Disclaimer: This dissertation has been written by a student and is not an example of our professional work, which you can see examples of here.

Any opinions, findings, conclusions, or recommendations expressed in this dissertation are those of the authors and do not necessarily reflect the views of UKDiss.com.

The Impact of Dyslexia on Stroop Task Performance

Info: 10156 words (41 pages) Dissertation
Published: 25th Feb 2022

Reference this

Tagged: EducationPsychology


The Stroop task allows interference between colour naming and reading to be studied. The interference found in Stroop task has been considered as an indicator of reading automaticity even though poor readers, who have been found to lack automaticity, display strong interference. This study aimed to test whether dyslexia would impact performance on the Stroop task. In order to test this, participants were required to participate in a series of tests including the Stroop test, reading, spelling, and working memory tasks. A 2 X 2 within-subjects factorial design was used to test 78 participants. The children recruited for this experiment were aged between 10-16 and the adolescent and young adults were between the ages of 16-24. The younger subjects were from a local Academy whereas the older students were from various Universities. The results suggested dyslexic individuals performed slower than non-dyslexics. However, the differences between the groups were not statistically significant. To conclude, although the differences between the groups were not significantly different, collecting more data may result in a more definitive answer.


1. Introduction

1.1. Stroop task in children

1.2. Reading, spelling and working memory

1.3. Stroop task and dyslexia

1.4. Hypothesis

2.0. Methods

2.1. Ethical Approval

2.2. Participants

2.3. Materials

2.4. Design

2.5. Procedure

2.5.1. Spelling test

2.5.2. Reading test

2.5.3. Working memory test

3.0. Results

3.1. 2-way ANOVA

3.2. Regression for facilitation excluding dyslexics

3.3. Regression for facilitation for both dyslexics and non-dyslexics

3.4. Regression for interference excluding dyslexics

3.5. Regression for interference for both dyslexics and non-dyslexics

4.0. Discussion

4.1. Conclusion

5.0. References

6.0. Appendices

Appendix A

Appendix B

Appendix C

Appendix D

Appendix E

Appendix F

Appendix G 

1.0. Introduction

The Stroop task is a psychological test which has been extensively investigated since its discovery in the 1930s (Stroop, 1935). It involves a decision about one dimension of a multidimensional stimulus in which other dimensions may disagree or agree with the judgement dimension (Logan, 1980). The Stroop effect refers to our tendency to experience difficulty naming a physical colour when it is used to spell the name of a different colour (the word blue written in red ink, usually referred to as the incongruent condition) but not when we simply read out colour words (the word red written in red ink, known as congruent condition), or when simply reading out words in black ink or colours of rectangle patches (neutral condition). The interference effect occurs in naming the print colour of a word when the word itself is the name of another colour (difference between incongruent and neutral condition), whereas interference within the congruent condition itself is named the incongruity effect (Wright & Wanley, 2003) and facilitation effect refers to the difference between congruent and neutral condition (Chen et al, 2001). Stroop task is considered as an assessment of interference and processing speed (MacLeod, 1991) and has been changed in a variety of different ways to the original list-based task. Thus, a large variety of studies have started to use pictorial versus words stimuli, card versus computer presentations, visual versus auditory stimuli, or list versus single stimuli, and tasks have been changed according to the vocal versus manual response format (MacLeod, 1991; Wright, 2017). These modified versions have been used in a wide area of psychological research, such as phobias, anxiety disorders and depression (Price & Karl-Hanson, 2007). The common finding is that a longer reaction time is observed in the incongruent condition compared to the congruent or neutral condition (Goldfarb & Henik, 2007; Verbruggen, Liefooghe & Vandierendonck, 2004; West & Alain, 2000) and greater Stroop interference has been regarded as an index of lower interference control (Ikeda et al., 2011).

1.1. Stroop task in children and adults

Stroop task, mainly the incongruent condition, is most often described as a measurement of ability to inhibit an overlearned response in favour of an unusual one (Armengol, 2002; Homack & Riccio, 2004; Strauss, Sherman & Spreen, 2006; Wecker et al., 2000). Previous studies have revealed that younger children are more affected by Stroop interference compared to older children (Carter, Mintun & Cohen, 1995; Vurpillot & Ball, 1979; Peru, Faccidi & Tassinari, 2006). For example, Bub, Masson and Lalonde (2006) conducted a study with 65 children aged 7-11 years using the congruent, incongruent, neutral conditions, as well as reading and spelling tests. The results demonstrated that the younger children (9 and under) showed larger Stroop interference effect compared to the older children (1267 ms vs 1042 ms respectively) as well as slowing of word reading in incongruent condition (943 ms vs 861 ms respectively).

In a typical Stroop task, interference control has an inverted U-shaped curve with age, increasing as 3-7-year-olds learn to read, then gradually decreasing (Leon-Carrion, Garcia-Orza & Perez-Santamaria, 2004). The most popular developmental study has been conducted by Comalli, Wapner and Werner (1962) from a sample of 235 subjects ranging from 7-80 years on tasks which included an incongruent and neutral condition. The results revealed that inference was greatest for 7 year olds, decreases with increasing age up to 17-19 year-old, remains constant during middle years (25 to 45) and then decreases again in the older group (65-80). Furthermore, the results demonstrated that response time is least for reading colour words, longer for naming actual colours and longest when there is interference.

