-338). New York, NY: American Society of Mechanical Engineers.
Dvorak, A. (1943). There is a better typewriter keyboard. National Business Education Quarterly, 12(2), 51-66.
Dvorak, A., Merrick, N.L., Dealey, W.L., and Ford, G.C. (1936). Typewriting behavior. New York, NY: American Book Company.
132
Evey, R.J. (1980). How typists type. IBM Report Number HFC-35. San Jose, CA: IBM General Products Division.
Fathallah, F.A. (1988). An experimental comparison of a ternary chord keyboard with the QWERTY keyboard. Unpublished masters thesis, Virginia
Polytechnic Institute and State University, Blacksburg, VA.
Gentner, D.R. (1983). The acquisition of typewriting skill. Acta Psychologica, 54, 233-248.
Glencross, D., and Bluhm, N. (1986). Intensive computer keyboard training programmes. Applied Ergonomics, 17(3), 191-194.
Goldstein, |.L. (1986). Training in organizations: needs assessment, development, and evaluation. Pacific Grove, CA: Brooks/Cole.
Gopher, D. (1986). Experiments with a two hand chord keyboard - the structure and acquisition process of a complex transcription skill. Report Number HEIS-86-5. Arlington, VA: Office of Naval Research, Personne! and Training Research Programs.
Gopher, D. and Eilam, Z. (1979). Development of the letter-shape keyboard: a new approach to the design of data entry devices. /n Proceedings of the Human Factors Society 23rd Annual Meeting (pp.40-44). Santa Monica, CA: Human Factors Society.
Gopher, D., Hilsernath, H., and Raij, D. (1985). Steps in the development of a new data entry device based upon two hand chord keyboard. /n Proceedings of the Human Factors Society 29th Annual Meeting (pp.132-136). Santa Monica, CA: Human Factors Society.
Gopher, D. and Raij, D. (1988). Typing with a two-hand chord keyboard: will the QWERTY become obsolete. /EEE Transactions on Systems, man, and cybernetics, 18(4), 601-609.
Green, L. (1940). A study of typewriting achievement in three high schools. Journal of Educational Research, 34(3), 209-217.
Lee, S.S., and Kroemer, K.H.E. (1991). Report on subjectively preferred one- hand chords on TCK #2. Blacksburg, VA: Industrial Ergonomics Laboratory, Virginia Polytechnic Institute and State University.
Leonard, J.A., and Newman, R.C. (1965). On the acquisition and maintenance of high speed and high accuracy in a keyboard task. Ergonomics, 8, 281- 304.
133
Lockhead, G.R., and Klemmer, E.T. (1959). An evaluation of an 8-key word- writing typewriter. IBM Report Number RC-180. Yorktown Heights, NY: IBM Research Center.
Killeen, P.R. (1978). Stability Criteria. Journal of the Experimental Analysis of Behavior, 19, 17-25.
Kinkead, R. (1975). Typing speed, keying rated, and optimal keyboard layouts. In Proceedings of the Human Factors Society 19th Annual Meeting (pp.159-161). Santa Monica, CA: Human Factors Society.
Kirschenbaum, A., Friedman, Z., and Melnik, A. (1986). Performance of disabled persons on a chordic keyboard. Human Factors, 28(2), 187-194.
Klemmer, E.T. (1958). A ten-key typewriter. IBM Report Number RC-65. Yorktown Heights, NY: IBM Research Center.
Kroemer, K.H.E. (1990). Final report on the experiments performed on VATELL Corporation's ternary chord keyboard research project 43716 (230-1 1- 110F-106-801189-1). Blacksburg, VA: Industrial Ergonomics Laboratory, Virginia Polytechnic Institute and State University.
McMulkin, M.L., and Kroemer, K.H.E. (1991). Report on the experiments performed on one-hand TCK #2. Blacksburg, VA: Industrial Ergonomics Laboratory, Virginia Polytechnic Institute and State University.
Morgan, A.S., Drake, J.B., Heck, S.K., and Long, C.H. (1981). Experimental study of effects on speed and accuracy of teaching alternate-hand method on a ten-key electronic calculator. Perceptual and Motor Skills, 52, 695-700.
Mussin, E.H. (1980). The microwriter. In Excerpt from JERE Electronic Office Conference Proceeding, (129-134). London, UK: Institute of Electronic and Radio Engineers.
