The Rey-Osterrieth (ROCF) and Taylor (TCF) complex figure tests are widely used to assess visuospatial and constructional abilities as well as visual/non-verbal memory. Normative data adjusted to the cultural and linguistic reality of older Quebec-French individuals is still nonexistent for these tests. In this article, we report the results of two studies that aimed to establish normative data for Quebec-French people (aged at least 50 years) for the copy, immediate recall, and delayed recall trials of the ROCF (Study 1) and the TCF (Study 2). For both studies, the impact of age, education, and sex on test performance was examined. Moreover, the impact of copy time on test performance, the impact of copy score on immediate and delayed recall score, and the impact of immediate recall score on delayed recall performance were examined. Based on regression models, equations to calculate Z scores for copy and recall scores are provided for both tests.

Introduction

The Rey–Osterrieth (ROCF; Osterrieth, 1944; Rey, 1941) and Taylor (TCF; Taylor, 1969) complex figure tests are commonly used in both clinical and research settings to assess a variety of cognitive processes, such as memory, motor function, perceptual abilities, and components of executive function (Meyers & Meyers, 1995). For both tests, the standard procedure requires participants to copy an 18-item figure and then, without prior warning, to reproduce it from memory after a 3-min delay (Osterrieth, 1944). Some investigators or clinicians may also include a delayed recall and/or a recognition trial (Strauss, Sherman, & Spreen, 2006).

Based on the fact that the ROCF and the TCF contain equal numbers of elements that are presumably of comparable complexity, both figures have long been assumed to be equivalent (Lezak, 1983). Accordingly, for both these tests, scores are generally interpreted using norms developed for the ROCF (Strauss et al., 2006). However, previous studies have shown that participants obtain significantly higher scores on both immediate and delayed recall trials of the TCF when compared with the ROCF (e.g., Strauss & Spreen, 1990; Tombaugh, Faulkner, & Hubley, 1992; Yamashita, 2006). These results were also observed in test–retest situations, even when the sequence of administration is counterbalanced (Tombaugh, Faulkner, & Hubley, 1992). Moreover, although one study in young adults found comparable scores between a modified version of the TCF and the ROCF (Yamashita, 2006), another study with older adults showed that scores at copy were higher on the modified TCF, but recall/memory performance was comparable between tests (Hubley, 2010). According to Strauss and Spreen (1990), using normative data established for the ROCF may result in an overestimation of memory levels in patients that are given the TCF. As such, norms should be developed for each figure separately. One study sought to establish norms specifically for the TCF in 407 participants aged between 20 and 79 years old (Tombaugh, Faulkner, & Schmidt, 1992). However, an intentional learning paradigm was employed with a subset of these participants (n = 63). Thus, participants knew that they would be asked to recall the figure from memory after a delay. Unfortunately, no normative data are currently available for the original implicit learning TCF task, which is the typical mode of administration for this test.

Sociodemographic variables such as age, education, and sex have been demonstrated to influence performance on neuropsychological tests. In previous studies, a relationship has been demonstrated between age and performance on the ROCF. While a few studies reported no change in copy performance with normal aging (e.g., Hartman & Potter, 1998; Mitrushina, Satz, & Chervinsky, 1990; Vingerhoets, Lannoo, & Wolters, 1998), several authors demonstrated an age-related decline in performance on every condition of the test (e.g., Boone, Lesser, Hill-Gutierrez, Berman, & D'elia, 1993; Caffarra, Vezzadini, Dieci, Zonato, & Venneri, 2002; Machulda et al., 2007; Meyers & Meyers, 1995; Rosselli & Ardila, 1991). Similarly, increasing age has been associated with a decline in the ability to learn and remember the TCF (Tombaugh, Faulkner, & Hubley, 1992; Tombaugh, Faulkner, & Schmidt, 1992;). Regarding the effect of education, some studies have found lower education levels to be associated with poorer scores on the ROCF (Caffarra et al., 2002; Fastenau, Denburg, & Hufford, 1999; Pontón et al., 1996; Rosselli & Ardila, 1991), while others have found performance to be unaffected by education (Boone et al., 1993; Meyers & Meyers, 1995). Finally, some studies revealed small or no differences between males and females on ROCF performance (Berry, Allen, & Schmitt, 1991; Boone et al., 1993; Chiulli, Haaland, LaRue, & Garry, 1995; Meyers & Meyers, 1995; Peña-Casanova et al., 2009), while other researchers reported that sex had an influence on performance, with males performing better on every conditions of the test (Gallagher & Burke, 2007) or on the delayed recall trial only (Caffarra et al., 2002). Neither sex nor education is associated with performance on the TCF (Tombaugh, Faulkner, & Hubley, 1992; Tombaugh, Faulkner, & Schmidt, 1992).

In addition to considering sociodemographic variables when interpreting non-verbal neuropsychological tasks, several authors have also highlighted the influence of culture on ROCF performance, which may be a result of specific training provided through formal education (e.g., Rosselli & Ardila, 1991; Rosselli & Ardila, 2003). Indeed, non-verbal tests may be culturally biased, and great caution is needed when interpreting the results of an individual belonging to a different culture than that of the normative sample (Rosselli & Ardila, 2003). The importance of developing norms adapted to a given culture has also been highlighted by a study showing that local norms are more sensitive and accurate than non-cultural specific norms for identifying cognitive difficulties in older adults (Arsenault-Lapierre et al., 2011). In this respect, normative studies have been carried out for the ROCF in community-dwelling older Italians (Caffarra et al., 2002), English (e.g., Boone et al., 1993; Fastenau et al., 1999; Machulda et al., 2007; Meyers & Meyers, 1995) and Spanish-speaking Americans (Pontón et al., 1996), as well as Spanish individuals (Peña-Casanova et al., 2009). Given these considerations, normative data are needed for Quebec-French individuals. Indeed, these individuals differ from other North-Americans subpopulations, as well as from other French-Canadians and French-speaking people worldwide on several aspects (e.g., spoken language and dialectal variation, education system).

