Stroop Effect

​

What is Stroop Effect?

Stroop Effect is the extra time taken to name the ink of a color word when the word names a different color, against the time to name those colors as patches.

if the word ‘red’ is printed in blue ink how will the interference of the ink-color ‘blue’ upon reading the printed word ‘red’ compare with the interference of the printed word ‘red’ upon calling the name of the ink-color ‘blue?’

That question is on page 647, after the words “In other words.” The 1935 paper does not print the two-word name Stroop Effect. Its title is “Studies of interference in serial verbal reactions.” This page keeps Stroop Effect for the ink-naming increase. It does not rotate his phrase “interference of conflicting word stimuli upon naming colors” into the page name after this section. Colin M. MacLeod’s review uses Stroop effect in the title. That review was not re-read past its abstract. No later millisecond is copied from it. The counted increase is narrow. A person speaks the ink, line by line, and must not speak the word. The word is one of five color names, and it is never printed in the color it names. The comparison sheet has the same colors as solid squares. A faster feeling is not the score. Processing Fluency is a feeling of ease used as a cue about the thing itself. A hard font changed a minute estimate on that page. It did not ask anyone to name an ink. Priming Effect is a faster correct decision that two letter strings are words, when the words are associates. Naming an ink is not that decision.

​

Stroop Effect in 3 Depths

• Beginner: On the naming sheets, 100 students took a mean of 63.3 seconds to name 100 colors printed as squares, and 110.3 seconds to name 100 inks when each word named a different color. The summary prints the gap as 47.0 seconds, or 74.3 percent of the square time. The medians are 61.9 and 110.4 seconds. Those medians are not the means.
• Practitioner: Before you adopt 74.3 percent, write the other sheet. Reading 100 color names took 2.3 seconds longer in conflicting ink than in black, which is 5.6 percent, and the ratio of that difference to its probable error is 3.64. Stroop calls that increase not reliable. A probable error, here, is his older gauge of how far a difference sits from zero. This page does not turn 3.64 into a modern probability.
• Advanced: The sentence that 99 percent of the word-sheet group sat above the square-sheet mean is his reading of a normal curve, from a gap of 2.5 standard deviations on the word sheet. It is not a count of 99 people who were each slower than their own squares. Eight days of practice on the conflicting sheet cut that day’s average by 16.8 seconds and, on his summary, did not remove the interference.

​

Origin

The sheets, the clock, and the later name are not one publication. Stroop prints the times. MacLeod’s 1991 title is where this page takes the two-word name, without taking his tables. J. Ridley Stroop wrote the paper at George Peabody College for Teachers. The problem grew out of color-naming and word-reading work in the Jesup Psychological Laboratory there. The transcription read here is the York University classics text of Journal of Experimental Psychology, volume 18, pages 643–662. It does not print a DOI or a month. The journal record is December 1935, issue 6, DOI 10.1037/h0054651. The manuscript was received on 15 August 1934. A footnote thanks Dr. Joseph Peterson for encouragement, suggestions, and criticism, and thanks William Fitzgerald of The Peabody Press for help printing the sheets. The same footnote thanks some three hundred college students who served as subjects. The three experiments print 70, 100, and 32. This page does not add those three figures into the three hundred. The colors are red, blue, green, brown, and purple. Stroop started from the Woodworth–Wells color sheet. He set black aside because the comparison words were printed in black, and he set yellow aside because yellow print was hard to match in strength to the other inks. He made both changes after consulting Peterson. No word is printed in the color it names. Each word appears equally often in the other four inks. No color, and no word, immediately follows itself in a row or a column. The conflicting words and the black words were printed from 14-point Franklin lower-case type. In the naming experiment the patch sheet used the same color order as solid squares from 24-point type. He defines the two increases himself. The extra time to read the colored words, over the same words in black, is the interference of color upon reading. The extra time to name the inks of the words, over the same colors in squares, is the interference of the words upon naming colors. Peterson had predicted that the reading increase would not be reliable. The first experiment agreed with that prediction. The paper also cites earlier accounts of why naming a color is slower than reading its name, including Cattell in 1886 and Peterson’s own work. Those papers were not re-read. No cell from them is on this page. MacLeod, “Half a century of research on the Stroop effect: An integrative review,” is Psychological Bulletin, volume 109, number 2, March 1991, pages 163–203, DOI 10.1037/0033-2909.109.2.163. The abstract says the review covers some 400 studies across more than 50 years, isolates 18 findings that a theory of the effect has to capture, and finds two candidate theories wanting: relative speed of processing, and automaticity of reading. It says accounts that put the weight on parallel processing of the irrelevant dimension and the relevant dimension are likely to do better than accounts that look for a single bottleneck. The review’s tables were not re-read. This page does not import a screen time, a clinical cutoff, or a list of the 18 findings.

​

Key Points

Four cuts keep a reading time, a penalized naming time, a normal-curve percentage, and a practice gain from inheriting one another. The squares, the penalty, and the half-sheets each have their own clock.

​

Applications

Use the two sheets when a slow color name is treated as proof that reading broke, or when 74.3 percent is treated as a property of every screen. Keep each clock on the question that produced it.