In terms of facilitation, it is evident that facilitation effect receives little investigation in comparison to the interference effect, (Wright & Wanley, 2003) as many developmental Stroop studies do not contain a congruent condition alongside neutral and incongruent conditions. Under single mechanism accounts, both facilitation and interference were assumed to be caused by same semantic process aiding versus delaying performance, respectively (Botvinick et al., 2001; MacLeod, 1991; Melara & Algom, 2003). However, MacLeod and MacDonald (2000) propose that single mechanism accounts are not sufficient to explain the findings on difference between interference and facilitation. This conclusion was reached on the basis that some tasks affect facilitation but not interference, while others affect interference without affecting facilitation (Wright, 2017). However, MacLeod and MacDonald (2000) later argued that interference and facilitation are not caused by the same process and rather proposed a theory suggesting that interference is the result of semantic conflict between word and colour, whereas facilitation is due to inadvertent word-reading.

1.2. Reading, spelling and working memory

Dyslexia can be defined as having difficulties with reading, writing, and spelling characterised by issues in literacy acquisition including reading speed, reading comprehension, and phonological decoding, ranging from mild to severe (Rayner et al., 2012; Shaywitz & Shaywitz, 2003; Vellutino et al., 2004). As dyslexic individuals have deficiencies in the skill to segment the written word into its underlying phonologic elements, they experience problems in decoding and identifying printed words (Høien & Sundberg, 2000; Joanisse et al., 2000; Rey et al., 2002). The way in which their brain codes phonology is less effective than that of normally developing individuals irrespective of their strengths in other cognitive abilities such as semantic processing (Lyon, Shaywitz & Shaywitz, 2003; Snowling, 2001). When dyslexic individuals are required to read rare or pseudo-words their accuracy is lower and their performance is slower compared to individuals without dyslexia as it takes dyslexics longer to decode words correctly due to their poor letter-sound knowledge (Griffiths & Snowling, 2002; Leinonen et al., 2001). These difficulties can be seen in dyslexic children from a very young age into adulthood (Gallagher, Frith & Snowling, 2000).

If dyslexics have serious phonological problems, these difficulties should also show themselves in spelling as well as reading. Dyslexics’ spelling problems are often more severe and persistent than their reading problems (Bourassa & Treiman, 2003; Caravolas & Volin, 2001). Studies on spelling compare the ability to segment, manipulate and identify phonemes in spoken words by children with dyslexia and age matched controls (Bernstein, 2009). Studies demonstrate that children with dyslexia perform worse than controls in the identification of spoken consonants (Breier et al., 2002) and tend to mistake some vowels for similarly articulated items when spelling (Bertucci et al., 2003). For example, Cassar et al., (2005) compared spelling abilities of 25 dyslexics aged 11, with 25 normally developing children aged 7-8, matched on their ability to spell. The children with dyslexia had difficulties with the same phonological structures that cause problems for 7 year-olds. Both groups had difficulties with consonant clusters, letter name spellings and reduced vowels in unstressed syllables.

Dyslexics have additional difficulties that are not restricted to reading and spelling. Deficiencies in short-term memory (STM) and working memory (WM) have been described as one of the crucial characteristics of dyslexia (Smith-spark & Fisk, 2007). WM is defined as a processing resource of limited capacity, involved in the preservation of information while processing the same or other information (Unsworth & Engle, 2007; Miyake & Shah, 1999). Researchers have suggested that there is a domain-general processing deficit in children and adults with dyslexia, with both verbal and visual WM being affected as additional demands are placed on the central executive (Cohen-Mimran & Sapir, 2007; Swanson & Howell, 2001; Swanson, Zheng & Jerman, 2009). For example, Smith-Spark et al., (2003) conducted a study with two groups of University students using the digit span and word span memory test. The results revealed significant differences between dyslexics and non-dyslexics on both the tasks, with dyslexics performing worse than the control group, especially as the numbers or words presented increased from 8 items to 12 items; thereby providing evidence for continuing dyslexic impairments of the WM into adulthood. The presence of deficits in the CE of WM in individuals with dyslexia is consistent with other deficits in the executive functions, such as selective and sustained attention, inhibition of routine responses, and inhibition of distracters (Brosnan et al., 2002).

1.3. Stroop task and dyslexia

The classic Stroop (1935) model is a typical conflictual situation, in which the physical appearance of alphabetic symbols is at odds with their meaning (incongruent condition). The notion of automaticity has been central for understanding and explaining the Stroop effect, since it is generally considered obligatory to read the word but not to name the colour (Protopapas et al., 2014). The concept of automatization refers to a gradual reduction in the need for conscious control as a new skill is learned. This leads to greater speed and efficiency and a decreased likelihood of breakdown of performance under stress, as well as the ability to perform a second task at little or no cost (Kapoula et al., 2010). It is suggested that reading is done for meaning and highly overlearned from 6 or 7 years of age (Braet et al., 2011; Seymour, Aro & Erskine, 2003). On this note, one would expect poor and less skilled readers to exhibit less interference than good readers as reading is less automatic, therefore they are more likely to skip the word and focus on the colour alone (Faccioli et al., 2008). Nevertheless, this prediction stands in contrast to empirically observed data. Children with dyslexia appear to be slower in reaching reading automation and full attentional-executive control (Facoetti et al., 2003). However, even if difficult and slow, reading is unavoidable also for these children. If they suffer from a reading automaticity problem, this should originate beyond the level where implicit reading determines Stroop interference (Everatt et al., 1997). The interference in turn would be larger than in normal readers due to a less effective control upon a reduced reading automaticity. This is in agreement with the model proposed by Coltheart et al., (1999) that predicts an increase in the size of Stroop effects whenever word processing is slowed.