Considering the impact of culture and sociodemographic variables on ROCF and TCF task performance, the present work aimed to establish normative data for both tests in a Quebec-French population. The impact of age, education, and sex was examined for the copy score, immediate recall and delayed recall of the ROCF (Study 1) and the TCF (Study 2). In addition, the impact of copy time on test performance was also examined since it implies longer encoding time. Moreover, performance at copy may influence performance at memory trials. Therefore, the impact of copy score on immediate and delayed recall score and the impact of immediate recall score on delayed recall performance were examined. The two studies included different samples of older adults.

Study 1

This study provides normative data for the ROCF in Quebec-French individuals.

Methods

Participants

Researchers across the province of Quebec (Canada) were invited to share anonymized data from healthy volunteers who had completed the ROCF as part of other research studies approved by local Research Ethics Boards. Secondary data from those studies were used in this normative study. Participants originated from four recruiting sites (i.e., three laboratories in Montreal and one in Quebec City).

All participants included in the study were Caucasian, at least 50 years of age, living independently in the community, and reported French as their mother tongue and usual language. All subjects scored within normal ranges on the mini-mental state examination (MMSE >26; Folstein, Folstein, & McHugh, 1975) or the Montreal cognitive assessment (MoCA >26; Nasreddine et al., 2005) and on the geriatric depression scale (GDS <11; Yesavage, Brink, Rose, & Adey, 1983), indicating no cognitive impairment or depressive symptomatology. A self-reported medical and psychiatric history was also obtained from each participant. All participants self-reported good mental and physical health (e.g., no history of neurological disease, past or current psychiatric illness, traumatic brain injury, medical condition that could interfere with cognitive performance).

The final normative sample consisted of 220 community-dwelling participants (64 males and 156 females), between 50 and 91 years of age (mean age = 67.56 years; SD = 7.51) who had between 3 and 24 years of formal education (mean education level = 14.43 years; SD = 3.69) (see Table 1 for normative sample description). Highly educated men and women of all age ranges are overrepresented in our sample compared with actual Quebec demographics. Also, women were overrepresented in our sample.

Table 1.

Normative samples description

 Age range N (F : M) Education
 
M (SDRange 
ROCF 50–59 31 (15 : 16) 14.65 (2.43) 10–19 
60–64 44 (32 : 12) 14.98 (3.75) 5–23 
65–69 43 (34 : 9) 14.33 (3.68) 6–22 
70–74 69 (52 : 17) 14.33 (3.68) 3–21 
75–79 19 (14 : 5) 14.79 (3.28) 11–22 
80–91 14 (9 : 5) 12.50 (5.03) 5–24 
TCF 70–74 128 (68 : 60) 13.30 (4.27) 5–23 
75–79 166 (78 : 88) 13.61 (4.64) 2–23 
80–91 138 (86 : 52) 11.86 (3.87) 3–20 
 Age range N (F : M) Education
 
M (SDRange 
ROCF 50–59 31 (15 : 16) 14.65 (2.43) 10–19 
60–64 44 (32 : 12) 14.98 (3.75) 5–23 
65–69 43 (34 : 9) 14.33 (3.68) 6–22 
70–74 69 (52 : 17) 14.33 (3.68) 3–21 
75–79 19 (14 : 5) 14.79 (3.28) 11–22 
80–91 14 (9 : 5) 12.50 (5.03) 5–24 
TCF 70–74 128 (68 : 60) 13.30 (4.27) 5–23 
75–79 166 (78 : 88) 13.61 (4.64) 2–23 
80–91 138 (86 : 52) 11.86 (3.87) 3–20 

Notes: These categories of age and education were not used to produce normative data and are for indicative purpose only. One should note that people aged <60 years and 80 years or older are underrepresented in the ROCF sample while people <70 years are underrepresented in the TCF sample. Therefore, results for these age clustered should be interpreted carefully.

ROCF = Rey–Osterrieth complex figure; TCF = Taylor complex figure; M = Mean; SD = standard deviation.

Materials and Procedure

The original French version of the ROCF (Meyers & Meyers, 1995; Rey, 1941) was individually administered to all subjects as part of a larger neuropsychological battery. The only deviation from the standard administration is that participants were given no time limit to copy the figure. They were asked to copy the figure as accurately as possible and while doing so, participants used a series of colored pencils. This procedure allowed the experimenter to preserve a record of the order in which elements of the figure were reproduced. Either 0 or 3 min following completion of the copy, subjects were asked to draw the figure from memory, as accurately as possible and with no time limit. Participants were not informed that they would subsequently be asked to reproduce the figure from memory. Finally, 20 min following completion of the immediate recall, subjects were asked to draw the figure from memory again. The delay interval was filled with verbal neuropsychological tasks. According to Osterrieth's scoring system (adapted by Taylor, 1969), the maximum possible score is 36 for the copy and recall trials. Doctoral students and trained psychometrists administered and scored the test, assigning 0–2 points for each of the 18 elements of the figure based on distortion and placement.