Neter, J., and Wasserman, W. (1974). Applied Linear Statistical Models. Homewood, IL: Irwin Inc.
Norman, D.A., and Fisher, D. (1982). Why alphabetic keyboards are not easy to use: keyboard layout doesn't much matter. Human Factors, 24(5), 509- 519.
Noyes, J. (1983a). Chord keyboards. Applied Ergonomics, 14(1), 55-59.
Noyes, J. (1983b). The QWERTY keyboard: a review. International Journal of Man-Machine Studies, 18, 265-281.
134
Ratz, H.C., and Ritchie, D.L. (1961). Operator performance on a chord keyboard. Journal of Applied Psychology, 45(5), 303-308.
Richardson, R.M.M., Telson, R. U., Koch, C.G., and Chrysler, $.T. (1987). Evaluation of conventional, serial, and chord keyboard options for mail encoding. In Proceedings of the Human Factors Society 31st Annual Meeting (pp.911-915). Santa Monica, CA: Human Factors Society.
Rochester, N., Bequaert, F.C., and Sharp, E.M. (1978). The chord keyboard. Computer, 11(12), 57-63.
Salthouse, T.A. (1984). The skill of typing. Scientific American, 250(2), 128-135.
Salthouse, T.A. (1986). Effects of practice on a typing-like keying task. Acta Psychologica, 62, 189-198.
SAS Institute Inc. (1985). SAS user's guide: statistics. Cary, NC: SAS Institute Inc.
Seibel, R. (1962). Performance on a five-finger chord keyboard. Journal of Applied Psychology, 46(3), 165-169.
Seibel, R. (1964). Levels of practice, learning curves, and system values for human performance on complex tasks. Human Factors, 6, 293-298.
Seibel, R. (1972). Data entry devices and procedures. In Van Cott, H., and
Kinkade, R. (Eds.), Human engineering guide to equipment design (pp.311-345). Washington D.C.: American Institute for Research.
Showel, M. (1974). A comparison of alternative media for teaching beginning typists. The Journal of Educational Research, 67(6), 279-285.
Sidorsky, R.C. (1974). ALPHA-DOT: a new approach to direct computer entry of battlefield data (Technical Report Number 249). Arlington, VA: U.S. Army Research Institute for Behavioral and Social Sciences.
Strong, E.P. (1956) A comparative experiment in simplified keyboard retraining and standard keyboard supplementary training. Washington D.C.: General Services Administration.
Thurstone, L.L. (1919). The learning equation. Psychological Monographs, 26, 1-52.
Towill, D.R. (1976). Transfer functions and learning curves. Ergonomics, 19(5), 623-638.
135
Towill, D.R. (1982). How complex a learning curve model need we use? The Radio and Electronic Engineer, 52(7), 331-338.
The purpose of the study is to gather information about a newly designed keyboard, called the One-Hand Ternary Chord Keyboard (TCK). You are expected to participate for 40 one-and-a-half hour sessions. There will be one session a day, five days a week for eight weeks; sessions will be scheduled for the same time each day.
The first session will be the "learning phase" and the following 39 will be the "performance phase”.
Learning Phase. In the first session, you will learn general facts about the One-Hand TCK and how to type characters with it. You should proceed through this section at your own pace, follow directions on the computer screen carefully, and consult the experimenter when suggested by the computer screen and also if you have any questions or problems.
You will learn to type the 10 numbers and 8 mathematical symbols during this session. To help you learn these characters, quizzes will be given at times to test your knowledge. Please, keep in mind that you are expected to memorize how to type these characters, and to type them with not more than 3% errors.
Performance Phase. In each one-and-a-half hour sessions, you will be given text files to type which will increase your speed. Follow the instructions displayed on the computer screen. Please keep errors down at 3%, but - - go for speed!
Keep in mind that the purpose of this experiment is to determine the highest typing speed that you can attain on the One-Hand TCK. During the sessions, please work to increase your speed and give the fastest performance you can. It is expected that you will have a general trend to increase your keying speed over the 40 sessions.
Once a week you will be asked to rate your performance on a 0 to 9 scale. You will still finish the full 40 sessions regardless of your ratings, but it is important to the experimenter how you feel you are progressing.
