usability issues concerning child restraint system harness design

Accident Analysis and Prevention 35 (2003) 341–348

Usability issues concerning child restraint system harness design

Christina M. Rudin-Browna,∗, Jason K. Kumagaib, Harry A. Angelb,Kim M. Iwasa-Madgeb, Y. Ian Noya

a Ergonomics Division, Road Safety and Motor Vehicle Regulation Directorate, Transport Canada,330 Sparks Street, 8th Floor, Ottawa, Ont., Canada K1A 0N5

b Human Systems Incorporated, 111 Farquhar Street, Guelph, Ont., Canada N1H 3N4

Received 6 June 2001; received in revised form 13 December 2001; accepted 14 December 2001

Abstract

A study was conducted to assess usability issues relating to child restraint system (CRS) harness design. Four convertible child restraintsystems representing a wide variety of design features were used. Forty-two participants installed two child test dummies in both forward-and rear-facing configurations either inside or outside a test vehicle. Observer-scored checklists determined the degree to which harnesseswere installed correctly. Participant-scored questionnaires evaluated the ‘ease-of-use’ of various design features. While the percentageof correct installations exceeded 83% for all designs when installed in the forward-facing configuration, in the rear-facing position (thatintended for children under 9–10 kg), there was a significant (between 65 and 89%) percentage of incorrect installations for all models.This finding is of particular interest and may be indicative of a more generalized problem with ‘convertible’ CRS designs when they areused in the rear-facing configuration. Furthermore, while certain design features were perceived by users as providing significantly betterprotection in the event of a collision, these also tended to be the features that were misused most often. The benefits and costs of variousdesign features are discussed, and a method to test harness design usability is presented.© 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Occupant protection; Road safety; Injury prevention; Infants; CRS

1. Introduction

In Canada, approximately 30% of accidental deaths ofchildren age five and under are the result of motor vehiclecollisions (Statistics Canada, 1997). In 1997, child restraintsystems (CRSs) were estimated to be improperly used 32.3%of the time (Transport Canada, 1998). During the same year,45 vehicle occupants under age five were killed and another79 injured in motor vehicle collisions (TRAID, 1997).

Transport Canada’s mandate includes regulating certainafter-market products such as CRSs. Accordingly, the RoadSafety Directorate of Transport Canada has undertaken re-search into various aspects of CRS design and use, con-tributing towards the development of a universal attachmentsystem (Pedder et al., 1994), and developing a set of regula-tions for user-ready tether, and universal lower, anchoragesfor child restraints (Canada Gazette, 1998, 2001).

A number of factors determine whether a child is prop-erly restrained in a motor vehicle. These include the choiceof restraint type for the child’s size (Weinstein et al., 1997),

∗ Corresponding author. Tel.:+1-613-998-1947; fax:+1-613-990-2913.E-mail address: [email protected] (C.M. Rudin-Brown).

the choice of vehicle seating position where the CRS is in-stalled (Braver et al., 1997), how the CRS is installed inthe vehicle (Turbell et al., 1993), and how the child is in-stalled in the CRS harness (Hummel et al., 1997). A 1992vehicle survey found that the most potentially dangerous er-ror relating to the improper restraint of children involvedthe incorrect installation of the CRS in the vehicle (Wilsonet al., 1994). Therefore, initial Transport Canada CRS us-ability research focused on this problem (Noy and Arnold,1995). While there remains an explicit need for CRS designsto allow easy installation in the vehicle (Noy and Arnold,1995; Wilson et al., 1994; Seyer, 1993; Turbell et al., 1993),their performance in a collision also depends on whetherthe child is correctly installed in the CRS harness (Hummelet al., 1997).

A correct harness installation is not limited to attachingand tightening the harness, but extends to other issues suchas selecting the appropriate shoulder strap height, the appro-priate configuration (forward- versus rear-facing) and reclineposition for a child’s size, and using available padding ap-propriately. Deficiencies in any of these areas can seriouslycompromise a child’s safety in the event of a collision, asdemonstrated in experimental crash tests using incorrectly

0001-4575/02/$ – see front matter © 2002 Elsevier Science Ltd. All rights reserved.PII: S0001-4575(02)00009-X

342 C.M. Rudin-Brown et al. / Accident Analysis and Prevention 35 (2003) 341–348

installed child test dummies (Hummel et al., 1997) andreal world studies assessing child occupants’ injury patterns(Weinstein et al., 1997; Gotschall et al., 1997). To reducethe likelihood of being installed incorrectly, CRS harnessdesigns should be ‘useable’, or convenient to use (Hummelet al., 1997; Wilson et al., 1994). This issue is particu-larly important in light of the tendency of some parents to‘graduate’ their children to a booster seat or seat belt beforethey meet the size and weight criteria (Transport Canada,2000; Hummel et al., 1997). This tendency may be due toa reluctance to use CRSs that are perceived as overly com-plicated and cumbersome compared to a vehicle’s seat beltsystem.