Statistical Analyses

First, to identify the variables influencing performance on the ROCF, a standard multiple regression analysis was performed for each dependent variable (copy score, immediate recall score, delayed recall score, and copy time) with Age, Sex, and Education as predictors. Second, as Copy time may impact performance on recall trials, it was also assessed as a predictor in the regression model for immediate and delayed recall scores. Similarly, Copy score was entered as a predictor in the regression model for immediate recall score. Third, Copy score and Immediate recall score were entered as predictors in the regression model for delayed recall score. Finally, interactions between predictors were tested. When significant, interactions were added as predictors. Because the length of the immediate recall delay (0 or 3 min) had no significant effect on delayed recall scores when entered as a predictor (p = .117), both procedures were grouped into a single analysis. All variables were entered in the analyses as continuous variables except Sex, which was coded as 0 for women and 1 for men. Visual and statistical analyses were conducted to verify the underlying assumptions of the regression model. All statistical analyses were performed using SPSS software (version 16.0) with the α level set at 5%.

Results

Using univariate analyses, it was found that participants who were administered the immediate recall 3-min delay procedure were older (t(218) 2.347, p = .020, d = 0.35), more highly educated (t(218) 2.665, p = .008, d = 0.40), and included a higher proportion of women compared with participants with the 0-min delay (χ2 = 32.27, df: 1, p < .001). However, despite these differences in characteristics, participants did not differ in terms of immediate recall scores (t(218) 1.324, p = .187, d = 0.20) or delayed recall scores (t(180) = 0.936, p = .350, d = 0.15). As such, both procedures were grouped into a single regression analysis.

In this section, results for the copy, immediate recall (0- and 3-min), and delayed recall (20-min) conditions will be presented separately. Descriptive statistics for each condition of the test as well as for copy time are presented in Table 2. One should note that data were missing for 38 participants on the delayed recall trial and for 8 participants regarding copy time.

Table 2.

Basic descriptive statistics for each variable

 Variable N M SD Range 
ROCF Copy score 220 31.30 3.78 17.5–36 
Immediate recall 0-s 64 14.58 6.50 1.5–28 
Immediate recall 3-min 156 15.75 5.76 4–31 
Delayed recall 182 15.25 5.83 2–30 
Copy time 212 239.46 92.62 69–480 
TCF Copy score 432 30.37 2.78 19–36 
Immediate recall 3-min 432 18.20 5.15 4–34 
Delayed recall 430 17.93 4.90 5–33 
Copy time 427 242.89 96.48 80–480 
 Variable N M SD Range 
ROCF Copy score 220 31.30 3.78 17.5–36 
Immediate recall 0-s 64 14.58 6.50 1.5–28 
Immediate recall 3-min 156 15.75 5.76 4–31 
Delayed recall 182 15.25 5.83 2–30 
Copy time 212 239.46 92.62 69–480 
TCF Copy score 432 30.37 2.78 19–36 
Immediate recall 3-min 432 18.20 5.15 4–34 
Delayed recall 430 17.93 4.90 5–33 
Copy time 427 242.89 96.48 80–480 

Notes: ROCF = Rey–Osterrieth complex figure; TCF=Taylor complex figure. Copy time is measured in seconds.

Regression analyses were first run with sociodemographic variables as predictors of copy time and copy, immediate recall and delayed recall scores. Analyses indicated that Age (β = 0.175, p = .010) and Education (β = −0.169, p = .014) explained a significant amount of the variance of the copy time (R2 = .077, F(3, 208) = 5.775, p = .001). More precisely, older and less educated people took more time to copy the figure. Also, analyses showed that Education (β = 0.185, p = .007) explained a significant amount of the variance of the copy score (R2 = .044, F(3, 216) = 3.312, p = .021); less educated people performed worse at copy trial. Copy time was significantly correlated with copy score (r = −.145, p = .035), with higher copy time associated with lower copy score. Next, Age (β = −0.179, p = .006) and Education (β = 0.257, p < .001) explained a significant amount of the variance of the immediate recall score (R2 = .113, F(3, 216) = 9.209, p < .001). Finally, Age (β = −0.187, p = .010) and Education (β = 0.249, p = .001) both explained a significant amount of the variance of the delayed recall score (R2 = .105, F(3, 178) = 6.959, p < .001). For both memory trials, higher age and lower education level were associated with poorer performance.

Regression analyses were subsequently re-run with sociodemographic variables and Copy time as predictors of copy score. Results indicated that Education (β = 0.165, p = .019) explained a significant amount of variance of the copy score (R2 = .057, F(4, 207) = 3.148, p = .015); less educated people performed worse at copy trial. Regression analyses were also run with sociodemographic variables, Copy time and Copy score as predictors of immediate recall scores. Three multivariate outliers were identified according to Mahalanobis distances (p < .001) and were removed from the analyses. Results showed that Age (β = −0.116, p = .042), Education (β = 0.150, p = .010), Copy score (β = 0.215, p = .001), and Copy time (β = −0.314, p < .001) explained a significant amount of variance of the immediate recall score (R2 = .391, F(6, 202) = 21.654, p < .001). Specifically, older, less educated people who took longer to copy the figure and obtained poorer copy score were those who obtained poorer score at immediate recall trial. Immediate recall score was also influenced by the Sex × Copy score interaction (β = 0.232, p < .001). This interaction indicated that immediate recall score was more strongly influenced by copy score in men than in women. Finally, regression analyses were run with sociodemographic variables, Copy time, Copy score, and Immediate recall score as predictors of delayed recall score. Analyses indicated that Copy time (β = 0.087, p = .011), Copy score (β = 0.127, p < .001), and Immediate recall score (β = 0.893, p < .001) explained a significant amount of the variance of the delayed recall score (R2 = .852, F(7, 166) = 136.345, p < .001). Participants who obtained higher copy and immediate recall scores and who took more time to copy the figure obtained higher score at delayed recall. Delayed recall score was also influenced by the Immediate recall score × Copy time interaction (β = 0.083, p = .007). This interaction indicated that longer copy time enhanced performance on the delayed recall in participants with high, but not low, immediate recall score. The bivariate correlation between immediate and delayed recall scores was very strong (r = .908, p < .001) and the semi-partial correlations from the regression model indicated that this relationship remained strong (r = .731) in presence of all other predictors, which had modest associations (p = .116).