The computer program will explain a great deal about the TCK, so wait until you have proceeded though the INTRODUCTION to the learning section before asking any detailed questions you might have. Consult the experimenter at any time if there are any questions or problems.
Initials
140
Appendix C
Informed Consent Form
141
CONSENT FORM
l, am participating in this research study because | want to participate. The decision to participate is completely voluntarily on my part. No one has coerced or intimidated me to participate.
The experimenters have adequately answered any and all questions | have asked about this study, my participation, and the procedures involved, which are described in the attached "EXPERIMENT INSTRUCTIONS," which | have initialed.
| recognize the research team as Mark McMulkin, Graduate Research Assistant (231-5359) and Dr. K.H.E. Kroemer, Principal Investigator (231-5677).
| understand that they will be available to answer any questions concerning procedures throughout this study. | understand that if significant new findings develop during the course of this research which may relate to my decision to continue participation, | will be informed. | further understand that | may withdraw this consent at any time and discontinue further participation in this study without prejudice to my entitlements. | also understand that the Principal Investigator, his assistants, or medical consultants for this study may terminate my participation in this study if he or she feels this to be in my best interest. | may be required to undergo certain further examinations, if they are necessary for my health or well being.
| do not have any disorders of my cardiovascular system, of my spinal column (particularly in the lower back), or any other disorders or deficiencies, which make it unadvisable for me to participate in this experiment.
| understand that in the case of physical injury, no medical treatment or compensation are offered under the research program, of by VA Tech-VPI.
| understand that | shall receive payment in the amount of $5 per hour plus a $50 bonus for showing a general trend of improvement in keying speed over all 40 sessions. However, | further understand that if | withdraw from the experiment before it is completed, | will be paid only for the time | actually spent performing in the experiment at $5 per hour. If for some reason the equipment used for the experiment malfunctions before the completion of the experiment, | understand that | will only be paid for the time | actually spent performing in the experiment at $5 per hour.
| understand that the results of my efforts will be recorded and that | may be photographed, filmed, or audio/videotaped. | consent to the use of this information for scientific or training purposes, and | understand that any records of my participation in this study may be disclosed only according to federal law, including the Federal Privacy Act, and its implementing regulations. This
142
means that personal information will not be released to an unauthorized third party without my permission. |
| understand that if | have any further questions about my rights as a participant, | may contact Dr. Ernest R. Stoudt, Chairman of the Institutional Review Board at VPI&SU, at 231-5281.
| FULLY UNDERSTAND THAT | AM MAKING A DECISION WHETHER OR NOT TO PARTICIPATE. MY SIGNATURE INDICATES THAT | HAVE DECIDED TO PARTICIPATE UNDER THE CONDITIONS DESCRIBED ABOVE.
Signature Date
Printed Name SS Number
Address Phone Number
Industrial Ergonomics Lab ISE Department VPI&SU Blacksburg, VA 24061
Protocol ERGLAB 1987 Date: Jan. 1987, renewed February 12, 1989 for two
143
Appendix D
Derivation of CPM; = C+T|B1
144
Derivation of the Log-Log function
Log(CPM)) = Bo + ByLn(T})
CPM, = eBo + BiLn(Ti)
CPM; = eBo eBiLn{T))
CPM; = eBo eb n(TiP1)
CPM; = eBo 7,81
Since eBo is constant let C = eBo
CPM; = C+T;1
145
Appendix E
Graphs of All Time Increments and Subjects for CPM; = Bp + ByP; + Bo/Pj +
Bs/P2 and CPM; = eBo P;B
146
LavieQll.dve
ovd/edtd/ldtd.d
lenpy
POLO, INU]
O€ 40}
uoIssasHay ‘|
yoalqns *1°9
eunBbi4
Poa
aINUIW OF
02 S9
09 SG
OS Sr
Ov Se
OF Sz
OZ SI
OL
|
| _ {
{ |__|
| |
| |
{ |
| I
t 1
1 I
| }
1 i
] !