Several researchers have looked specifically at the issueof CRS usability. Bell (1991), adapting a usability methodoriginally designed to test the child-resistance of packaging,approached couples in a store and asked them to install theirchild in a booster seat after first installing it in a vehiclemock-up. Ten of the 20 couples made mistakes, includingtwo that were considered to be very serious. Consequently,and based on these findings, Bell (1993) proposed an In-ternational Organization for Standardization (ISO) standardthat specifies requirements and test methods for evaluatingthe installation of CRSs in vehicles, and children in CRSs.The standard was recently published (ISO, 1999) and usescorrectness and speed of installation as criteria for judgingwhether a CRS is ‘useable.’ It also provides a method forpredicting the effects of CRS misuse. Adapted from thefailure mode and effects analysis (FMEA) procedure usedto predict product failures in the aerospace and automobileindustries, misuse mode and effects analysis (MMEA) isspecific to CRSs (Czernakowski and Müller, 1991, 1993).By using this method, a researcher is able to predict andassess the likelihood of potential instances of CRS misuse,as well as their effect on safety. While providing a goodbasis for assessing CRS misuse, the ISO standard does notallow for the complete assessment of problem areas, particu-larly those that are not immediately obvious to the observer.A more thorough method would assess not only installationperformance, but also subjective ratings and feedback fromusers, to obtain a deeper understanding about CRS usability.

Decina and Knoebel (1997) assessed patterns of CRS mis-use in four American states and found that CRSs were cor-rectly installed in only 20.5% of observed cases. A correctinstallation was defined as demonstrating the appropriate useof the following elements: CRS direction (forward- versusrear-facing), vehicle seat belt routing, harness buckle, strapsand chest clip, and locking clip (on vehicle seat belt). Rec-ommendations included: focusing public safety messages onthe proper use of CRSs, providing clearer instruction forproperly using CRSs, increasing the awareness of the CRSmisuse problem, and considering CRS design modifications,especially on features that have frequent misuse.

A more recent study assessed the likelihood of CRS mis-use among three types of forward-facing harness design andfound that at least 89% of 438 observed cases demonstrated

at least one form of misuse (Arbogast et al., 1999). The mostprevalent mistake related to the installation of the CRS inthe vehicle, but also included not securing the child snuglyin the seat (50% of cases). While there was no significantdifference in overall misuse rates by harness type, twistedharness straps were more common among 5-point harnessesthan among swing-shield or T-shield designs. Furthermore,there was a trend towards not using the chest clip (the fas-tener that holds the harness straps over the child’s shoulders)in both swing- and T-shield, as opposed to 5-point designs.These results point to a need for more systematic researchinto the usability of CRS harness designs.

The Insurance Corporation of British Columbia (ICBC)recently assessed the usability of a number of availableCRSs, infant restraint systems and booster seats (ICBC,1999). Using predefined criteria, the seats were rated asgood, average, or fair in a number of categories. Categoriesrelating to installing the child in the harness included easeof tightening and loosening the harness, number of harnessheight adjustment positions, and ease of securing and re-leasing the harness buckle or latch plate. The ICBC ratingsystem was designed to inform CRS users of the importanceof purchasing a CRS that is not only effective, but easy touse as well. There remains, however, a paucity of empiri-cal data characterizing what, in actual fact, makes a CRSharness usable. Furthermore, while evaluations of specificCRS models can provide useful information to consumers,it would be even more helpful if all CRSs were designed tomeet basic usability criteria.

The purpose of the present study was to assess theease of installing children in representative CRS harnessesand to generate a reliable test method for assessing CRSharness usability. The usability of a variety of currentlyavailable harness design features was assessed by having arepresentative sample of users install three child test dum-mies (dressed in either summer or winter clothing) in fourCRSs, both inside and outside of a vehicle.

2. Method

2.1. Participants

A total of 42 people participated in the study (18in-vehicle, 24 out-of-vehicle). The average age was 37, andranged from 23 to 72 years. Fifty-five percent of participantswere female and the ratio of experienced to inexperiencedusers was approximately 2:1. This number reflects the ratioof other-than-firstborn to firstborn births in Canada. Parti-cipants were selected to represent the anthropometry of theCanadian user population (5th percentile female to 95thpercentile male), with extremes of the population beingover sampled. Participants were recruited from a variety ofsources, including child care centres, hospitals, universities,local businesses and a seniors’ centre, and were paid fortheir participation. Screening ensured that participants had

C.M. Rudin-Brown et al. / Accident Analysis and Prevention 35 (2003) 341–348 343

Fig. 1. Harness types (from left to right: swing-shield, 5-point andT-shield).

no physical or mental impairments, no formal or extensiveCRS training, and no previous involvement in CRS designor evaluation. All test groups were balanced across age,gender, experience level (none, some, or significant), andanthropometry.