Based on the results from the regression models, equations to calculate Z scores for the copy, immediate and delayed recall scores are presented in Tables 3 and 4. Table 3 includes regression equations to calculate Z scores with sociodemographic variables alone, while Table 4 includes regression equations to calculate Z scores with all variables (i.e., sociodemographic, copy time, copy score, and immediate recall score). The application of the equations in clinical practice is explained in the General discussion.

Table 3.

Information to calculate Z scores with sociodemographic variables for the Rey–Osterrieth complex figure

 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.187E − 0.946S − 0.029A + 30.858 3.664 
Immediate recall 0.417E + 0.568S − 0.143A + 18.905 5.679 
Delayed recall 0.393E + 0.000S − 0.145A + 19.388 5.557 
 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.187E − 0.946S − 0.029A + 30.858 3.664 
Immediate recall 0.417E + 0.568S − 0.143A + 18.905 5.679 
Delayed recall 0.393E + 0.000S − 0.145A + 19.388 5.557 

Notes: E = Education in years; A = Age in years; S = Sex (1 = men and 0 = women).

Table 4.

Information to calculate Z scores with all predictors for the Rey–Osterrieth complex figure

 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.166E − 1.018S − 0.019A − 0.005CT + 31.673 3.648 
Immediate recall 0.243E + 0.966S − 0.095A − 0.021CT + 0.353 (CS − 31.2955) + 0.941 (SX(CS − 31.2955)) + 23.147 4.705 
Delayed recall 0.010E − 0.280S − 0.030A + 0.006 (CT − 238.1975) + 0.199CS + 0.868 (IRS − 15.4114) + 0.001 ((IRS − 15.4114) × (CT − 238.1975)) + 11.205 2.289 
 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.166E − 1.018S − 0.019A − 0.005CT + 31.673 3.648 
Immediate recall 0.243E + 0.966S − 0.095A − 0.021CT + 0.353 (CS − 31.2955) + 0.941 (SX(CS − 31.2955)) + 23.147 4.705 
Delayed recall 0.010E − 0.280S − 0.030A + 0.006 (CT − 238.1975) + 0.199CS + 0.868 (IRS − 15.4114) + 0.001 ((IRS − 15.4114) × (CT − 238.1975)) + 11.205 2.289 

Notes: A = Age (years); E = Education (years); CT = Copy time (s); CS = Copy score (max = 36); IRS = Immediate recall score (max = 36); S = Sex (1 = men and 0 = women).

Discussion

The aim of Study 1 was to assess the effect of sociodemographic variables on ROCF performance and establish normative data for the Quebec-French population. Age was found to be associated with copy time as well as immediate and delayed recall scores. The positive association between age and copy time can be explained by the fact that processing speed is known to be highly associated with age, with older adults being generally slower than younger adults (Verhaeghen & Salthouse, 1997). Also, the negative associations between age and immediate and delayed recall scores are consistent with the fact that memory for visuospatial stimuli decreases with age (Haaland, Linn, Hunt, & Goodwin, 1983; Koss, Haxby, DeCarli, Shapiro, & Rapoport, 1991; Tombaugh, Faulkner, & Hubley, 1992). Overall, the results obtained here are in line with previous work that showed an age-related decline in ROCF performance (Boone et al., 1993; Caffarra et al., 2002; Machulda et al., 2007; Meyers & Meyers, 1995; Rosselli & Ardila, 1991).

Results also indicated that participants' education level was associated with all test measures (i.e., copy time, copy score, and immediate and delayed recall scores). Higher education level was associated with faster copy time and better scores on the other measures, which is generally in line with results of previous normative studies (Caffarra et al., 2002; Fastenau et al., 1999; Pontón et al., 1996; Rosselli & Ardila, 1991). Finally, consistent with previous normative studies (Berry et al., 1991; Boone et al., 1993; Chiulli et al., 1995; Meyers & Meyers, 1995; Peña-Casanova et al., 2009), sex was not directly associated with performance on any condition of the test. However, one should note that sex had an impact on the relationship between copy time and immediate recall, which was not examined in previous studies.

Copy time was found to be associated with immediate recall score. Indeed, longer copy time was associated with poorer performance at immediate recall. According to Bennett-Levy (1984), this may be due to the fact that subjects using less effective strategies tend to “lose their way” while copying the figure, resulting in higher copy time. In other words, longer copy time may reflect a loss of set, in that patients may lose track of initial objective. Accordingly, Newman and Krikorian (2001) suggested that the use of an organized strategy at copy, resulting in more efficient encoding, might yield less demanding recall. Consequently, longer copy time may be a result of less efficient organization strategies and therefore lead to lower recall scores. Copy time was also found to be associated with delayed recall score. However, here longer copy time was associated with higher performance at delayed recall trial in participants with higher immediate recall score. This association is more intuitive and can be explained by the fact that longer encoding time may strengthen the memory trace for subsequent retrieval performance. In fact, according to Bennett-Levy (1984), longer copy time may also be the result of better care taken to copy the figure, or better attention paid to details. Consequently, different cognitive processes may underlie immediate and delayed recall; the former might rely more on working memory and organizational abilities and the latter on visual episodic memory consolidation. However, further research is needed to investigate this hypothesis.

Finally, immediate recall score was more strongly related to copy score in men than in women. That is, a higher copy score is more likely to result in a higher immediate recall score in men than in women, a result for which literature does not allow us to provide any possible explanation. Also, immediate recall score was more strongly related to delayed recall when copy time was longer. In other words, as copy time increases, so does the strength of the relationship between immediate and delayed recall scores.