A number of studies showed that poor readers produce more interference (Everatt et al., 1997; Helland & Asbjørnsen, 2000; Mano et al., 2016). Protopas, Archonti and Skaloumbakas (2007) revealed that reading ability is negatively related to Stroop interference. In their first study, they compared children with dyslexia (mean age 12.5) to age matched controls and reported greater interference for children with dyslexia. In a second study they examined the relationship between interference and reading skills in the general school population and found that poorer reading skills were associated with larger interference. Faccioli et al., (2008) confirms more interference in dyslexic children (7-11) relative to control of similar age. They demonstrated that children with dyslexia and normal readers performed very well on rectangles and congruent words (mean overall accuracy: 97.5% children with dyslexia, 99.6% normal readers), but they made several errors on incongruent words (mean accuracy: 84.9% children with dyslexia and 85.8% normal readers). With respect to speed of response, children with dyslexia were slower than normal readers (mean overall: 941.7 ms vs. 738.8 ms). In terms of facilitation, like normal readers, children with dyslexia were slower in responding to congruent names than rectangles, thus showing no facilitation.

1.4. Hypothesis

The current study aimed to develop a better understanding of the impact of dyslexia on individuals’ performance on the Stroop task on a variety of age groups. Our first hypothesis (H1) predicts that dyslexics will perform worse on all cognitive tasks (reading, spelling and working memory) than non-dyslexics. The second hypothesis (H2) predicts that adults will have a faster reaction time on all five of the Stroop conditions in comparison to children. Thirdly, (H3) anticipates a faster reaction time and better performance on Stroop task in non-dyslexics compared to dyslexic individuals. Finally, (H4) predicts that facilitation and interference would be predicted by different variables for dyslexic and non-dyslexic participants.

2.0. Methods

2.1. Ethical Approval

This study received ethical approval from the Department of Life Sciences Research Ethics board at Brunel University on 12/12/2017 (Appendix A).

2.2. Participants

A total of 78 participants were recruited for this experiment. Of these, 68 (87.18%) were children between 10 and 16 years of age (M= 13.07, SD= 1.20) and 10 (12.82%) were adults (M= 20.55, SD= 2.48). None of the participants reported any problems related to anomalous colour perception, and had normal or corrected-to-normal visual acuity. The sample of participants recruited for this project were mixed; some participants were students from a local school and the rest were volunteers from Universities.

2.3. Materials

The participants participated in a total of four tasks which included the Stroop task, reading, spelling, and working memory tests. The Stroop task consisted of 5 blocks with 1-dimensional blocks that did not present Stroop stimuli, and three mixed blocks each containing 48 words. Blocks one and five were both 1-dimensional conditions (colour and word have nothing to do with each other) however both had different randomised order. The words ‘car’, ‘plug’, ‘jigsaw’, ‘sheep’ and the colours ‘red’, ‘green’, ‘yellow’, and ‘blue’ were all written in black ink, or the colours were displayed in a rectangle (following Wright, 2017). The three other blocks were both a mixture of congruent (colour and word agree) and incongruent conditions (colour and word disagree). In these trials, participants were required to ignore the written word and read out the colour as fast as possible.  For both the congruent and incongruent conditions, the same colours were used. In addition, the task consisted of one dimensional trials in order to measure the speed of colour naming in isolation, word reading in isolation, and word reading of colour in isolation. Word reading was either neutral words (one third), colour words (one third) or colour patches (one third).

For both reading and spelling, Wechsler Objective Reading Dimension tests were used. There was a total of (55) words for reading and (50) for spelling and these words ranged from easy words to more difficult words. The working memory test contained 14 trials of which two were practice trials. Of these 12 trials, four trials had three numbers to memorise, four trials had four numbers and four trials had five numbers.

Both the pre-tests and actual experiment were conducted using a Toshiba laptop. All the stimuli were presented on a low intensity white background to eliminate any biases and reduce fatigue. Colour patches were rectangular in shape, with on-screen dimensions of 2.2 cm high and 3.5 cm wide. The words were written in a font equivalent to Time New Roman with the participants sat approximately 60 cm away from the screen.

2.4. Design

This study employed a 2 X 2 within-subjects factorial design. Participants performed the same computer task and participated in the same conditions. The first factor was the condition (i.e. three levels corresponding to the incongruent, neutral and congruent conditions) while the second factor was whether the participants had dyslexia or not. Measures of interference and facilitation were calculated from the condition times.

2.5. Procedure

In advance to starting the actual experiment, consent was obtained from the school (Appendix B) and parents (Appendix C) for those participants in school. Participants in the school were separated on the basis of those in nurture class and average performing children. Upon arrival participants were presented with an information sheet (Appendix D) and a consent form (Appendix E). For most of the individuals in the nurture group, the information sheet had to be read verbally to them. Before starting the actual experiment, participants were required to take part in a spelling, reading and working memory test. This was to distinguish participants with dyslexia from those without. Reaction times were collected using a headset microphone. Error rates were noted in a hard copy format by sitting behind the participants during the experiment. All participants gave written informed consent before starting the experiment and were debriefed after the experiment (Appendix F).