0c
OV
09
08
OOL
Ocl
OVI
O91
O8l
002
elnuily Jed suajoeeyy
147
Lavibul,. ave
ovd/cdtd/ldtd.d
lenjpy
GS
POUedY a}NUI
Op 40}
UoIssesBayY ‘|
Joalqns °2z°9
aunbi4
OS
Poudd
aINUIW OF
Sb Ob
GSE O&
Ge O06
Gt _
Ol S
0
0¢
Ov
09
08
001
0c}
OVI
091
O8l
002
o]nulyy sed syajoeseyyg
148
POS
S}NuUIPy OS
40} uoISsaiHay
‘| yoolqns
‘Ey eunbi
POHSd aiNuUIl|
OS
GY Ov
GE O€
Ge 0¢
Sl Ol
S 0
| |
| |
| ]
l |—
| +
0 Ov
09
LavibQll,dyve ———-—~—
08
ovd/cdtd/l dtd.
--------- 001
jenjoy Ocl
SINUIW Jed syajoeeYyD
Or |
O91
+ O81
—— 002
149
bavielt .ave
ovd/cdtd/tdtd.d
Jenjov
[EAUalU] AINUI|
O9 40)
SuoisseuBay ‘|
yalqns ‘py
aunbi4
POHad aINUIW
O9
OV Ge
0€ Se
0¢ Sl
Ol S
| |
—__| |
L {
| |
I 1
| {
|
+
0c
OV
09
08
00}
0c |
e]nul Jed suajoeieyD
Ovl
O91
O81
00¢
150
jeAsaju| Aeq
10} uoissaiBHay
‘1 yoelaqns
*s°y eunbi4
Keq
OV Ge
0€ Ge
02 Gl
Ol G
0
| |
| |
| |
{ |
I t
I |
i |
| |
0
LavAed,dve ——-———
cevAeq/zqtAeq/LatAed,d ---------
+ O01
leny +
Ob
alnulj Jad syejoeieyy
7 OVt
+ O91
7 O8t
—- 00¢
151
bavibl1, ave
ovd/edtd/l dtd.
jenpy
OZ
POU,
SINUIW] OE
40} uoISsesBeyY
‘Zz yoalqns
-9°9 aunbi4
POuad SyNuUI
OF
S9 09
GS OS
Sh OF
GE OF
Ge 02
StL OF
GS O
| |
| {st
| {
| |
| |
{
| i
! ]
J |
I |
{ I
| 1
I
0c
Ov
09
08
OOt
Oct
OVI
O9t
O8l
00¢
e}Null\ 49d sugjOeBeYyy
152
bavieHl.dve
ovd/edtd/|atd.d
lenjoy
GS
POudY s}NUI-
OP 10}
uoIsseuBay ‘2
yalqns “7°
eunbi4
POUdd aINUIN
OF
0S Sb
Ov Sse
O0& Ge
O06 Gt
Ol S
| |
| |
jt |
| |
| {
I |
I |
| |
T |
1 1
0¢
OV
09
08
00}
Ocl
Ovi
O91
081
00¢
anu Jad syajoeseuy
153
POMS
a}NUIW OG
40} uoIssasHayY
‘Zz elgns
-g°y eunbi4
poued aINUIW!
OS
Sy CO Oh:Ssi‘<i«sSC“‘<«té‘i:SC“‘wSSC‘iSCC(‘éiR(SSCSG
0 |
| l
a |
| |
i 0 09
LdvieWil.dve8 —————
08
ovd/cdtd/ldtd.d ---------
OO}
jenjoy Oc}
Ori
O91
081
— 00¢
154
ejnull sed syajoeeuy
bavieut.dve
ovd/cdtd/ldtd.d
jenjpoy
[EAIBIU] S}NUII|
OQ 410}
SUOISSasHay ‘zg
yoalqns ‘6°9
eunbi4
POUSd 9}NUI
09
OV GE
O& Se
0c Sl
Ol G
0
0c
OV
09
08
00l
Ocl
OV lL
091
O8t
00¢
eINUIW sed SuBjORIeUD
155
LavAeqd, dave
ovAeq/za+Aeq/1g+Aed,
fenjoy
jeAraju Aeg
40} uoissaiBay
‘g yelqns
‘o1°y sunbi4
Keq
Ov GE
O¢ Ge
0d Gl
Ol G
| |
| |
| {_
{ |
| }
| i
I |
0¢
Ov
09
08
001
Ocl
e]nuIw 16d suajoeeyyD
Ovl
O91
O8l
00¢
156
LaviGlt.dva
ovd/edt+d/tdatd.d