Study 2

This study was aimed at providing normative data for the TCF in the Quebec-French population.

Methods

Participants

Participants were recruited as described in Study 1. The same inclusion and exclusion criteria were used, but participants from Study 2 were recruited independently from Study 1. Thus, no participants were administered both the ROCF and the TCF. In this study, participants originated from only one recruiting site (i.e., one laboratory in Montreal). The final sample consisted of 432 community-dwelling older adults (233 women and 199 men) between 71 and 86 years old (mean age = 77.38 years; SD = 3.92). Years of education varied between 3 and 23 years (mean education level = 12.92 years; SD = 4.23) (see Table 1 for a full description of the normative sample). Highly educated men and women of all age ranges are overrepresented in our sample compared with actual Quebec demographics (Institut de la statistique du Québec, 2006).

Materials and Procedure

The TCF (Taylor, 1969) was administered individually to all subjects using the same procedure, instructions, and scoring method as in Study 1 (Osterrieth, 1944; Rey, 1941). However, in the present study, the 0-min immediate recall trial was not used.

Statistical Analyses

The analyses were identical to those of Study 1. The dependent variables were copy time, copy score, immediate recall score (3-min), and delayed recall score.

Results

Descriptive statistics for each condition of the test as well as for copy time are presented in Table 2. Data were missing for two participants on the delayed recall trial and for five participants regarding the copy time.

Regression analyses were first run with sociodemographic variables as predictors of copy, immediate and delayed recall scores. The results of the regression analyses indicated that Age (β = 0.134, p = .005) and Sex (β = −0.107, p = .029) explained a significant amount of the variance of the copy time (R2 = .046, F(3, 423) = 6.764, p < .001), with women and older people taking more time to copy the figure. Results also indicated that Age (β = −0.144, p = .002), Education (β = 0.216, p < .001), and Sex (β = −0.095, p = .048) explained a significant amount of the variance of the copy score (R2 = .073, F(3, 428) = 11.193, p < .001). More specifically, lower age and higher education were associated with higher copy score. Moreover, women were found to score higher than men. Copy time was not significantly correlated with copy score (r = .075, p = .120). Regarding recall scores, results revealed that only one predictor explained a significant amount of the variance of the immediate recall (R2 = .047, F(3, 428) = 6.972, p < .001) and delayed recall (R2 = .057, F(3, 426) = 8.629, p < .001) scores. It was found that only Age significantly predicted immediate (β = −0.198, p < .001) and delayed (β = −0.213, p < .001) recall scores. On both trials, the oldest participants obtained the lowest scores.

Regression analyses were also performed with sociodemographic variables and Copy time as predictors of copy score. Results showed that Age (β = −0.160, p = .001), Education (β = 0.226, p < .001), and Copy time (β = 0.117, p = .014) explained a significant amount of the variance of the copy score (R2 = .0864, F(4, 422) = 9.905, p < .001). More precisely, older and less educated people who took less time to copy the figure obtained lower scores. Regression analyses were then performed with sociodemographic variables, Copy time, and Copy score as predictors of immediate recall score. Results indicated that Age (β = −0.129, p = .004) and Copy score (β = 0.377, p < .001) explained a significant amount of the variance of the immediate recall score (R2 = .204, F(6, 420) = 17.977, p < .001). Specifically, higher age and lower copy score were associated with lower immediate recall score. Immediate recall was also significantly influenced by the Copy time × Sex interaction (β = −0.213, p < .001), indicating that Copy time was negatively associated with immediate recall in men but positively associated with this variable in women. Finally, regression analyses were performed with sociodemographic variables, Copy time, Copy score, and Immediate recall score as predictors of the delayed recall score. Results indicated that only Immediate recall score (β = 0.887, p < .001) significantly predicted delayed recall score (R2 = .838, F(6, 418) = 360.331, p < .001). More specifically, higher immediate recall score was associated with higher delayed recall score. The bivariate correlation between these variables was very strong (r = .913, p < .001) and immediate recall score remained strongly associated with delayed recall score even when adjusting for all other predictors, as shown by the semi-partial correlation from the regression model (r = .894).

Based on the results from the regression models, equations to calculate Z scores for the copy, immediate and delayed recall scores of the TCF are presented in Tables 5 and 6. Table 5 includes regression equations to calculate Z scores with sociodemographic variables alone, while Table 6 includes regression equations to calculate Z scores with all variables (i.e., sociodemographic, copy time, copy score, and immediate recall score). The clinical use of these equations will be explained in the General discussion.

Table 5.

Information to calculate Z scores with sociodemographic variables for the Taylor complex figure

 Equations to calculate Z scores
 
Z = (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.142E − 0.102A − 0.528S + 36.700 2.685 
Immediate recall 0.083E − 0.261A − 0.184S + 37.378 5.046 
Delayed recall 0.105E − 0.267A − 0.501S + 37.438 4.778 
 Equations to calculate Z scores
 
Z = (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.142E − 0.102A − 0.528S + 36.700 2.685 
Immediate recall 0.083E − 0.261A − 0.184S + 37.378 5.046 
Delayed recall 0.105E − 0.267A − 0.501S + 37.438 4.778 

Notes: E = Education in years; A = Age in years; S = Sex (1 = men and 0 = women).

Table 6.