2.2. Equipment

Four commercially available CRSs (labeled CRS A–D)representing a range of available features were used. Allwere convertible seats, meaning that they could be usedin both forward- (that intended for children weighing be-tween 10 and 18 kg and standing 67–102 cm tall) andrear-facing (that specified for infants weighing up to 9–10 kgand measuring less than 67 cm) configurations. Harnesseswere swing-shield (CRS A), 5-point (CRS B and C) orT-shield (CRS D) designs (see Fig. 1). All identifyinglabels and markings on the CRSs and in the instructionmanuals were covered to eliminate potential sources ofbias.

To minimize variability in the test procedure and be-cause test sessions were 3 h in length, a decision wasmade to use child test dummies rather than real children.Two child test dummies were used in each phase. Thein-vehicle portion of the study used a 6 (8 kg, 66 cm) anda 9-month (11 kg, 72.5 cm) dummy. The out-of-vehicleportion used the same 6- and a 36-month (17 kg, 98 cm)dummy. Dummy sizes were selected to represent the up-per manufacturer-recommended weight and height rangefor rear-facing CRSs, and the upper and lower limits forforward-facing CRSs. Each dummy wore a tag indicat-ing its height and weight. For the 9-month dummy, theactual weight measurement was exaggerated (from 9 to11 kg) so that participants would be able to assign it to aforward-facing configuration in all cases (one CRS spec-ified a minimum weight of 10 kg for its forward-facingconfiguration).

Test dummies were dressed in either summer or win-ter clothing. Summer clothing consisted of a lightweightsweatshirt-type top and bottom, and winter clothing consis-ted of undergarments, a sweat suit, sweater, and a one-piecesnowsuit. All dummies wore diapers in all conditions.

The test vehicle used for the in-vehicle phase was a 4-door1998 Toyota Camry. The front seats were moved as far for-ward as possible and two CRSs installed in the right and

left rear outboard seating positions. For the out-of-vehiclephase, participants installed test dummies in the CRSs on atable.

2.3. Procedure

For both the in-vehicle and the out-of-vehicle phases, eachparticipant evaluated all CRS designs within the assignedclothing condition for two dummy sizes, corresponding toone forward- and one rear-facing configurations. Participantswere instructed to imagine they had purchased the CRS,brought it home, and had only the manufacturer’s instruc-tions and their own abilities to determine how it should beinstalled. There was no time limit specified. Once the par-ticipant had completed each installation, an observer com-pleted anEvaluation Checklist. The participant then filledout aSubjective Questionnaire that assessed how easy theyfound each task to complete, how confident they were thatthey had performed each task correctly, how confident theywere with respect to the safety of each feature, and the phys-ical demand of each task. Once a participant had installedboth test dummies in all four CRSs, they completed aFea-tures Questionnaire, which assessed how acceptable theyfound each CRS feature to be and which of the availablefeatures in a category they preferred overall. They then par-ticipated in an exit focus group with a facilitator and threeother participants.

2.4. Severity scores

The severity of potential usability errors for each CRSwas determined a priori using Czernakowski and Müller’s(1991, 1993) MMEA procedure. Three subject matter ex-perts with backgrounds in CRS forensics and usability wereasked to rate, on a scale from 0 to 10, an error’s probableeffect on safety, with 10 indicating the most negative ef-fect. Final severity scores for each potential error were de-termined by averaging the subjective, independent ratings ofall experts (Table 1). Severity scores of four or more wereconsidered unacceptable and likely to compromise the effec-tiveness of the CRS in the event of a collision (Czernakowskiand Müller, 1993).

2.5. Risk priority numbers (RPN)

The RPN is a composite measure of a potential error’sseverity score and the frequency with which the error actu-ally occurs during testing. Typically, this number is derivedusing a subjective scale similar to the severity score ratingscale, with 0 representing ‘no misuse’ and 10 represent-ing ‘misuse almost inevitable’ (ISO, 1999). In the presentstudy, however, the RPN was based on an actual count ofthe number of participants demonstrating a particular errorand did not depend on a subjective rating. Due to the un-equal number of participants in each phase (in-vehicle versusout-of-vehicle) and in order to be able to compare results to


Table 1Severity scoresa

Error Severity score (rear-facing) Severity score (forward-facing)