The experiment took place in a cubicle located at the local school or the library to ensure a quiet and confidential environment. Participants were given verbal instructions regarding which responses to make and how to make them. Each participant was reminded that their verbal response was being recorded, therefore they had to speak in a loud and clear manner. After each verbal response, the next would appear, so participants were instructed to not make any noise other than their response to the stimuli.

The following tests were used to distinguish participants with dyslexia from participants without dyslexia. Wechsler Objective Reading Dimensions are individually administered tests designed for the assessment of children aged from 6 to 16 years.

2.5.1. Spelling test

The spelling test was used to determine the literacy ability of each participant. For this test, the participant was sat away from the laptop, therefore they could not see the screen. The researcher was placed in front of the laptop, facing the screen. The test required the participants to spell out the last word read out by the researcher. For example, the researcher would say ‘Cat. Anne’s cat had kittens. Cat’ and participants would have to spell out the word ‘cat’. The words ranged in difficulty, starting with easy words such as ‘cat’ or ‘no’ to moderate words such as ‘apparently’ and  ‘assistants’ to more challenging words such as ‘pharmaceutical’ and ‘conscience’. Reaction times were recorded as soon as the researcher read out the last word. As soon as a correct response was given by the participant, the researcher pressed 6 on the keyboard however, if the participant gave an incorrect answer the researcher pressed 4 on the keyboard. If none was given, the researcher pressed 4 on the keyboard to record an incorrect response and to stop the timer.

2.5.2. Reading test

For this test, participants had to sit in front of the laptop, facing the screen, while the researcher was seated next to the participant with an external keyboard. In this test, participants had to read out the words on the screen. These words ranged from easy words such as ‘the’ and ‘up’ to more moderate words such as ‘accordion’ and ‘ridicule’ to more difficult words such as ‘euphemism’ and ‘hierarchical’. The researcher recorded reaction times by pressing 0 to bring up the word, followed by pressing 4 if the participant read the word incorrectly or 6 if they read it correctly.  The researcher had to wait for the participant to completely say the word as they needed to factor in incorrect pronunciations.

2.5.3. Working memory test

The working memory test assessed numerical working memory. In this test, participants were informed that the experimenter would read out a group of numbers and they would have to remember the biggest, or smallest digit from that list. The decision of whether the participant recalled the biggest or the smallest number was pre-set on the laptop. In the first few trials, the task comprised of three numbers which increased to four, and later five and participants had to remember each of the numbers to figure out the smallest or largest from the group. Reaction times were recorded by the experimenter. As soon as the researcher said smallest or biggest, reaction times were recorded by pressing 0 on the keyboard. After the participants gave a response, incorrect responses were recorded by pressing 4, and correct responses were recorded by pressing 6. The reaction time records the time it takes between the researcher pressing 0 to pressing 4 or 6 depending on the response given by the participant.

3.0. Results

To test the basic features of the data in this study, descriptive statistics were run. The descriptive statistics were run on the analyses of cognitive ability (reading, writing, and working memory); analyses of children versus adults’ performance on the Stroop task; and finally analyses of dyslexia on Stroop task performance.

Participants were identified as dyslexic or non-dyslexic based on spelling and reading scores. Individuals who performed greater than 2sd below the average performance (for children) on both spelling and reading tests were identified as dyslexic. It is important to acknowledge that this is not a formal diagnoses of dyslexia, participants are identified as likely dyslexics due to performing extremely low on spelling and reading tasks. This gave a total of 12 participants identified as dyslexic. However, as three participants did not provide either complete reading or complete spelling scores they were eliminated automatically by SPSS for cases where they had relevant data that was blank, which left us with a total of as few as nine dyslexics, depending on the analysis that was run.

Table 1: Summary of Percentages of Cognitive Performance According to Dyslexia Status

  Dyslexia Status  
  Non-Dyslexic Dyslexic Both Groups
Reading 78% (1) 39% (4) 58% (2)
Spelling 69% (1) 28% (3) 48% (2)
WM 81% (2) 51% (5) 66% (2)
Overall Cognitive 76% (1) 39% (3) 57% (2)

Note: Figures in parentheses are standard errors.

3.1. 2-Way ANOVA

As depicted in Table 1, the overall performance on reading, spelling and working memory was 37% better in non-dyslexics than dyslexics. In terms of spelling there was a massive difference on the performance of dyslexics compared to non-dyslexics, with non-dyslexics performing 41% better than dyslexics. Similarly, non-dyslexics performance on reading task was 39% better than dyslexics. The smallest difference was on working memory task, with non-dyslexics performing 30% better than dyslexics. Nevertheless, it is important to recognise the fact that reading was out of 55 and spelling was out of 50 whereas working memory was out of 12 therefore, it is better to compare them as percentages as these can be more meaningfully compared to each other in analyses than raw scores that have different maximums. See Appendix G for Table 2 redrawn with raw scores rather than percentages.

The tendency for the non-dyslexic group to do better than the dyslexic group on reading, spelling and working has been confirmed as statistically significant (F(1,73) = 57.398, p = 0.001, partial eta squared = 0.440, observed power = 1.000)  . This confirms that dyslexic participants, as identified by the basis of standard deviations, did much worse than non-dyslexics. Overall, the difference between domain of cognition were statistically significant (F (2,146) = 363.199, p= 0.001, partial eta squared = 0.833, observed power =1.000). This demonstrates that performance on working memory task was highest, followed by reading task, and followed by spelling being lowest.  Finally, a significant interaction was observed between dyslexia and domain of cognition (F(2,146) = 39.252, p=0.001, partial eta squared = 0.350, observed power = 1.000).