Information to calculate Z scores with all predictors for the Taylor complex figure

 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.148E− 0.113A − 0.459S + 0.003 (CT − 242.891475) + 36.624 2.669 
Immediate recall −0.023E − 0.170A + 0.032S + 0.699CS + 0.002 (CT − 242.891475) − 0.017 ((CT − 242.891475) XS) + 10.273 4.627 
Delayed recall 0.022E − 0.034A − 0.335S − 0.001CT + 0.074CS + 0.845IRS + 3.163 1.988 
 Equations to calculate Z scores
 
Z= (real score − expected score)/square root of the mean square residual
 
Expected score Square root of the mean square residual 
Copy 0.148E− 0.113A − 0.459S + 0.003 (CT − 242.891475) + 36.624 2.669 
Immediate recall −0.023E − 0.170A + 0.032S + 0.699CS + 0.002 (CT − 242.891475) − 0.017 ((CT − 242.891475) XS) + 10.273 4.627 
Delayed recall 0.022E − 0.034A − 0.335S − 0.001CT + 0.074CS + 0.845IRS + 3.163 1.988 

Notes: A = Age (years); E = Education (years); CT = Copy time (seconds); CS = Copy score (max = 36); IRS = Immediate recall score (max = 36); S = Sex (1 = men and 0 = women).

Discussion

The aim of Study 2 was to assess the effect of demographic variables on TCF performance and establish normative data for the Quebec-French population. Results indicated an age-related decline in the ability to copy the TCF and recall it after 3- and 20-min delays, which corroborate results from previous studies using the intentional learning variant of the TCF (Tombaugh, Faulkner, & Hubley, 1992; Tombaugh, Faulker, & Schmidt, 1992). These findings are also in line with those of Study 1 as well as with previous studies using the ROCF (Boone et al., 1993; Caffarra et al., 2002; Machulda et al., 2007; Meyers & Meyers, 1995; Rosselli & Ardila, 1991). In fact, the negative effect of age on all conditions of the test is consistent with previous work that showed an age-related decrement in organizational abilities (Mitrushina et al., 1990), visuospatial skills, and visual memory (Haaland et al., 1983; Koss et al., 1991).

Education was found to be positively associated with copy score, but not with immediate or delayed recall performance, which is in line with the results reported by Tombaugh, Faulkner, and Schmidt (1992). Moreover, the positive association between education and copy score is consistent with the results obtained in Study 1 as well as with previous studies using the ROCF (Caffarra et al., 2002; Fastenau et al., 1999; Pontón et al., 1996; Rosselli & Ardila, 1991).

In the present sample, sex was not associated with immediate or delayed recall of the TCF. This finding is consistent with results from a previous study using the TCF with an intentional learning paradigm (Tombaugh et al., 1992). On the copy trial, however, women obtained higher scores than men, which might be associated with the fact that women generally took more time to copy the figure than men. However, further research is needed to examine the association between copy time and copy score on the TCF.

Finally, it was shown that longer copy time was associated with better immediate recall score in women, but with worse immediate recall performance in men. It is possible that longer copy time may reflect increased attention to detail or scrupulousness in women, whereas it may be a mark of impairment in men. However, further studies are needed to investigate and clarify this hypothesis.

General Discussion

Considering the impact of culture and sociodemographic variables on ROCF and TCF performance, the present studies aimed to establish Quebec-French normative data for the copy as well as immediate and delayed recall scores of both tests. The impact of sociodemographic factors on task performance was investigated. Also, the impact of copy time on test performance was examined. Because it has been determined that the ROCF and TCF are not equivalent memory measures (Strauss & Spreen, 1990; Tombaugh, Faulkner, & Hubley, 1992; Yamashita, 2006), normative data were developed for each test separately. To our knowledge, this study is the first to establish normative data for the ROCF and the TCF in a French-Canadian sample.

An original aspect of both studies was to measure the association between copy time and subsequent recall scores. Copy time was found to be an important variable in the prediction of performance on immediate and delayed recall for both figures. However, its effect on both was not consistent. For the ROCF, copy time was found to be positively associated with delayed recall score, but negatively associated with immediate recall score. For the TCF, the direction of the association varied as a function of sex. More precisely, it was found that longer copy time was associated with better immediate memory performance in women but worse memory performance in men. Consequently, copy time may represent a contaminated index of performance (Bennett-Levy, 1984) and may be influenced by numerous factors such as motor slowing, cognitive difficulties or simply thoroughness. As such, it is important in clinical practice to carefully interpret Z scores calculated using copy time as a predictor of task performance, and use these measures in conjunction with clinical observations and other test results.

Another original aspect of the present study was the estimation of the association between copy score and subsequent recall (immediate and delayed), and the association between immediate and delayed recall performance. Results showed that higher copy score is predictive of better immediate recall performance which, in turn, leads to better delayed recall performance. In clinical practice, this can have implications for the accuracy of estimation of recall performances. For instance, taking copy score into account may improve clinicians' accuracy in estimating memory performance in patients with poor constructional abilities. Without including copy score as a predictor of performance, memory for the ROCF and/or the TCF could be incorrectly appraised. To illustrate this, let us imagine a 65-year-old man with 19 years of education, who took 300 s to copy the ROCF, scored 20 on the copy trial, obtained 10 on the immediate recall trial, and 10 on the delayed recall score. First, based on the regression equations for the copy, immediate and delayed recall scores from Table 3, the patient's Z scores are −3.16, −1.43, and −1.34, respectively. These results appear to be indicative of impaired constructional abilities and mild memory difficulties. However, based on the regression equations from Table 4, which considers all the variables (i.e., sociodemographic variables, copy time, copy score, and immediate recall score), the patient's Z score are now −3.04, 1.78 and 0.66, respectively. These results are still indicative of impaired constructional abilities. However, Z scores for the immediate and delayed recall trials now fall within normal limits. This example shows how important it is to take into account copy performance while interpreting subsequent ROCF memory performance. The previous example applies only to ROCF performance. To interpret TCF performance, one should refer to Tables 5 and 6. While Table 5 includes sociodemographic variables alone, Table 6 includes all variables (i.e., sociodemographic, copy time, copy score, and immediate recall score). To ease calculation of Z scores, an Excel spreadsheet containing automatic formulas has been prepared and is available by writing to the corresponding author of the manuscript. (Please contact the corresponding author for the Excel spreadsheet containing automatic formulas.)