Jenjoy
DOO
SINUI
OF 10}
UoIssalBay ‘¢E
yalqns “1
1°9 eunbi4
poua aynuiW
O€
0Z S9
09 SG
0G Sb
Ov SE
O€ GZ
O02 St
OL ¢
O |
| |
| |
{|
| |
| |
i |
| |
1 |
j |
1 |
1 I
t |
| 1
1
Od
OV
09
08
001
0c}
Ovi
091
O8l
00¢
alnui sed syajoeueyy
157
POWdd aINUIW
OF 40}
UoIssesHay ‘E
yoelqns *z1°Q
eunbi4
Poued aINUIW
OF
S¢ 0S
Sr Or
Se of
Sze of
St OL
¢ oO
|
| |
| |
I 1
I IF
0 09
havieQHl.dve —————
08
évd/cdtd/ldtd.d
--------- 0Ol
yenjoy Oct
OF |
091
O8l
— 00¢
158
e]nul~ 4ad syejoeseyD
POUSAY aINUIP]
OG 404
UOIssalHaY ‘€
yelqns ‘e1°4
aunbig
pouad ainui~w
Og
Sp Ohsti‘<iésESC“<«ié‘iOE:C“‘sSSSC‘i
SC C:t‘éisék
S 0
|
| |
| |
—
| |
) 0 09
bdvieHl,dye —-—-———
08
Cvd/cdtd/ldtd.d ---------
00}
jenjoy Och
Ort
O91
O8 |
— 00¢
159
enul, 18d suapeueyyD
baviebll.dve
ovd/cdtd/|dtd.d
renyoy
jeAlaju] aINUIWY
OO 40}
SuOIssasBayY ‘E
yoaelqns ‘p1"g
eunbi
poued eINuIW
09
Op Ge
O€ GZ
02 Sh
Ol S
0¢
Ov
09
08
001
Ocl
orl
091
O81
00¢
einul~— Jed suajoeieuy
160
jeAlaju| Aeg
10} uoissaiHay
‘¢ yoalqns
“¢4°y aunbi4
Keg
OV Ge
O€ Se
02 Gt
Ol G
0
| —t
| |
| |
| 0
| |
| 4
| i
] |
T Od
09
LavAed,a@ye ————~—
08
evAeq/edtAeq/Lg+Aed,g ---------
001
jenpy
OZL
a
091
08}
—- 00¢
161
ainul Jed syajoeeuy
Laviel1l.dye
ovd/cdtd/tatd.d
lenjoy
POHdd ajNuIWy
O€ 10}
UoISSauBayY ‘p
yoalqns ‘91°9
aunbi4
pouad eInuIW
OF
02 G9
09 SG
0G Sr
Or SE
Of Gz
Of Gt
OL
0
02
OV
09
08
001
Ocl
Ovl
091
O8l
00¢
Slnull\ Jed syajpeweyy
162
POW,
BINUIW OF
10} UoISsalBayY
‘p yoelqns
"74°94 aunbi4
pouad aInuIW
OF
cS 0S
Gr Orv
se oe
G2 of
St oO
¢
O |
| 4
| |
| |
| ]
| |
0
OV
09
LdvieHl,dve —————
08
Cvd/cdtd/ld+d.d ---------
00}
jenjoy Och
alnuljy Jad srajoeseuy
OVI
+ O91
+ O8t
—- 002
163
baviell. ave
ovd/cdtd/l dtd.
lenjpy
Sv OV
POS
SINUIW] OG
JO) UdISsaiBay
‘p joalqns
-g1°5 eunbi4
pouad anu]
OS
sé o€
G2 02
st ors
0
0c
Ov
09
08
001
Ocl
Or |
O91
081
00¢
SINuIW 16d syejoeieUy
164
havieul.dve
ovd/cdtd/ld*d.d
lenjoy
|EAJ}U] SINUIW
OQ 40}
SUOIssaJBayY ‘p
yoelans “61°9
aunbi
POUdd a}NUlW
O9
OV Se
O€ Go
0¢ Sl
OL S
| {.
| I
1 i
1 \
| |
|
Oc
Ov
09
08
001
Ocl
Ovl
091
08 |
00¢
anu Jad sr9joeieyyD
165
LavAeqd, ave
evAeq/zatheq/1g+Aeq,g
enjoy
OV
GE O€
Sd
Aeq
0c Gl ! ft
jealajuy Aeg
40} uoissaiBay
‘p yoalqns
‘ogy eunbi4
Oe
OV
09
08
00!