Table 3: Summary of Stroop Performance According to Age Group.
  Children (under 16) Adults (over 16) Both Groups
Congruent 843 (18) 789 (47) 816 (25)
Neutral 894 (17) 800 (45) 847 (24)
Incongruent 973 (23) 911 (62) 942 (33)
Overall Performance 903 (18) 833 (48) 868 (26)

Note: Number of participants = 68 children and 10 adults. Figures in parentheses are standard errors.

The difference between children and adults for the congruent condition was 54ms, with the children responding slower than adults. For the neutral condition, the difference between children and adults was 94ms, with children responding slower than adults. Finally, the children responded 62ms slower than adults in incongruent condition. For both groups, the responses slowed down from neutral to congruent condition by 31ms, and from incongruent to neutral condition by 95ms. The conditions showed the typical Stroop profile, with the congruent condition faster than the neutral condition, and the incongruent condition slowest. The overall performance difference between children and adults was 70ms, with children performing slower than adults. For children the difference between congruent to neutral was 51ms, with responses being faster in the congruent condition whereas the difference for adults was 11ms; thus, showing children had higher facilitation compared to adults. The difference between neutral to incongruent was 79ms for children and 111ms for adults thus, demonstrating that interference was higher for adults. Nevertheless, the differences between interference are more similar in both groups compared to facilitation.

Overall difference between children and adults on the Stroop task was not statistically significant (F(1,76) = 1.794, p= 0.184, partial eta squared = 0.023, observed power = 0.262. This confirms that age does not impact the performance on the Stroop task thus, we combined the Stroop data for adults and children and analysed all 78 participants in all further analysis. Overall, the difference between Stroop condition was statistically significant (F(2,152) =  27.002, p= 0.001, partial eta squared = 0.262, observed power =1.000). This demonstrates that performance on congruent condition was fastest, followed by the neutral condition, with the incongruent being slowest. The interaction between Stroop condition and age was not statistically significant (F(2,152) = 0.687, p= 0.505, partial eta squared = 0.009, observed power = 0.164).

Table 4: Summary of Stroop Performance According to Dyslexia Status
  Dyslexia status  
  Non-dyslexics Dyslexics Both Groups
Congruent 835 (18) 841 (43) 838 (23)
Neutral 873 (17) 930 (41) 901 (22)
Incongruent 955 (24) 1017 (56) 986 (30)
Overall Performance 888 (19) 929 (44) 908 (24)

Note: Number of participants = 66 non-dyslexics and 9 dyslexics. Figures in parentheses are standard errors.

As seen in Table 4, dyslexics tended to respond slower than non-dyslexics on the Stroop task, by 41ms. For non-dyslexics, the difference between neutral and congruent condition was 38ms, however for dyslexic participants this was 89ms thus, demonstrating facilitation is higher in dyslexics compared to non-dyslexics. The difference between incongruent and neutral was 82ms for non-dyslexics and 87ms for dyslexics, thereby demonstrating that interference effect looks smaller compared to facilitation. The difference for congruent condition between dyslexic and non-dyslexics was 6ms, whereas it was 57ms for neutral condition and finally, 62ms for incongruent condition, with dyslexics responding slower than non-dyslexics in all three conditions.

Overall difference between dyslexics and non-dyslexics on Stroop performance was not statistically significant (F(1,76) = 0.731, p=0.395, partial eta squared = 0.010, observed power = 0.135. This demonstrates that even though dyslexics performed slower than non-dyslexics on all three of the condition, this difference was not effective enough to be significant. The overall difference on the Stroop task was statistically significant (F(2,152) = 40.874, p=0.001, partial eta squared = 0.350, observed power = 1.000. There was no significant interaction between dyslexia and the Stroop task (F(2,152) = 1.778, p = 0.172, partial eta squared = 0.023, observed power = 0.368. This reveals that although the two groups performed closely on the congruent condition and further apart for neutral and slightly further apart for incongruent condition, the overall tendency to be further apart as the condition got harder was not statistically significant.

Table 5: Summary of regression for facilitation for non-dyslexics
Variable Name Beta Unstandardized Beta Standardised Partial Correlation P Level
Model 1
Ages -2.923 -0.132 -0.127 0.245
Errors1 d av -7.055 -0.045 -0.043 0.753
Errors2 d av 4.833 0.158 0.140 0.299
Spell50full -3.660 -0.323 -0.142 0.294
Wm all 12 score -3.101 -0.084 -0.068 0.615
Read55score 3.825 0.404 0.199 0.137
Gender1f0m 50.582 0.369 0.366 0.005
Item Removed at Step 2, Errors1 d av 3, Wm all 12 score 4, Ages 5, Errors2 d av
Model 6
Spell50full -5.597 -0.494 -0.242 0.060
Read55score 4.241 0.448 0.221 0.086
Gender1f0m 44.601 0.325 0.329 0.010

3.2. Regression for facilitation excluding dyslexics

Our next hypothesis was assessed using regression analyses, two analyses were conducted for facilitation and another two analyses for interference. For facilitation, the first analysis was a linear regression run using the backward stepping method. This would allow the 7 predictor variables (Errors1 d av, Wm all 12 score, ages, Errors2 dAV, Spell50full, Read55score, Gender1f0m) to be reduced to only the few variables that work best together to predict Stroop RT performance. The first regression was based on only the participants who were not classified as dyslexics earlier in our first analysis (N = 66).