Interestingly, results from this study have shown that immediate recall score was strongly related to delayed recall score, calling into question the usefulness of the delayed recall score. However, the present study was conducted in healthy older adults and the immediate and delayed recall scores may be more likely to differ in older adults with memory impairment, especially in those with weakened storage processes. Further studies are needed to verify this hypothesis.

Finally, the study used regression equations to calculate Z scores for performance on the ROCF and TCF. These equations differ from typical normative methods (e.g., percentiles or standard Z scores calculated from a mean and a standard deviation). According to Fastenau (1998), these regression-based norms can lead to higher rates of false negatives when used in older women and less educated people. However, this normative method has the advantage of better estimating the expected performance of a participant given his/her specific characteristics.

The main limitation of the present studies was the use of an incidental sampling method, resulting in an underrepresentation of people aged 50–59 and 80 years or older. Also, for the ROCF, women in all age ranges were overrepresented. Although a random sampling method would have been ideal, the sampling method used here is a practical starting point in the establishment of ROCF and TCF norms for the Quebec-French population which were nonexistent until now. To our knowledge, this study is the first to provide normative data for the ROCF and the TCF in a Quebec-French population, which is necessary to allow accurate interpretation of test results to distinguish between normal and pathological cognitive states. Also, it is the first to provide separate normative data for the standard procedure of the TCF and the ROCF. However, it is important to note that these normative data were developed with different samples of older adults. As such, one should be careful when comparing performance on both tests in a patient. Also, it should only be used for individuals who have been tested using the standard, incidental learning protocol, and results should be interpreted carefully with individuals within the youngest (50–59) and oldest (≥80 years) age clusters, for lower educated people as well as for men. Moreover, as these norms were developed in older Quebec-French individuals, results should be interpreted carefully in individuals from other groups (e.g., English-speaking older adults in Quebec, French-speaking older adults from other countries). The fact that no time limit was given to copy the ROCF or the TCF could also be considered as a limitation of the study. For instance, some participants took up to 8 min to copy the figures and this brings into question the possibility that some participants could be considered impaired. All participants were screened using brief cognitive tests (i.e., MMSE and MoCA) but the visuocontructional items of these instruments are easier than the ROCF or the TCF. Thus, one cannot exclude the possibility that the samples included a few participants with some degree of visuospatial impairment. Finally, although all researchers used the same task instructions and versions, and subjects from each site met the same inclusion/exclusion criteria, some variability may exist between the recruiting sites. Moreover, although all examiners were individually trained to administer these tests, no common training was conducted and no inter-rater reliability estimates were tested. As such, despite the standardized administration and scoring of the tests, some variability may exist between the examiners.

Funding

This work was funded by the Réseau québécois de recherche sur le vieillissement, the Canadian Institutes of Health Research and the Natural Sciences and Engineering Research Council of Canada. In addition, BC was supported by a doctoral research award from the Alzheimer Society of Canada. OP was supported by a Fellowship award from the Canadian Institutes of Health Research. MPT was supported by a scholarship from the Centre de recherche sur le cerveau, le comportement et la neuropsychiatrie. CH, JFG, and NC were supported by a salary award from the Fond de recherche du Québec – Santé. JFG is also supported by a New Investigator award from the Canadian Institutes of Health Research.

Conflict of Interest

None declared.

References

Arsenault-Lapierre
G.
Whitehead
V.
Belleville
S.
Massoud
F.
Bergman
H.
Chertkow
H.
Mild cognitive impairment subcategories depend on the source of norms
Journal of Clinical and Experimental Neuropsychology
 
2011
33
596
603
Bennett-Levy
J.
Determinants of performance on the Rey-Osterrieth complex figure test: An analysis, and a new technique for single-case assessment
British Journal of Clinical Psychology
 
1984
23
109
119
Berry
D. T.
Allen
R. S.
Schmitt
F. A.
Rey-Osterrieth Complex Figure: Psychometric characteristics in a geriatric sample
The Clinical Neuropsychologist
 
1991
5
143
153
Boone
K. B.
Lesser
I. M.
Hill-Gutierrez
E.
Berman
N. G.
D'elia
L. F.
Rey-Osterrieth complex figure performance in healthy, older adults: Relationship to age, education, sex, and IQ
The Clinical Neuropsychologist
 
1993
7
22
28
Caffarra
P.
Vezzadini
G.
Dieci
F.
Zonato
F.
Venneri
A.
Rey-Osterrieth complex figure: Normative values in an Italian population sample
Neurological Science
 
2002
22
443
447
Chiulli
S. J.
Haaland
K. Y.
LaRue
A.
Garry
P. J.
Impact of age on drawing the Rey-Osterrieth figure
Clinical Neuropsychologist
 
1995
9
219
224
Fastenau
P. S.
Validity of regression-based norms: An empirical test of the comprehensive norms with older adults
Journal of Clinical and Experimental Neuropsychology
 
1998
20
906
916
Fastenau
P. S.
Denburg
N. L.
Hufford
B. J.
Adult norms for the Rey-Osterrieth complex figure test and for supplemental recognition and matching trials from the extended complex figure test
The Clinical Neuropsychologist
 
1999
13
30
47
Folstein
M. F.
Folstein
S. E.
McHugh
P. R.
Mini-mental state: A practical method for grading the state of patients for the clinician
Journal of Psychiatric Research
 
1975
12
189
198
Gallagher
C.
Burke
T.
Age, gender and IQ effects on the Rey-Osterrieth complex figure test
British Journal of Clinical Psychology
 