Ocl
OV l
O9l
O8l
00¢
alnulw sad suajoeieyyD
166
bavi. ave
ovd/cdtd/ld+d.d
fenjpy
POUdd aINuUIP
OE 10}
UOISsalHayY ‘s
oalqns *12°9
eunbi4
POLS, 9INUI|
OF
OL S9
09 GS
OS Sb
OF SE
OF G2
O02 Gt
OL G
O
| |
| I
{ |
| |
| |
| {
{ {
I t
I {
1 {
i |
I j
| |
I i
0c
OV
09
08
00!
Oc |
Ovl
091
08}
002
eynully Jed suajoeueyy
167
bavibl lave
ovd/cdtd/ldtd.d
enjoy
POS
SINUIW Ov
40} uoIssalBaY
‘sg yalqns
-zg"yQ eunbi4
OS Gv
OV
POWdd 9INUIW
OV
Ge SE
0€ 0c | __
St | {——
Ol
0¢
OV
09
08
OOL
Ocl
Ovi
O91
O8l
00¢
enully 18d susjoeeYyyD
168
baviGW1.dve
ovd/cdt+d/ldtd.d
lenjoy
POUAd SINUIW]
OS 40}
UdISsalBaYy ‘G
yoalqns ‘Eez’y
eunbi4
SY OV
+
GE
pousd einuiy
OS
06 Gz
oc St
or gs
0
| |
| |
1 |
\ I
J I
I
0c
OV
09
08
O0Ol
Ocl
Ort
O91
081
002
elnul~-, Jad syajoeseyy
169
[eAJa}U] SINUI,
OY 40};
SuoIssaiBayY ‘sg
yoalqns “pz°y
eunbi4
pousd elnuiyy
09
OP ce
o¢€ Sz
02 SI
Ol c
0
, |
| |
l l
L |
| 0
LavilGWl,dva
—_-——
—
—
Cvd/cdtd/ldtd.d
--------- 7
OO-
lenjoy +
OZt
+ OvVl
+ O91
+ O8t
— 00¢
170
ejnul sed susjoeeYyD
jeAraju Aeg
10} uoissaiBay
‘g yoalqns
°sz°y aunbi4
Keg
OV Ge
O€ Ge
0¢ GI
Ol S
0
LavAed,@ve —————
ovAeq/zat+Aeqg/lgt+Aed,g ---------
+ OOL
Gans
+ Oct
+ Ov
7 O9t
+ O81
— 002
171
einuljy Jed syajoeeyy
Appendix F
Average Response Times of Each Chord by Trial
172
EL puoyy
seul Aq
ow) asuodsay ebesaay
“1°44 ounbi4
JaquUNN jeu
00S OST
OO0r OSE
O00€ 0S¢
00¢ OSI
001 OS
0
[| |
| |
| |
| |
| )
t |
| |
| I
I |
I
+ O01
+ Q0¢
7 OO€&
+ O00
+ 00S
+ 009
+ O0OZL
+ 008
+ 006
- OOOL
- OOLL
- O0cl
+- OO€1
+ OOF1
+ OOSI
+ 0091
+ OOZI
7 0081
+ OO6L
+ 000¢
+ OOLe
— 00¢¢
173
(DeSswW ul) awl] esuodsey abeusay
00S OSL
OOP OSE
OO0€
Pl Puoyy
‘yeu Aq
euly asuodsay ebelaay
‘2°'4 eunbi4
JEQUINN jeuL
0S¢ |
00¢ | I
OSt { i
001 | I
0S | l
001
00¢
OO0€
OOF
00S
009
002
008
006
0001
OO-L
00¢1
OO€L
0O0r |
00S |
0091
OOZ 1
0081
0061
000¢
0012
00¢¢
(Oesw ul) aul, asuOodsey sbeusay
174
00S OSP
007 OSE
00€ | |
G| puoyy
:yeu Aq
eu, asuodsay
ebesany ‘¢'4
ounbi4
JEqUINN Jeu,
0S¢ | |
00¢ OS|
0Ol 0S
OOL
00¢
OO0€&
00v
00S
009
002
008
006
0001
OOLL
00c1
OOE |
00V |
00S1
0091
0021
0081
0061
000¢
00Le
00c¢
(O@sw ul) euly esuodsey obesoAy
175
00S OSt
00r OSE
O00€
Q| puoyy
seus AQ
auly asuodsay
ebeleny “p’4
aunbi4
JEquUNN jeuL
0S¢ | I
00¢ OSI | |
001 { I
OS {| I
OOl
00¢
OO&
00V
00S
009
002
008
006
0001
OOLt
001
00€ |
007+ |
00SI
0091
0021
008 |
0061
000¢
0Ole
00¢e¢
(Oesw ul) sully esuodsey abeisay
176
00S OST
0O0V OSE
00€
Z| puoyy
‘yeuy Aq
awl, asuodsay ebeaay
*s’4 aunbi4
JEQuINN jeu lL
0S¢ | i
00¢ | T
OS 1 | |
OOL 0S
001
00¢
00€
00V
00S
009
002
008
006
0001
OOL |
00c1
OO€ 1
007 |
OOS 1
009 |
OOZ1
0081
0061
0002
0Ol¢e
00¢¢
(Oesw ul) euly asuOdsay sbeusay
177
00S OSt
00v OSE
O0€ | i
Eg puoyy
‘jeu Aq
awl, asuodsay
ebessay -9°+4
aunbi4
JEqUINN jeu
0S¢ 00¢
0S |
001 0S | |
001
00¢
00€
007
00S
009
002
008
006
0001
0011
00¢ |
00€ |
00r |
00S 1
0091
OOZ1
OO08t
0061
000¢
001d
00c¢e
(osu ul) Swi, BsuOdseY eHbeuaay
178
00S OSP +
00v OSE
O0€
pz psouy
:jeuL Aq
euiy asuodsay
sbeyany °7°4
aunbiy
JOQUINN jeuL
0S¢ | |
00¢ OSI
O0Ol OS
OOl
00¢
O0O0€
007
00S
009
002
008
006
0001
001}
00¢ |
OOE |
0O0r |
00S |
0091
OO |
008 |
0061
0002
0012
00¢¢
(O@sw ul) swly asuodsay aBeusay
179
00S OST
OOP OGE
00€
Gz puoyuy
yeu
Aq awiy
asuodsay ebesaay
“g°+4 aunbi-4
JOQUNN jeu]
0S¢ | i
00¢ OS| | ]
001 | I
OS
001
00¢
00€
00V
00S
009
002
008
006
0001
0011
00c1
OOEL
OOVl
00SI
0091
OOZI
0081
0061
000¢
0Ole
00c¢
(O@sw ul) euuly eBsuOdsey sHbeieAy
180
00S OS?
OOP OSE
OO0€
{ |
QZ ps0yy
‘yeu Aq
owl asuodsay ebeuaay
*6°4 aunbi4
JOquny seul
0S¢ 00¢ | |
OSI | |
OOL { |
0S
001
00¢
O0€
00V
00S
009
002
008
006
0001
OOLL
00cl
OOEL
Or
00S |
0091
OOZ1
0081
0061
000¢
0OLe
00é¢
(Oasw ul) dw] ssuOdseYy ebessaay
181
00S OS?