As presented in Table 5, the top of the table shows the initial model with all predictors entered simultaneously, with the middle of the table illustrating the variables that were excluded at each step. Finally, the bottom of the table displays the variables that survived to remain in the final model. In total, SPSS produced 6 models for this analysis.

The R value for the model was R = 0.377 and this model was statistically significant (F(3,59) = 3.265, p = 0.028). The R squared value for this model was R squared = 0.142, indicating that the model accounts for 14.2% of the variability of the facilitation RTs about the regression line.

Overall, there were 3 variables remaining in the final model. These variables were spelling score (Spell50full), word reading score (Read55score) and also gender (Gender1f0m).

The largest standardised beta was for spelling score however this coefficient was negative. This demonstrates that a lower spelling score predicted a higher amount of Stroop facilitation for non-dyslexic participants.

Read RT had the second largest coefficient but this was positive. This reveals that a higher reading score predicted a greater amount of Stroop facilitation. This further suggests that for normal readers, the better they are at reading, the bigger the difference between the congruent and incongruent condition (i.e., facilitation).

Finally, the last variable retained in the model was gender with a standardised beta coefficient that was positive. As males were coded 0 and females were coded 1, a positive beta illustrates that females tended to exhibit greater Stroop Facilitation.

Having done a regression for facilitation with only the non-dyslexic participants, it was intended to now do the same analysis for dyslexic participants. However, because there were at most nine dyslexic participants in the sample, and only seven of these had enough data to be included in the regressions, we could not run this analysis. An alternative to this was to combine dyslexic participants’ data with previous sample (non-dyslexics) and run a new regression. Although this would not allow us to assess the predictors of dyslexia very directly, it would allow us to look at these predictors indirectly. This is because adding the dyslexic participants to the non-dyslexic participants would allow us to see how this changed which predictors now were significant in a final model, as well as how many models there now were plus what the model R now became (e.g., is it higher or lower than before).

Table 6: Summary of facilitation for both groups combined
Variable Name Beta Unstandardized Beta Standardised Partial Correlation P Level
Model 1
Ages -3.328 -0.132 -0.135 0.288
Errors1 d av -7.310 -0.044 -0.042 0.739
Errors2 d av -4.110 -0.174 -0.150 0.236
Spell50full -4.854 -0.482 -0.207 0.100
Wm all 12 score -7.616 -0.193 -0.165 0.193
Read55score 2.975 0.337 0.167 0.187
Gender1f0m 57.623 0.380 0.389 0.001
Item Removed at Step 2, Errors1 d av 3, Ages 4, Wm all 12 score 5, Errors2 d av
  6, Read55score      
Model 6
Spell50full -2.443 -0.243 -0.255 0.035
Gender1f0m 52.352 0.345 0.351 0.003

3.3. Regression for Stroop facilitation for both dyslexics and non-dyslexics

The second regression models used facilitation RT as the independent variable. This analysis was based on both the dyslexic and non-dyslexic participants (N=70). SPSS produced a total of 6 models for this analysis. As displayed in Table 6, there were 2 variables that remained in the final model; these were gender (Gender1f0m) and spelling score (Spell50full).

The R value for this model was R = 0.402 and this model was statistically significant (F(2,67) = 6.444, p = 0.003). The R squared value for the model was R squared = 0.161, illustrating that the model accounts for 16.1% of the variability of facilitation RTs.

The Beta coefficient for this model are summarized in Table 6. The largest standardised beta was for gender and this was positive thus, implying that females tended to have higher facilitation than males.

The second variable retained in the model was spelling score, however this had a negative coefficient which was the same for non-dyslexic participants. This demonstrates that whether or not we assess non-dyslexics on their own, or assess dyslexic and non-dyslexic participants combined, spelling features an important predictor of the facilitation RT effect.

Table 7: Summary of regression for interference for non-dyslexics
Variable Name Beta Unstandardized Beta Standardised Partial Correlation P Level
Model 1
Ages 0.253 0.009 -0.010 0.942
Errors1 d av 46.574 0.241 0.242 0.070
Errors2 d av -19.213 -0.510 -0.444 0.001
Spell50full 3.749 0.270 -0.128 0.342
Wm all 12 score 4.698 0.103 0.091 0.501
Read55score -5.584 -0.481 -0.254 0.057
Gender1f0m -24.488 -0.146 0.166 0.217
Item Removed at Step 2, Ages 3, Wm all 12 score 4, Gender1f0m 5, Spell50full
Model 5
Errors1 d av 47.335 0.245 0.262 0.041
Errors2 d av -21.024 -0.559 -0.508 0.000
Read55score -2.485 -0.214 -0.228 0.077

3.4. Regression for interference excluding dyslexics

It is also important to assess interference in the same way as facilitation. The regression for interference for just the non-dyslexic participants on their own had a model R vale of R = 0.518 and this model was significant (F(3,59) = 7.217, p < 0.001). The R squared value was 0.268, indicating that the model accounted for 26.8% of variance of the interference data. This is approximately 10% higher than in either of the facilitation models above.