2007
46
35
45
Haaland
K. Y.
Linn
R. T.
Hunt
W. C.
Goodwin
J. S.
A normative study of Russell's variant of the Wechsler memory scale in a healthy elderly population
Journal of Consulting and Clinical Psychology
 
1983
51
878
881
Hartman
M.
Potter
G.
Sources of age differences on the Rey-Osterrieth complex figure test
The Clinical Neuropsychologist
 
1998
12
513
524
Hubley
A. M.
Using the Rey-Osterrieth and modified Taylor complex figures with older adults: A preliminary examination of accuracy score comparability
Archives of Clinical Neuropsychology
 
2010
25
197
203
Institut de la statistique du Québec
Population de 25 ans et plus, selon le plus haut degré de scolarité atteint, le sexe et le groupe d’âge Québec, 2006
2006
Koss
E.
Haxby
J. V.
DeCarli
C.
Shapiro
M. B.
Friedland
R. P.
Rapoport
S. I.
Patterns of performance preservation and loss in healthy aging
Developmental Neuropsychology
 
1991
7
99
113
Lezak
M.
Neuropsychological assessment
 
1983
New-York
Oxford University Press
Machulda
M. M.
Ivnik
R. J.
Smith
G. E.
Ferman
T. J.
Boeve
B. F.
Knopman
D.
et al.  
Mayo's older Americans normative studies: Visual form discrimination and copy trial of the Rey–Osterrieth complex figure
Journal of Clinical and Experimental Neuropsychology
 
2007
29
377
384
Meyers
J. E.
Meyers
K. R.
The Meyers scoring system for the Rey complex figure and the recognition trial: Professional manual.
 
1995
Odessa, FL
Psychological Assessment Resources
Mitrushina
M.
Satz
P.
Chervinsky
A. B.
Efficiency of recall on the Rey-Osterrieth complex figure in normal aging
Brain Dysfunction
 
1990
3
148
150
Nasreddine
S. Z.
Phillips
N. A.
Bédirian
V.
Charbonneau
A.
Whitehead
V.
Collin
I.
et al.  
The Montreal cognitive assessment, MoCA: A brief screening tool for mild cognitive impairment
Journal of the American Geriatrics Society
 
2005
53
695
699
Newman
P. D.
Krikorian
R.
Encoding and complex figure recall
Journal of the International Neuropsychological Society
 
2001
7
728
733
Osterrieth
P. A.
Le test de copie d'une figure complexe: Contribution à l’étude de la perception et de la mémoire
Archives de Psychologie
 
1944
30
286
356
Peña-Casanova
J.
Gramunt-Fombuena
N.
Quiñones-Úbeda
S.
Sánchez-Benavides
G.
Aguilar
M.
Badenes
D.
et al.  
Spanish multicenter normative studies (NEURONORMA Project): Norms for the Rey–Osterrieth complex figure (copy and memory), and free and cued selective reminding test
Archives of Clinical Neuropsychology
 
2009
24
371
393
Pontón
M. O.
Satz
P.
Herrera
L.
Ortiz
F.
Urrutia
C. P.
Young
R.
et al.  
Normative data stratified by age and education for the Neuropsychological Screening Battery for Hispanics (NeSBHIS): Initial report
Journal of the International Neuropsychological Society
 
1996
2
2
96
104
Rey
A.
L'examen psychologique dans les cas d'encéphalopathie traumatique
Archives de Psychologie
 
1941
28
286
340
Rosselli
M.
Ardila
A.
Effects of age, education, and gender on the Rey-Osterrieth complex figure
The Clinical Neuropsychologist
 
1991
5
370
376
Rosselli
M.
Ardila
A.
The impact of culture and education on non-verbal neuropsychological measurements: A critical review
Brain and Cognition
 
2003
52
326
333
Strauss
E.
Sherman
E. M.
Spreen
O.
A compendium of neuropsychological tests: Administration, norms, and commentary
 
2006
3rd ed.
New York
Oxford University Press
Strauss
E.
Spreen
O.
A comparison of the Rey and Taylor figures
Archives of Clinical Neuropsychology
 
1990
5
417
420
Taylor
L. B.
Localization of cerebral lesions by psychological testing
Clinical Neurosurgery
 
1969
16
269
287
Tombaugh
T. N.
Faulkner
P.
Hubley
A. M.
Effects of age on the Rey-Osterrieth and Taylor complex figures: Test-retest data using an intentional learning paradigm
Journal of Clinical and Experimental Neuropsychology
 
1992
14
647
661
Tombaugh
T. N.
Faulkner
P.
Schmidt
J. P.
A new procedure for administering the Taylor complex figure: Normative data over a 60-year age span
The Clinical Neuropsychologist
 
1992
6
63
79
Verhaeghen
P.
Salthouse
T. A.
Meta-analyses of age-cognition relations in adulthood: Estimates of linear and non-linear age effects and structural models
Psychological Bulletin
 
1997
122
231
249
Vingerhoets
G.
Lannoo
E.
Wolters
M.
Comparing the Rey-Osterrieth and Taylor complex figures: Empirical data and meta-analysis
Psychologica belgica
 
1998
38
109
119
Yamashita
H.
Comparability of the Rey–Osterrieth complex figure, the Taylor complex figure, and the modified Taylor complex figure in a normal sample of Japanese speakers
Psychological Reports
 
2006
99
531
534
Yesavage
J. A.
Brink
T. L.
Rose
T. L.
Adey
M. B.
Crook
T.
Ferris
S.
Bartus
R.
The geriatric depression rating scale: Comparison with other self-report and psychiatric rating scales
Assessment in geriatric psychopharmacology
 
1983
New Canaan, CT
Mark Powley Associates
153
165