00P OSE
00€ | |
8g PuoUy
:jeuy Aq
aul) asuodsay sbeisay
‘O|'+ aunbi4
JOquNN |e
0S¢ | I
00¢ |
OSI OO}
OS
OOL
00¢
O0O0€
00V
00S
009
002
008
006
0001
OOL |
00cl
OO0€ L
0Ovl
00S1
0091
0021
0081
0061
000¢
0O0Le2
00¢c2
(OesSw ul) eluly eBSUOdseY BbeisAy
182
GE puoyy
:jeuL Aq
oul asuodsey sHesany
*11°4 aunbis
Jaquinn jeu
00S OST
00” OSE
OO0€ 0S¢
00¢ OSI
O0Ol OS
0
| |
| |
| |
| 4
| 0
+ 001
+ 002
+ 00€
+ 00¢
+ 00S
+ 009
+ 004
+ 008
- 006 OOO}
i OO}
+ 001
+ O0€1
+ 00v1
+ 00S}
+ 0091
+ OOLI
+ 0081
+ 0061
+ 0002
+ 0012
+ 00z2
183
(Oesw ul) sul) asuodsey sbeisAy
00S OSV
00r OSE
00€
9€ POY
jeu, AQ
owt, esuodsey ebeyaAy
‘Z1°4 eunbi4
JOqUINN [eu
0S¢ 00¢ | |
OSI 001
00}
00¢
00€
007
00S
009
002
008
006
OO00L
OOl |
00cl
00€ |
007 |
00S1
0091
OOZ1
0081
0061
000¢
00l¢e
00¢¢
(Desw ul) ewuly eSuodseyY ebeieay
184
00S + OSP
OO0V OSE
00€ { |
Z€ puouy
yeuy Aq
ewiy asuodsey sbeseny
“eE1°4 eunbi4
JOQUINN JeuL
0S¢e
| I
00¢ OS |
OOL OS
OOl
00¢
00€
OOP
00S
009
002
008
006
0001
OOl}
00ct
OOEl
OOrl
00S |
0091
OOZL
0081
OO6l
000¢
001¢e
00¢¢d
(Oesw ul) awl, esuodsey sBbeusay
185
00S OSt
0O0r OSE
O00€
Gp Puouy
:jeuy Aq
euly esuodsay
ebeieay ‘p1'4
aunbi
JOQuUNN jeu,
OSe {| l
00¢ OS}
001 | —|—
OS | t—
O01
00¢
OO0&
00v
00S
009
002
008
006
000t
OOl tL
00cl
OOEL
OOF |
00S I
0091
OOZ1
008 |
0061
000¢
O0l¢e
00e¢
(DeSuw ul) eu] easuodsey eBbeieay
186
00S OST
OOV OSE
OO0&
9p puoyy
‘jeu, Aq
ewly asuodsay ebeiaay
“s1°4 aunbi4
JOQUINN jeu
0S¢ 00¢
OS | 001
OS | | : i
001
00¢
O00€
OOP
00S
009
002
008
006
0001
OO} |
00c |
O00 |
OOr
00S |
0091
OOZI
0081
0061
000¢
0OLe
00c¢
(Dasw ul) sul] esuodsey ebeusay
187
00S OS
OOr OSE
O0€ | |
8p Psoyy
:jeu, Aq
aul asuodsay
ebeuany ‘91°44
eunbi4
JEQuNN [el
0Ge 002 |
OS | | {
OOL OS
001
00¢
OO0€
007
00S
009
002
008
006
000l
OOl |
00d}
O0€ |
00r I
00S |
0091
OOZI
008 t
0061
000¢
001d
00¢¢
(oasw ul) eu, esuodsey sBessAy
188
00S OSV
00V OSE
00€ | |
ZG psouy
:jeuy Aq
au) asuodsay
ebesny
*/}°4 eunbi
JequnN jeuL
OS¢ { i
00¢ OS 1
OOL
002
00&
007
00S
009
002
008
006
0001
OOLL
00cl
OOE L
ooVl
OOSt
0091
OOZI
O08 t
OO6I
000¢
00l¢e
00c¢
(Oaswi ul) sul, esuodsaey aBbeisay
189
89 POU
‘yeu Aq
euly asuodsay sbeieay
“g}°4 eunbi-y
Jaqunn
seul
00S OSY
OOP OSE
OO0€ 0S¢
00¢ OSI
00} OS
0
| |
| |
| |
| 0
| i
| |
i |
| ]
| I
OOL
7 00¢
+ OO0€
+ OOD
+ 00S
+ 009
+ 002
+ 008
+ 006
- O00}
r OOLL
- O0Zc!
7 OOEl
qt O01
y OOS1
y~ OO9L
+ OOZt
+ 0081
+ OO6I
+ 000¢
+ O0l2
—— 00¢¢
190
(DeSwW ul) su] esuodsey eHbeusay
VITA
Mark Lee McMulkin was born on May 17, 1966 in Great Bend, Kansas.
He received a B.S. degree in Mechanical Engineering from the University of
Idaho in May 1989. While pursuing his Master Degree at Virginia Polytechnic
Institute and State University (Virginia Tech), he worked as a Graduate
Research Assistant for two years performing research on the Ternary Chord
Keyboard and finger speed. He is a member of Tau Beta Pi and the Human
Factors Society. He will be working on his Doctoral Degree in Human Factors