The Beta values for this model are presented in Table 7. Overall, SPSS produced 5 models, however only 3 variables remained in the final model. These were the number of errors made in total on the 1-dimnesional stimuli (Errors1 d av), the errors made on the 2-dimensional Stroop condition (Errors2 d av), and finally the word reading score (Read55score).

The largest standardised beta was for errors made on the 2-dimensional Stroop condition (Errors2 d av) however the coefficient was negative. This demonstrates that the less errors made on the 2-deminesiona condition predicts a higher interference for non-dyslexic participants.

Total errors made in 1-dimensional stimuli (Errors 1 d av) had the second largest positive coefficient. This implies that the more errors participants made, the more interference they had.

Finally, the last variable maintained in the model was word reading score, however the coefficient was negative.  This reveals that lower scores on reading predicted higher interference.

Table 8: Summary of Interference for both groups combined
Variable Name Beta Unstandardized Beta Standardised Partial Correlation P Level
Model 1
Ages -0.564 -0.017 -0.017 0.894
Errors1 d av 45.855 0.211 0.192 0.128
Errors2 d av -12.476 -0.405 -0.321 0.010
Spell50full -1.415 -0.108 -0.045 0.721
Wm all 12 score 3.418 0.067 0.055 0.665
Read55score -2.289 -0.199 -0.096 0.452
Gender1f0m 8.992 0.046 0.048 0.704
Item Removed at Step 2, Ages 3, Gender1f0m 4, Spell50full 5, Wm all 12 score
Model 5
Errors1 d av 48.715 0.225 0.214 0.080
Errors2 d av -12.636 -0.410 -0.359 0.003
Read55score -3.002 -0.261 -0.261 0.031

3.5. Regression for interference for both dyslexics and non-dyslexics

In order to gain an understanding of how interference is for dyslexic participants, we combined dyslexic participants’ data to the non-dyslexic participants as we did for facilitation earlier. This final model is summarized in Table 8. The model R value of R = 0.394 and this model was significant (F(3,66) = 4.050, p = 0.011). The R squared value was 0.115, showing that the model only accounted for 15.5% of the variability of interference, which is much lower than the previous model of interference which did not contain the dyslexic participants.

As depicted in Table 8, for the analysis with all the data combined there were 5 models and only 3 remained in the final model. These variables were the same 3 variables in the previous analysis of non-dyslexic participants alone (i.e., Errors1 d av, Errors2 d av , and Read55score).

The largest standardised beta was for 2-dimensional Stroop condition, which had a negative coefficient. This means that less errors on the 2-dimensional Stroop condition predicts higher interference regardless of whether they are dyslexic or not dyslexic.

Errors made in 1-dimensional stimuli had the second largest with a positive coefficient. This means that for both dyslexic and non-dyslexic, more errors results in higher interference.

Finally, the last variable retained in the model was word reading score with a standardised beta coefficient that was negative. For all the participants, lower scores on reading meant a higher interference.

4.0. Discussion

The aim of the present study was to investigate the impact of dyslexia on the performance on Stroop task in individuals aged between 10 and 23. Findings for this study have provided partial support for the proposed hypotheses. The first hypothesis (H1) was fully supported by the findings, as the non-dyslexic group outperformed the dyslexic group on all cognitive tasks. With regards to (H2), although children gave slower responses on all of the Stroop conditions, this was not statistically significant. For (H3), although responses were fastest in the congruent condition, followed by neutral and slowest in the incongruent condition, and these responses were made faster by the non-dyslexic group, the overall difference on performance between the two groups were not statistically significant. For our final hypothesis (H4), facilitation was predicted by spelling score, word reading score and gender for typically developing participants. On the other hand, for dyslexic participants, facilitation was predicted by spelling score and gender. In terms of interference, the predictors remained the same for both typically developing participants and those with dyslexia, these were errors made in total in 1-dimensional stimuli, errors made on 2-dimensional Stroop condition and word reading.

The results gathered from this study are in support of previous research conducted on the impact of dyslexia on reading, spelling and working memory. As discussed previously, dyslexics have issues with reading and spelling due to the ineffective coding of phonology compared to typically developing individuals (Snowling, 2001) as well as problems with working memory due to inability to retain information actively in mind (Swanson, Zheng & Jerman, 2009). Data from studies of children with dyslexia who have been followed prospectively support the concept that in adults, difficulties with reading fluency and spelling continue (Shaywitz & Shaywitz, 2005). For example, Lindgren and Laine’s (2011) study on University students, who have been identified as dyslexics from a young age demonstrated that dyslexic impairments were most visible in word and sentence segmentation, correctness in oral text reading and phoneme-to-grapheme awareness. These studies demonstrate that symptoms of dyslexia are persistent throughout the lifespan.