Inappropriate recline position 3 4.6Harness latch plate/buckle not secured 10 10Chest clip not attached 2.3 2Chest clip level too high 2.5 2.5Chest clip level too low 2 1.5Harness loose; 2 fingers (3.2 cm; 1.25 in.) 1.7 1.3Harness loose; 3 fingers (4.8 cm; 1.87 in.) 4.3 3.7Harness loose;≥4 fingers (6.35 cm; 2.5 in.) 6.7 6.3Shoulder height; slots too high 6.3 1.7Shoulder height; slots too low 2.3 2.3Splitter plate not fully attached 7.3 7.3Harness shoulder strap twisted 2.7 1.3Harness leg/lap strap twisted 2 2.7Harness crotch strap twisted 3.5 3.5Inappropriate use of padding 4.7 –Overhead bar not adjusted snugly 0.7 (position 1), 1.3 (position 2) 1.3

a 0: no effect on safety; 10: very severe failure.

other studies, the number of participants was normalized ton = 100. By doing this, an RPN value for each error couldpotentially range from 0 to a maximum of 1000, if 100% ofparticipants committed an error of severity 10.

2.6. Statistical analyses

The percentage of correct installations and total RPNs forall CRSs were subjected tot-tests for dependent samples.Participants’ ratings of the overall ease of adjusting the har-ness were subjected to a four-way repeated measures mixedANOVA with clothing condition, study phase, installationconfiguration, and CRS as dependent variables. Individualfeature preferences from theFeatures Questionnaires arepresented in terms of percentages and were not subjected tostatistical analysis.

3. Results and discussion

3.1. Percentage of correct installations

An indicator of CRS harness usability is the percent-age of installations that are performed correctly. To derivethis measure for each CRS and configuration, the severityscores for all of the usability errors that occurred duringeach participant’s installation were evaluated. A CRS wasconsidered correctly installed if there were no errors hav-ing a severity score of four or more. The percentage ofcorrectly installed CRSs was significantly greater for theforward-facing, than the rear-facing, configuration,t(6) =11.45, P < 0.01. In terms of overall usability, a CRS wasconsidered to be acceptable if at least 85% of the instal-lations were done correctly (Fig. 2). This percentage, con-sistent with the ISO standard, was chosen prior to datacollection, and identifies CRS designs that are problematic,while at the same time accounting for individual differences

Fig. 2. Percentage of correct installations.

in user performance. While all but one CRS met the cri-terion when installed in the forward-facing configuration,none met it when installed rear-facing.

3.2. Risk priority numbers (RPNs)

RPNs for errors that were common to all four CRSs weresummed to create a ‘total RPN’ score for each CRS. Thesescores were significantly higher when the CRS were ad-justed in the rear-facing, as opposed to the forward-facingconfiguration (794 versus 414),t(3) = −9.04, P < 0.01.This indicates that more errors were made, and the overallpotential severity of the errors was greater, when the CRSswere installed rear-facing.

To assess which errors are most serious, it is helpful tocompare error RPNs to a criterion, or threshold value. Basedon results of an earlier usability study done by TransportCanada (Noy and Arnold, 1995) and in consideration of thecriterion RPN value suggested by Czernakowski and Müller(1993), a normalized criterion RPN value of 42 was chosenbefore testing began. Values exceeding this critical numberare considered unacceptable and likely to compromise the


effectiveness of a CRS in the event of a collision. All fourCRSs had several features that generated RPN values abovethis criterion. For example, inappropriately positioning therecline, positioning shoulder straps too high, not fully attach-ing the splitter plate (the steel plate at the back of the CRSwhere the harness straps meet) that had only one slot for bothharness straps, and not adjusting the 5-point harness on CRSC tightly (≥4 fingers or 6.35 cm width of slack) contributedmost to the overall poor performance in the rear-facing con-figuration. The latter three errors also had the highest asso-ciated severity scores (6.3, 7.3, and 6.7), adding to the poorobserved performance.

3.3. Ease-of-use ratings

Participants rated the ease-of-use of a number of tasksrelating to harness installation using a seven-point Likertscale. Usability ratings indicate that participants experiencedmore difficulty with the CRSs when they were rear-facingthan forward-facing,F(1, 38) = 31.1, P < 0.001, whenthey were used in-vehicle than out-of-vehicle,F(1, 38) =11.4,P < 0.002, and when the dummies were dressed inwinter, rather than summer clothing,F(1, 38) = 5.5, P <

0.03. Not surprisingly, difficulty in adjusting the harnesswhen the CRS were rear-facing was more pronounced whenparticipants were required to adjust it in the vehicle, as op-posed to outside the vehicle,F(1, 28) = 32, P < 0.001.Several usability issues were revealed that might have con-tributed to the unsatisfactory performance of the CRSs inthe rear-facing mode:

• In the vehicle, access to the harness tightener was limitedin all cases due to the front of the CRSs being pushed upagainst the vehicle seat back. Because of this, not onlydid participants have difficulty getting enough leverageto tighten the harness, but the angle at which they wererequired to pull on the harness tightener actually causedthe ‘flip’-type harness release (where the user has to liftup the release tab) to open, making the harness loosenslightly before locking.

• In the vehicle, access to harness adjusters that were lo-cated on the front of the CRS seat base (CRSs A and D)was restricted, making it difficult to press the latch releasebutton.

• The swing- (CRS A) and T-shields (CRS D) tended toobstruct access to the CRS, making it difficult for parti-cipants to adequately position, secure, and remove the testdummy, especially inside the vehicle.

• In certain models (CRSs A and D), it was necessary toroute the vehicle seat belt directly over the seat of theCRS, which seriously limited participants’ ability to in-sert the harness’ metal tongue in its slot on the CRSseat.

Focus group comments suggested a number of limitationsof ‘convertible’ CRSs when used in the rear-facing con-figuration. In general, most participants felt that the CRSs

were too big for infants, leaving excessive space between the6-month dummy and the harness. In particular, many parti-cipants commented that the size of the T-shield (CRS D),combined with the chest clip above it, resulted in the harnessbeing unacceptably loose, especially when the dummy wasdressed in summer clothing. Repeated attempts to tighten theharness failed due to the chest clip contacting the dummy’sneck. Participants perceived that in the event of a collision,this limitation of the T-shield would adversely affect thesafety of a child. Most participants reported that they wouldchoose a dedicated infant restraint system for smaller in-fants. Apart from the perceived better fit offered by thesesystems, they allow the child to be positioned and the har-ness adjusted before entering the vehicle, thus improvingusability and reducing the likelihood of disturbing the in-fant. Moreover, participants liked the added convenienceof being able to use an infant restraint system as a babycarrier.

3.4. Ease-of-use ratings and RPNs

An important consideration when designing CRS har-nesses is whether there is feedback provided to the userwhen installation errors are made. If not, a harness may berated as easy to install without the realization that a taskhas been performed incorrectly. Total RPNs for each CRSin each configuration were cross tabulated with ease-of-useratings (see Fig. 3). This cross tabulation resulted in fourseparate categories of usability tasks: (1) those that re-sulted in few errors (low RPNs) and their ease-of-use wasrated as acceptable (true positives), (2) those that resultedin few errors but their ease-of-use was rated as unaccept-able (apparent to user if done incorrectly; false negatives),(3) those that resulted in errors and their ease-of-use wasrated unacceptable (true negatives), and (4) those that re-sulted in errors and their ease-of-use was rated acceptable(false positives).

It is this last category of usability tasks that, for two rea-sons, is of particular concern with regards to safety. First,if it is not clear to the user that a task has been completedincorrectly, s/he may experience false confidence in his/herperformance of the task. This, in turn, might make the userless likely to attempt to correct the error than if the featurehad been rated difficult to use in the first place. Second, CRSmanufacturers may not attempt to redesign features that arerated as acceptable by users, even if errors associated withthese features could result in serious consequences. Severalusability errors fell into the false positive category and wereconsidered to be likely to compromise the effectiveness ofthe CRS in the event of a collision. The five most seriouserrors are indicated with numbers in Fig. 3. They include:

(1) adjusting the shoulder strap height too high when usingthe moveable backrest in the rear-facing mode (CRS C);

(2) not fully attaching the one-slot splitter plate (CRS B)when rear-facing;


Fig. 3. Ease-of-use rating by risk priority number (RPN) for all errors.

(3) adjusting the 5-point shoulder strap height too high inthe rear-facing mode (CRS B);

(4) adjusting the 5-point shoulder strap height too high usingthe moveable backrest forward-facing (CRS C);

(5) adjusting the 5-point harness with≥4 fingers (6.35 cm)of slack (CRS C).

Other serious errors in the false positive category included:

• positioning the recline inappropriately in the CRSs thatwere limited to one rear-facing and two forward-facingrecline positions (CRS A and D);

• incorrectly attaching the insertion-type chest clip (one thatrequires the user to thread the harness straps over it) (CRSsA and D);

• positioning the insertion chest clip too high (CRS D) andthe combination (buckle and insertion) chest clip too low(CRS C);

• not tightening the swing- and T-shield harnesses ade-quately when in the forward-facing mode (CRSs A andD);

• using the padding inappropriately in the swing-shield(CRS A) when rear-facing.

3.5. Feature questionnaires

Most participants (44%) chose photographs in the instruc-tion pamphlet as their preferred instruction format; however,all of the instruction types were rated as being acceptableby at least 88% of participants. Indicators on the CRS thatused arrows to show the correct recline position were pre-ferred to level indicators by 70% of participants, while 61%of participants felt that having no indicators on the CRS wasunacceptable.The 5-point harness design was preferred by53% of participants, compared to 38% for the swing-shieldand 10% for the T-shield, with 32% of participants rating theT-shield negatively. Despite these preferences, the T- andswing-shield harnesses (CRSs A and D) were adjusted cor-rectly more often than either of the 5-point designs (CRSs Band C) in the rear-facing mode (Fig. 2). Furthermore, error

RPNs relating to harness tightness were highest for the two5-point harness designs. Focus group comments indicatedthat participants perceived the 5-point and swing-shield har-nesses as being easier to use than the T-shield. As well,the 5-point harnesses were believed to be safer and moresecure for smaller children, and to offer better weight dis-tribution over the harness (more impact absorption) in theevent of a collision. It is possible that participants also be-lieved the 5-point harness design would be better able tokeep real children in position than either the swing- or theT-shield. Interestingly, a recent survey of CRS use found thatwhile 31% of heavier children (those over 13.5 kg or 30 lb.)were installed in swing-shield harnesses, and 29% were inT-shields, only 7% of heavier children were in 5-point har-nesses (Pedder and Legault, 1999). This raises the interest-ing possibility that, in reality, 5-point harnesses may not beas easy to use (at least for heavier children) as reported. Thiswould explain the finding that parents of heavier childrenchoose to use other CRS harness types or else ‘graduate’their children to booster seats and seat belts earlier thanrecommended.

A harness adjuster located on the CRS seat, as opposedto the seat base, was preferred by 66% of participants, dueto its enhanced accessibility. A slight majority of partici-pants (59%) preferred the ‘flip’-type style of harness release(found on CRSs A, C, and D) to the pull strap, which wasrated unacceptable by 17% of participants. Focus groupcomments and experimenter observations indicated that thepull strap harness release strap was often confused with theharness tightener strap, which was located directly below it.Because the strap ends looked similar and were positioned inclose proximity, participants tended to pull on the incorrectstrap to tighten or loosen the harness. Also, if the harnesswas adjusted very loosely, the strap ends would withdraw tothe edge of the seat, making it difficult to locate the straps,particularly when the CRS was installed rear-facing in thevehicle.

The moveable backrest on CRS C was, by far, the mostpreferred type of shoulder strap height adjuster, with 79%


of participants ranking it first. Unfortunately, incorrectadjustment of this backrest fell into the false positive cat-egory of error (see (1) and (4) in Fig. 3), indicating thatimproved feedback regarding the backrest’s position is re-quired. Interestingly, this type of backrest alleviates thepotentially serious problem that can occur if the shoul-der strap height is selected incorrectly on forward-facingCRSs. As only the uppermost shoulder strap slot on mostmodels of CRS is reinforced (and thus, able to withstandphysical forces generated in the event of a collision), thisslot must be used in the forward-facing configuration. Ifthe CRS harness is not adjusted in this manner, there isa risk that the slot will fail in the event of a collision.A CRS with a moveable backrest, therefore, is not onlyeasier to use, it also eliminates potential errors in select-ing the correct slot or the need to reinforce all availableslots. Conversely, most participants expressed difficulty inre-threading harness straps through slots. Performing thetask was so difficult for most participants that experimenterseventually had to provide a supplementary aide (a butterknife).

Most participants (95%) preferred a splitter plate with twoslots for inserting the harness ends. RPNs for the single-slotsplitter plate (CRS B) not being fully attached were highwhile the ease-of-use of this feature was rated as acceptable,indicating a false positive situation. Because of this finding,a single slot splitter plate is not recommended. Although thesplitter plate on CRS B was not preferred, its double-loopharness length attachment was. Sixty-seven percent of par-ticipants rated this feature as being preferable to a singleloop system. Interestingly, however, all participants who in-stalled this seat outside of the vehicle (and who, therefore,had the opportunity to adjust the harness using the differentloops) overlooked the feature.

Although observations suggested that most participantswere unaware of its dual functionality, the combination in-sertion and buckle chest clip on CRS C was the preferreddesign for 67% of participants, despite some comments thatthis type of chest clip could be dangerous to a child andthat it was ‘flimsy.’ The preference for the combination typemay have arisen due to limitations found with the other twotypes, which were also viewed as being ‘flimsy.’ As well,the dedicated buckle type was found to jam and pinch theuser’s fingers.

Finally, 63% of participants preferred the CRS designsthat offered only three, as opposed to four, recline positions.Focus group comments indicated that participants felt that inorder to minimize the number of required adjustments andto prevent opportunities for wear and tear, CRSs should bedesigned with as few moving parts as possible.

4. Conclusions and recommendations

In the forward-facing configuration, all CRS harnesseswere correctly adjusted in at least 83% of cases. On the

other hand, when rear-facing, all four CRSs were incorrectlyadjusted between 65 and 89% of the time. This findingquestions the appropriateness of using ‘convertible’ CRSs inthe rear-facing configuration (i.e. that specified for childrenunder 9–10 kg). Further, while the 5-point harness was theoverall preferred harness design, the T- and swing-shieldharnesses were correctly adjusted more often in therear-facing configuration.

Results highlight the importance of providing adequateperformance feedback to the user. A number of potentiallyserious usability errors that would have gone unnoticed bythe user (false positives) were identified when ease-of-useratings were examined in relation to error RPNs. Most ofthese errors could have been avoided if feedback in theform of indicators or instructions directly on the CRS, hadbeen provided. For example, arrow indicators that indi-cate appropriate recline position and shoulder strap heightwould have reduced the frequency with which participantsinappropriately adjusted these features. Not attaching theinsertion-style chest clip properly and incorrectly position-ing all chest clip types would similarly have been avoided ifa pictogram specifying correct installation and positioninghad been affixed directly to the chest clip. Errors resultingfrom inadequate tightening of the T- and swing-shield har-nesses could have been reduced by affixing labels to theharness straps that specify the appropriate harness tightness(i.e. less than one finger-width of slack). Another, moreexpensive, method might be to install some kind of ten-siometer that would indicate appropriate harness tensionon the pull-strap harness tightener. It is recommended thatmanufacturers identify the minimum limits of force that arerequired to make the appropriate adjustments and make surethese are reasonable for the general population. If possible,appropriate harness tension should be obvious to the user,regardless of the clothing that the child is wearing. For ex-ample, harness tension indicators on the pull-strap harnessadjuster could be used to indicate the amount of tensionachieved. Clearer, more conspicuous, instructions relatingto the appropriate use of padding would preclude errorsrelating to this feature. A label or pictogram directly on thesplitter plate would likely have increased the probability ofusers being successful when attaching the harness straps toit. Also, as stated earlier, single-slot splitter plates are notrecommended.

Finally, results indicate that correct adjustments should:(a) not act in a manner contrary to user expectations, and (b)should be kept as simple as possible to perform. The reclineon CRS B did not adjust into a secure position. Despite thisfeature being highlighted in the instruction manual and onthe CRS itself as being appropriate, it was counter-intuitiveto most participants, and resulted in frustration and lowerratings of usability.

Manufacturers are encouraged to assess current CRS mod-els, as well as ongoing designs, using the methodology de-scribed in this report. An impartial observer should performthe usability tests, using a predetermined checklist that is


specific to each CRS and that assesses all of its features.For a participant’s CRS harness installation to be consid-ered correct, there should be no errors having an associatedseverity score of four or more. For a CRS to ‘pass’ the us-ability criteria, at least 85% of installations shall be per-formed correctly. Also, at least 85% of participants must ratethe ease-of-use of each feature as acceptable (i.e.≥4 on theLikert scale). Where ease-of-use is rated as acceptable butusability errors are present, appropriate safeguards shouldbe implemented, either by improving instructions and label-ing or by re-designing the feature. Tests should be repeatedonce a safeguard has been implemented.

Feature categories should include: (1) clarity of instruc-tions, (2) overall user acceptance, (3) ease of placing thechild (dummy) in the CRS, (4) ease of adjusting the seat re-cline, (5) ease of adjusting shoulder strap height, (6) ease oftightening the harness straps, (7) ease of loosening the har-ness straps, (8) ease of securing latch plate/buckle and chestclip, (9) ease of releasing latch plate/buckle and chest clip,(10) ease of preventing the harness straps from twisting, (11)occupant/user physical comfort. Design improvements willnot only improve the ease with which users install childrenin CRS harnesses, but they will also increase the effective-ness of these devices in protecting children from serious in-jury in the event of a collision.

Acknowledgements

The authors would like to thank Don Day, Murray Dance,and Jennifer Stalmach for acting as subject matter experts,and France Legault and Valerie Lee for providing technicalassistance. The opinions expressed in this manuscript reflectthose of the authors, and are not necessarily those of Trans-port Canada.

References

Arbogast, K.B., Durbin, D.R., Morris, S.D., Winston, F.K., 1999. Childsafety seat misuse: influence of harness type and accuracy of parentalreport. In: Child Occupant Protection in Motor Vehicle Crashes.pp. 165–167.

Bell R., 1991. Misuse test panels: a technique for the identification andprevention of misuse of child restraints. In: Proceedings of the 35thAnnual Conference on Association for the Advancement of AutomotiveMedicine. pp. 45–55.

Bell, R., 1993. ISO/WD 13215-2: Road Vehicles—Child Restraint Sys-tems—Reduction of Misuse Risk—Requirements and Testing Proce-dures for Correct Installation (Panel Method). Report 289, 1993.

Braver, E.R., Witfield, R., Ferguson, S.A., 1997. Risk of death amongchild passengers in front and rear seating positions. SAE Paper 973296.In: Proceedings of the 2nd Symposium on Child Occupant Protection.pp. 9–15.

Canada Gazette, 1998. User-Ready Tether Anchorages (Part II). SOR/DORS/98-457, 30 September.

Canada Gazette, 2001. Universal Lower Anchorages for Child Restraints(Part I). 3 March, pp. 751–811.

Czernakowski, W., Müller, M., 1991. Misuse mode and effects analysis(MMEA): an approach to predict and quantify misuse of child restraintsystems (CRSs). In: Proceedings of the 35th Annual Conference onAssociation for the Advancement of Automotive Medicine. pp. 27–43.

Czernakowski, W., Müller, M., 1993. Misuse mode and effects analysis:an approach to predict and quantify misuse of child restraint systems.Accident Anal. Prev. 25, 323–333.

Decina, L.E., Knoebel, K.Y., 1997. Child safety seat misuse patterns infour states. Accident Anal. Prev. 29, 125–132.

Gotschall, C.S., Eichelberger, M.R., Morrisey, J.R., Better, A.I., Reardon,J., Bents, F., 1997. Injury patterns associated with child restraint misuse.In: Proceedings of the 2nd Symposium on Child Occupant Protection.SAE Paper 973311. pp. 187–193.

Hummel, Th., Langwieder, K., Finkbeiner, F., Hell, W., 1997. Injuryrisks, misuse rates and the effect of misuse depending on the kindof child restraint system. In: Proceedings of the 2nd Symposium onChild Occupant Protection. SAE Paper 973309. pp. 165–180.

ICBC, 1999. Buying a Better Child Restraint. ICBC Road Sense Pamphletno. PI265 (101999).

ISO, 1999. ISO 13215: Road Vehicles—Reduction of Misuse Risk ofChild Restraint Systems. Part 2. Requirements and Test Proceduresfor Correct Installation (Panel Method), and Part 3: Prediction andAssessment of Misuse by Misuse Mode and Effect Analysis (MMEA).ISO, Geneva.

Noy, Y.I., Arnold, A.-K., 1995. Installing child restraint systems in vehi-cles: towards usability criteria. Transport Canada Ergonomics DivisionTechnical Memorandum no. TME 9501.

Pedder, J., Legault, F., 1999. Parking lot survey of child restraint use.In: Proceedings of the 3rd Symposium on Child Occupant Protection.Barcelona, Spain.

Pedder, J., Legault, F., Salcudean, G., Hillebrandt, D., Gardner, W.,Labrecque, M., 1994. Development of the CanFIX infant and childrestraint/vehicle interface system. In: Proceedings of the 38th Confer-ence on Stapp Car Crash. SAE Paper 942221. pp. 235–244.

Seyer, K., 1993. Development of user friendly child restraint attachmentsystems. In: Proceedings of the 2nd Symposium on Child OccupantProtection. SAE Paper 933086. pp. 43–62.

Statistics Canada, 1997. Causes of Death: Shelf Tables. Catalogue no.84F0208XPB.

TRAID (Traffic Accident Information Database), 1997. Transport Canada,Road Safety Directorate.

Transport Canada, 1998. Child Restraint Use in Canada: 1997 SurveyData. Transport Canada Road Safety Leaflet CL 9804(E).

Transport Canada, 2000. Keep Kids Safe: Car Time 1–2–3–4. TransportCanada Road Safety Publication no. TP13511(E).

Turbell, T., Lowne, R, Lundell, B., Tingvall, C., 1993. ISOFIX: a new con-cept of installing child restraints in cars. In: Proceedings of the 2ndSymposium on Child Occupant Protection. SAE Paper 933085. pp.35–41.

Weinstein, E.B., Sweeney, M.M., Garber, M., Eastwood, M.D., Osterman,J.G., Roberts, J.V., 1997. The effect of size appropriate and properrestraint use on injury severity of children. In: Proceedings of the2nd Symposium on Child Occupant Protection. SAE Paper 973310.pp. 181–186.

Wilson, R.J., Grant, B., Hurley, J., 1994. 1992 Observational and Tele-phone Survey of Restraint Use by Children in Motor Vehicles. Trans-port Canada Publication no. TP11943(E).

usability issues concerning child restraint system harness design

Documents