​

Case Study

In the naming experiment of the 1935 paper, Stroop asked how much a conflicting color word slows the naming of its ink, relative to the same colors printed as squares. The readers were not clicking a screen. They were 100 students, speaking aloud, timed with a stopwatch to a fifth of a second, in daylight from the left. Eighty-eight were undergraduates, 29 men and 59 women. Twelve were graduate students, all women. Everyone named two full sheets of squares and two full sheets of conflicting words. Half began on the squares, in the order squares, words, words, squares. The other half began on the words. A ten-item sample was named before the first full sheet of each kind. The instruction was to name the colors in reading order as quickly as possible and to correct every error. Where the word red was printed in blue, the spoken answer was blue. The squares used the color order of the word sheet, as solid squares from 24-point type. The words were red, blue, green, brown, and purple, and no word was printed in the color it named. Uncorrected errors were penalized. For each one, twice the average time per item on that sheet was added. Fifty-nine percent of the group left an average of 2.6 such errors on the word sheets, across 200 namings. Thirty-two percent left an average of 1.2 on the squares. Before the penalty the means were 108.7 seconds and 63.0. After it they were 110.3 and 63.3. The paper’s interference is the later pair: 47.0 seconds, called 74 percent in the results section and 74.3 percent in the summary. The spread grew with the mean. Standard deviations are 10.8 on the squares and 18.8 on the words. Dividing each by its mean, he gets .171 both times. Medians are 61.9 and 110.4. He places the 47-second gap at 2.5 standard deviations of the word sheet and says, from the normal curve, that 99 percent of the word-sheet scores sat above the square-sheet mean. He does not print how many of the 100 were slower on their own word sheet. The lesson is a spoken list of 100 inks. Conflicting color words raised the mean from 63.3 seconds to 110.3, and a penalty of the sort he calls arbitrary accounts for 1.3 seconds of that 47.0. It is not a reading score, and the reading experiment’s 2.3 seconds was the increase he did not call reliable.

​

Boundaries and Failure Modes

Stroop Effect is the wrong summary when 74.3 percent is applied to every mismatched sign, when the reading sheet is described as the same kind of failure, or when eight days of practice are described as the end of the conflict. The first boundary is the unit, and the penalty inside it. The 47.0 seconds are a mean for 100 spoken names, squares against conflicting words, in this group of 100. The unpenalized pair is 63.0 and 108.7. The penalty, twice the sheet’s average time per item for each uncorrected error, adds 0.3 seconds to the squares and 1.6 seconds to the words. A screen that reports a millisecond, a clinic that reports an error rate, or a child naming a picture book has left this stopwatch. Google Effect includes a color-naming delay for computer words after hard questions, 712 milliseconds against 591 in that paper’s first study. Those words were not color names printed in the wrong ink, and 712 milliseconds is not 47.0 seconds. The second boundary is the practice sample, and it is where the clock changes size. The Arizona group started at 32. Two were dropped for the schedule, and one was set aside after Ishihara’s test. The day’s score is the average of four half-sheets. The summary’s 200 reactions a day are also printed. This page does not divide 200 by 4 and then place 16.8 seconds on the same scale as 47.0 seconds for 100 names. The conflicting-sheet gain is 16.8 seconds, 33.9 percent of the first practice day. The patch-sheet gain is 4.0 seconds. He says less than one fourth of the conflicting-sheet gain is faster color naming. The variability of the practice scores rose from a coefficient of .14, plus or minus .012, to .19, plus or minus .015. That increase divided by its probable error is 2.60, which he calls not reliable. There were two days with no work between test days 3 and 4, and again between test days 8 and 9, because of weekends. The eight practice days are not eight uninterrupted calendar days. The third boundary is the common misuse, including a matched word, a word that is not a color, and a sex story that the paper does not measure. No word was printed in the color it named, so a matching word was not the control. The control in Experiment 2 was a solid square. In Experiment 3 the patch was printed as the hooked figure he names a swastika, so that white sat inside the color the way white sits inside a letter, and so the shades could match the word sheet more closely. A threat word, or any word that does not name a color, was not on these sheets. After the ink practice, reading the conflicting words got slower by about as many seconds as the ink sheet had gained, and the correlation between a person’s ink gain and that person’s reading loss was .262, plus or minus .11. The correlation of the two percentages was .016, plus or minus .17, which he calls zero. One further day of reading practice brought that reading mean from 34.8 seconds to 22.0. His suggestion that girls have more practice naming colors, because talk about dress uses color more often, is an interpretation. The measured sex change he prints is a drop, on the patch sheet, from a difference 5.38 times its probable error to one 2.99 times its probable error. That is not a count of conversations about clothing. Von Restorff Effect is better memory for one item that stands out in a list. On these sheets every word conflicted. Nothing was the single odd item.

​

Common Misconceptions

Three retellings turn one spoken list into a universal percentage, a small reading gap into a second effect of the same size, or a practice gain into the disappearance of the conflict.

​

Related Concepts

The ink-naming increase sits next to a feeling of ease, a faster word decision, a spoken name in an ignored ear, a color name for a computer word, and memory for one odd neighbor in a visual list.

​

One-Line Takeaway