Performance on Stroop task has been tested extensively in the field of cognitive psychology. Nearly all existing models propose that interference results from competition between colour and word information and the need to suppress word information (West & Alain, 2000). A study by Wright and Wanley (2003) reveals that RT for interference looks similar in both adults and children whereas facilitation was greater in children than in adults. This implies that facilitation arises from a different system to interference. Following MacLeod and MacDonald’s (2000) inadvertent word-reading hypothesis, it can be predicted that children are more vulnerable to inadvertent word reading than adults (Ikeda et al., 2011; Imbrosciano & Berlach, 2005). One possibility that arises from these developmental differences is that younger children may be less able to suppress irrelevant stimulus dimensions and therefore may experience more difficulty than adolescents or adults, and that this skill may have a developmental trajectory during childhood (Bub, Masson & Lalonde, 2006). As stated previously, in a typical Stroop task, the performance has an inverted U-shaped pattern with age, the same can also be seen for inhibitory control which may explain the differences in performance according to age. Studies using other suppressions models, such as the stop-signal and negative priming provide evidence of significant improvements in the ability to inhibit a proponent course of action through childhood, but little change throughout adulthood (Bedard et al., 2002). For example, Williams et al., (1999) found that on average, older children (9-12 years) were 50ms faster in stopping their proponent responses than younger children (6-8 years), and younger adults (18-29 years) were 20ms faster than older adults (60-82 years).

Furthermore, children face difficulties guiding their actions by rules held in mind which conflict their learnings (Diamond, Kirkham & Amos, 2002). Consequently, the greater facilitation effect may be because children, especially younger ones, are less able to repeatedly apply the task set for colour naming, rather than that they are less able to inhibit incompatible word responses (Proulx & Elmasry, 2015).

Increased Stroop effect among dyslexic individuals may be due to poor cognitive control resulting in difficulty stopping the need to read the word instead of stating the colour.  Many researchers attribute Stroop interference found in dyslexic groups to impaired executive functions (Altemeier, Abbott & Berninger, 2008; Varvara et al., 2014). Executive functions refer to a collection of cognitive abilities such as mentally playing with ideas, thinking before acting, resisting temptations (self-control), and interference control (selective attention and cognitive inhibition) (Diamond, 2013). Dyslexic and normal readers both have difficulty stopping word processing, however, automatic readers can control their proponent response much better than dyslexic individuals (Helland & Asbjørnsen, 2000). For example, Everatt et al., (1997) noted that dyslexics are incapable of stopping word processing prior to the point of interference.

There are several methodological issues which may limit the significance of our study results. First, the sample size of the groups compared was not equal. There was a high number of children (N = 68) compared to adults (N = 10) and higher amount of non-dyslexic participants (N = 66) compared to dyslexic participants (N = 9). Sample sizes are important as small samples can undermine the internal and external validity of the study and very large samples can transform small differences into statistically significant differences (Faber & Fonseca, 2014). Therefore, future research in this area should ensure an equal or similar sample size for each group.

Furthermore, the current study used a vocal response for Stroop task. The headset used in this experiment picked up any noise made around instantly and the Stroop task moved on. Therefore, with some participants, as there was noise in the background, the Stroop moved on before the participant was actually able to answer which was then recorded as an incorrect answer. In order to make sure that the Stroop task is only moving on because of the answers provided by participants, future research can use a Stroop task which relies on manual mode of response where the participant needs to signal the correct answer by pressing a predefined key. This will confirm that the differences recorded in errors are actually due to the answer provided by the participant and not to other, external factors.

The results from this study can be used in future research to advance our understanding of working memory capacity and attention as it demonstrates that such cognitive abilities can be assessed without such complex tasks. The Stroop task has several advantages as part of a larger test battery. Generally, the method employed to examine the nature of working memory, reading and spelling has involved assessing performance on a variety of complex skilled tasks, including language comprehension, complex learning, and reasoning (Long & Prat, 2002). In the case of children who are not really good at reading or spelling, when they are being tested using such complex tasks they may feel anxious and perform worse. The Stroop task gets around the potential anxiety children feel when asked to read as it does not necessarily require reading and does not test intelligence; hence it will not impact how children perform. However, in order to get a better understanding of how reading changes with age or to see how individuals with learning disabilities cope with their disability by tracking students over time (Golden & Golden, 2002). The Stroop task can also be used to advance our understanding of changes in executive function, attention, concentration effectiveness and gender differences with age and education level (Penner et al., 2012). For example, this has already been done by Van der Elst et al., (2006) to assess the changes of executive functions by age and level of education and the results revealed that executive function, as measured by Stroop test, declines with age and that the decline is more pronounced in people with a low level of education.

4.1. Conclusion

Previous research has found that individuals with dyslexia perform worse on the Stoop task. This study aimed to enlighten the current research regarding the effects of dyslexia on Stroop performance. Contradictory to previous research, the results from this study showed that dyslexia does not seem to have an effect on an individual’s ability to perform well on a Stroop task. Although the results did not vary significantly, the results suggested that dyslexic participants performed slower on all Stoop conditions compared to non-dyslexic participants. This is in line with previous research which has suggested that the inability of dyslexic participants to perform as well as non-dyslexic participants is due to deficiencies in cognitive controls in dyslexics. However, what was significant was that individuals with dyslexia did perform worse on cognitive measures including reading, spelling, and working memory tasks. The findings from this study can be used in future research to test the effects of age on dyslexia by employing a longitudinal study. This can measure the same participants across different milestones and test if increased age impacts performance on the Stroop task.

Cite This Work

To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Related Services

View all

Related Content

All Tags

Content relating to: "Psychology"

Psychology is the study of human behaviour and the mind, taking into account external factors, experiences, social influences and other factors. Psychologists set out to understand the mind of humans, exploring how different factors can contribute to behaviour, thoughts, and feelings.

Related Articles

DMCA / Removal Request

If you are the original writer of this dissertation and no longer wish to have your work published on the UKDiss.com website then please: