Sentence level errors by students have been investigated in two different ways. First, researchers have had informants judge errors that students commit. Connors and Lunsford (1988, 1992) identified the most common kinds of errors in students' writing, and Hairston (1981) documented that employers in the business community are disturbed by surface feature errors. Errors that almost all of her informants found highly irritating were labeled "status marking." The second line of research used the writing of English as a Second Language students to look at how faculty judge errors. This line of inquiry revealed that faculty judge some errors as more "grievous" than others (Vann, Meyer, & Lorenz, 1984; Janopoulos,1992) and that the relative degree of displeasure with which professors greet particular errors varies with discipline and age (Santos, 1988). To our knowledge, no study has attempted to relate judgments about errors or how frequently errors are committed.
These are real questions because facility and accuracy in writing come from years of practice rather than from intensive labor in single courses, and they are habits that must be reinforced by writing in a variety of contexts and for a variety of purposes. Skills acquired in one context do not automatically transfer to another, and even when skills may seem to be mastered, students often write carelessly because they believe that writing errors do not matter outside of English classes.
It would be nice to be able to send a clear message to our students that their teachers, regardless of discipline or other variables, do in fact value accurate writing, and that some consensus exists about what kinds of errors matter most. Accordingly, we asked the questions listed on page 1.
We decided to have three examples of most errors so that we could judge whether our subjects were responding to the error or to the sentence in which the error occurred. Three examples is not a large sample, but it is more informative than the uncategorized single examples used by Hairston. Because we wanted our survey to be short enough that our colleagues would be willing to complete it, we did not attempt to suggest the context of the sentences or include the passages that they were taken from. To ensure an even distribution of error types, we divided the survey instrument into thirds, placing one example of each error type in each third, using a random number table to determine placement of the items within each section.
Because many errors, even when organized into subcategories, may appear in almost infinite variety, we exercised some arbitrary control over the kinds of errors we would use. For example, we wanted three different kinds of tense shift instead of three examples of one kind, e.g., present to past, and different conjunctions (and, but, so) in the category of no comma in compound sentences. We wanted different words as both examples of nonstandard verb forms (reran, wrote, sang) and use of a wrong word (enterprises, cites, intertwine). For the missing comma after an introductory element, we decided to use elements of one word, a phrase, and a clause. Technically, these decisions mean that we actually have only one example of each of these errors. But it could also be argued that since every sentence and every reader is different, every sentence-reader interaction is a unique subtype. We thought our decision allowed us to maximize both rigor and comprehensiveness.
For the six confused-pairs errors and the it's/its errors, we thought that two examples of each was enough to make the point. We collected one example of each version of the error, e.g., one incorrect affect and one wrong effect. The four commonly misspelled words, we thought, could be represented by a single example of each. These eighteen items were distributed evenly throughout the survey, again, using the random number table to determine the exact placement.
To ensure that our colleagues would rate the items honestly instead of just marking every sentence as having a serious error, we inserted six correct (OK) sentences, taken from Connors and Lunsford's St. Martin's Handbook, one on each page of the survey but randomly placed on the page. This plan resulted in a survey containing 78 items, with 13 items per page (Appendix A).
Except for the six OK sentences, the survey items all came from typed student papers, i.e., papers written out of class, meaning that the writers had had opportunity to revise their work. The sentences were all written by native speakers of English, mostly students enrolled in our own classes; almost half of the items were written by upperclassmen. We shortened or otherwise modified some sentences to eliminate other errors and expressions that might confuse or mislead the raters; for example, the OK sentence "Pens poised in anticipation, the students waited for the test to be distributed" was revised to begin "With pens poised" because subjects in pilot tests kept marking it incorrect. The survey was pilot tested on nine English graduate assistants and then, in a second round, on eight upperclassmen enrolled in a section of Advanced Rhetoric taught by a colleague. The pilot subjects were asked to circle the error in each incorrect sentence. We interviewed the pilot raters about their marking of the survey items; sentences whose error was consistently misidentified were thrown out, and new sentences were tested and inserted. The OK sentences were also pilot tested to ensure that readers would recognize them as being correct. Getting such agreement proved surprisingly difficult! We informed the pilot raters and faculty of the presence of these correct sentences but not of their number or location.
When pilot subjects ignored our instruction to circle the error that they were responding to, we could not be sure that they were in fact responding to the intended error. Interviews showed that sometimes they were responding to some other aspect of the sentence. We therefore decided to ask our faculty raters to circle the error in each sentence where they detected one.
Because of the length of the survey and the large number of error types being surveyed, we anticipated that raters might have extra trouble with the first part of the survey. We therefore created three versions of the survey, with the parts arranged 123, 231, and 312. We also hoped that this strategy would discourage our colleagues from doing the survey collaboratively.
The survey, with the demographics sheet and a cover letter, was administered to 27 of the 34 academic departments, including their deans, and to the department of Educational Services (n = 381). Because of the survey's length and the potentially sensitive nature of the request to mark errors in sentences, no forms were distributed until one of the researchers had first met with either the department chair or, preferably, with the department, to explain the purpose of the survey and the reasons for the instructions. Departments not included in the survey were those with whom we were unable to schedule a meeting. We endeavored to distribute equal numbers of each type of form to each department. The participating groups were:
Evidence for a hierarchy of errors. Our results, as indicated both by mean level of irritation and by the number of "correct sentence" responses, indicate that a hierarchy does indeed exist. Figure 1 shows the mean irritation score for each error category contrasted with the total number of zero ratings (overlooks) for the error category, plotted on an X-Y axis. Highly irritating errors, like nonstandard verbs, have both high means and few zeroes, meaning that not only did raters consider them serious, but they also spotted the errors whenever they occurred. By contrast, the OK sentences and comma errors have low means and large numbers of zeroes, indicating that people often overlooked these errors and did not consider them very serious when they did see them. Because the most irritating errors were seen by almost all of the raters, we interpreted the large number of zeroes on the less serious errors as confirmation that readers do indeed consider these mistakes relatively unimportant. The spelling errors had especially high standard deviations (X = 1.4, vs. the X S.D. of 1.3) and low numbers of zeroes (X = 12.9, vs. the overall X of 25.5 zeroes), indicating that although most of the raters detected the misspellings, they disagreed about the significance of these errors.
Relationship between frequency and mean irritation score. Figure 1 also shows a clear negative correlation between error frequency and irritation level (r = -.77, p = < .00001), with the eleven most irritating errors clustered in the upper left corner and the six OK sentences in the lower right corner. This means that the most highly irritating errors, as determined by this study, occur relatively seldom in students' writing (Connors & Lunsford, 1988), and the most frequently occurring errors were judged by this faculty to be relatively innocuous.
Agreement of English teachers with teachers in other disciplines about the hierarchy of errors. Again, the results indicate a high level of agreement across disciplines as to which errors are more or less serious. In a preliminary evaluation, we looked at English faculty and Educational Development Center faculty separately, and grouped the other faculty together as "other." Spearman Rank correlations show a highly significant agreement among the three groups, as shown in Figure 2:
Figure 2: Spearman rank correlations for error rating among English, Educational Development, and all other faculty.
_____________________________ Other X EDC X (n = 118) ENG X .80* .84* (n = 19) EDC X .88* (n = 7) ___________________________ *p = < .00001Sample correlations showed less high but equally significant correlations among the three groups. The Educational Development faculty had higher mean irritation scores for the most serious errors than did the other two groups, and the content area faculty had the lowest irritation scores. The EDC faculty were also the most accurate of the three groups at detecting OK sentences.
Because the number of faculty from each department that participated in the study was often very small, we were unable to run other correlations among individual departments. However, we thought that it might be instructive to examine correlations among disciplines. Accordingly, we divided the participating academic faculty into four "disciplinary" groups, as follows:
Hard Science: Chemistry & Physics, Computer and Office Information Systems, Biology & Earth Sciences (n = 12)
Liberal Arts: Art, Communication, Economics & Finance, Educational Development Center, English & Philosophy, Music, Political Science & Geography, Psychology & Counselor Education, Sociology & Social Work, Theatre (n = 59)
Vocational Technology: Agriculture, Electronics Technology, Graphics, Human Environmental Sciences, Manufacturing & Construction, Nursing, Power & Transportation (n = 29)
Demographic Factors Influencing Response to Errors. As Table 1 shows, certain groups of faculty were more irritated by the errors in student writing than other groups were. However, when the errors were plotted on an X-Y axis by each separate demographic factor, the same pattern of overall agreement appeared. For example, although women, as in Hairston's study, were consistently and significantly more irritated by specific errors than men were, by almost .3 of a point, the sexes agreed on which errors were relatively more or less serious, as shown in Figure 4.
Likewise, although full professors were more irritated by errors than associate professors, by .45 of a point, and although "other" faculty, e.g., graduate assistants, were markedly less attuned to errors than were the fulltime teachers, respondents of all academic ranks indicated broad agreement about the relative seriousness of individual errors (see Figure 5).
The only demographic factor that revealed what we considered a meaningful difference in response was amount of writing seen in a semester: Teachers who reported seeing "a lot" of writing were not only more annoyed by errors than were teachers who reported seeing "some" or "little" writing, but they were also much more accurate in detecting the OK sentences than were teachers who see relatively little writing (see Figure 6).
Accuracy of Responses to the Survey. One of the most common student complaints about teachers is that the marking of errors is idiosyncratic, i.e., that what one teacher will mark as an error, another teacher will allow. Our results tend to confirm that impression, within certain limits. To see how consistently teachers found the errors in the sentences, we counted "overlooks," i.e., sentences incorrectly marked zero (correct). And to see if teachers were in fact responding to the errors that we left in the sentences, we counted "misdiagnoses," i.e., marking of anything in the sentence other than the intended error.
The results, given in Table 2, show that the women in our study were more accurate than the men in spotting errors, that more experienced teachers overlooked far fewer errors than inexperienced teachers, that full professors overlooked fewer errors than did their lower-ranking colleagues, and that teachers in the Hard Sciences and teachers who assign much writing spotted more errors than did teachers in other disciplines and teachers who assign relatively little writing. Thus, teachers who saw errors tended to be more highly annoyed by them than were teachers who were more likely to overlook errors.
In the accuracy of diagnosis, we found other patterns. Women were more attuned to our categories of error than were men, as were teachers who assign much writing. But we found high percentages of "many" misdiagnoses by older teachers, higher-ranking teachers, and teachers in the vocational technology departments. In other words, many teachers have their own ideas about what constitutes an error. Some of the corrections clearly involved stylistic preferences, as in this response to OK sentence #1 (Connors & Lunsford, 339; the hypercorrections and comments are boldfaced) "There are many people who fear success because they fear they do not deserve it. lacks economy." We found objections to correctly placed commas (the "correct" errors are in bold; "Being a nonmember, they allowed me to attend one of their meetings, and I was quite surprised") and correctly used prepositions ("It was a place were my big brother and I could spend hours teasing and talking to [with] each other").
It is worth noting that no teacher correctly diagnosed each error in the survey. In other words, every participant reacted at least once to some facet of a sentence other than the specific error that got the sentence chosen for the survey in the first place.
Teachers in all disciplines agree on what the error hierarchy is. Although the large standard deviations show that almost any error will receive a spectrum of responses ranging from "this is accurate" to "this is unacceptable," there is a broad consensus -- at least on this campus, among those faculty who were interested enough in this project to give an hour of their time to it -- that a hierarchy of errors exists and that we agree on what that hierarchy is. This is welcome news.
Moreover, the scoring differences among the three groups (English, Educational Development, everyone else) are what one would hope and expect to see. Since EDC faculty teach developmental writing, with heavy emphasis on surface feature accuracy, one would expect them to be vigilant and intolerant of error. It is reassuring to see that the English faculty agreed with their colleagues in other departments about what errors matter; perhaps predictably, the English faculty were the most picky about the "rightness" of the OK sentences.
Of course, these results do not "prove" that content area faculty are just as competent to detect and respond to errors as are English teachers; the low response rate suggest that perhaps only those faculty who felt competent to do this work participated in the survey. Clearly, however, many faculty in a variety of disciplines both can and do do a good job of responding to errors. This is a message that we want to share with each other and with our students.
The negative correlation between frequency and seriousness of error is also reassuring, since it suggests that students mostly avoid making the most serious status-marking errors. But that is not a message that this team of researchers wants to send to students, since we suspect that they might take it as permission to continue making "less irritating" kinds of errors. We are thus heartened to see the presence of four homophonic spelling errors among the eleven most serious errors (to/too, were/where, there/their, your/you're). These are words that everyone learns in first grade, and mistakes in using them will not be caught by computer spellchecking programs (although some text-editing programs will catch them). We faculty can therefore send our students a firm and unanimous message that we see these misspellings, including those that do not irritate us, and that continued carelessness and failure to edit manually have a clear negative effect on the perceived quality of writing.
As for the least serious errors, which involved commas, we account for their relative acceptability in three ways: 1) Conventions of comma usage may be changing. 2) There is no clear, easily teachable rule for, say, when to insert a comma after an introductory word or word string. 3) People may be confused about how to use commas and may have decided that they are just too much trouble to worry about. Given Sloan's [1990] finding that professional writers made as many comma errors as college freshmen, this last explanation seems especially likely.
Women and teachers who see much student writing every semester are more aware of errors, more annoyed by them, and more accurate in detecting them than are other instructors. It would be surprising indeed if vigilance and accuracy regarding writing errors did not correlate positively with the amount of writing assigned. Apparently, the more sensitized one becomes to error, the more of it one sees, and vice versa. We are at a loss, however, to account for the gender difference in attitude toward error. When we presented these results to our colleagues, they suggested that women may be socially conditioned to care more about writing errors, that women may be more in touch with their feelings than men, and that men may be less willing to acknowledge being bothered by things generally, including writing errors. We also note that the different percentages of women in our four "disciplines" might account for the variations in how annoying members of these disciplines, as groups, considered the errors.
Teachers do not agree about what constitutes correctness. Although we found clear agreement about certain matters of grammar (e.g., nonstandard verbs), spelling, and some matters of punctuation (apostrophe usage), the fact that so many teachers objected to the OK sentences and identified "nonmarked" errors in the test sentences suggests that teachers have personal preferences that they apply to students' writing. Cognitive reading theory explains why the teachers gave every possible response, from 0 to 6, to almost every item on the survey. Since the concepts of "correctness" and "style" are enacted through innumerable instances of daily usage, the distinction between them must be blurred; on a scale of relative seriousness, probably, specific categories of, say, grammar errors and stylistic lapses would overlap. For example, one respondent circled both the "you" and the missing comma in the sentence "If you observe one of the concert choir's classes you will see the hard work that the students and instructor put in their practicing," and commented "generalized 'you' = 3, comma needed = 3," apparently meaning that he considered the punctuation error and the use of the second person as equally serious faults. The discussion by Flower et al. (1986) of how writers think about their texts applies, we believe, equally to professorial readers: They
1) The existence of a cross-disciplinary agreement about a hierarchy of error means that students need to know that these errors count against them in all of the college writing that they do.
2) The lack of agreement about standards of correctness means that we owe it to our students to tell them what we as individual teachers consider correct.
3) The lack of accuracy in doing the survey means that we should perhaps express our judgments about correctness with a bit of humility.
On the other hand, students must accept that different writing contexts evoke broadly similar but by no means identical expectations. But they cannot learn this lesson unless we teach it to them by telling them clearly what we expect.
Implications for Writing and Research. We plan to extend this study by replicating it with employers. We want to learn whether the cross-disciplinary consensus about error extends beyond the boundaries of this campus and whether nonacademics place the same relative importance on errors as we do. In pursuit of this study, we expect to be contacting various offices on campus for names of area employers, especially businesses that employ CMSU graduates. We also think that it might be worthwhile for individual departments, as part of their CPI work, to survey their alumni about their current attitudes toward writing, how their attitudes about writing have changed since graduation, and their satisfaction with the training in writing that they received here.
___. (1992). Exorcising demonolatry: Spelling patterns and pedagogies in college writing. Written Communication, 9 (3), 404-26.
Flower, L., Hayes, J.R., Carey, L., Schriver, K., & Stratman, J. (1986). Detection, diagnosis, and the strategies of revision. College Composition and Communication, 37,16-55.
Hairston, M. (1981). Not all errors are created equal. College English, 43, 794-806.
Hacker, D. (1991). The Bedford handbook for writers. 3rd edition. Boston: Bedford Books of St. Martin's Press.
Janopoulos, M. (1992). University faculty tolerance of NS and NNS writing errors: A comparison. Journal of Second Language Writing, 1,109-121.
Lunsford, A. and Connors, R. (1989, 1992). The St. Martin's handbook. 1st and 2nd editions. New York: St. Martin's Press.
Santos, T. (1988). Professors' reactions to the academic writing of nonnative-speaking students. TESOL Quarterly , 22, 69-90.
Sloan, G. (1990). Frequency of errors in essays by college freshmen and by professional writes. College Composition and Communication, 41, 299-308.
Vann, R., Meyer, D., & Lorenz, F. (1984). Error gravity: A study of faculty opinion of ESL errors. TESOL Quarterly, 18 (3), 427-440.
Appendix B: Categories of Errors Studied, Listed in Order from Most to Least Bothersome, with Mean Score and an Example.
Table 1: Demographic groups ranked by mean irritation scores.
Table 2: Overlooks and "misdiagnoses," arranged by demographic groups.
Figure 1. X-Y axis showing average response to each error category, contrasted with the total number of zeroes (overlooks) for each category, including OK sentences.
Figure 2. Spearman rank correlations for error rating among English, Educational Development, and all other faculty.
Figure 3. Responses to the error categories factored by four academic "disciplines".
Figure 4. Responses to the error categories factored by gender.
Figure 5. Responses to the error categories factored by academic rank.
Figure 6. Responses to the error categories factored by amount of writing seen per semester.
Rank Category Mean Irritation Score 1. nonstandard verb forms 4.5 These movies seemed to be reran over and over. 2. you're/your 4.4 I'm not saying that you're child definitely suffers from this phobia, just that it's a possibility. 3. their/there 4.3 Students should tap into there teachers' knowledge and expertise. 4. sentence fragment 4.3 No explanation as to why one phrase sounds better than the other, leading me to guess that grading is just subjective. 5. subject-verb agreement 4.2 The family and home is important in making this country a wonderful place to live. 6. wrong preposition in verb phrase 4.2 We could of run to breakfast faster than the bus got us there. 7. too/to 4.2 I felt these remarks to be to general to give me a better understanding of ways to improve my writing abilities. 8. were/where 4.2 It was a place were my big brother and I could spend hours teasing and talking to each other. 9. pronoun agreement 4.1 An ordinary kid would come home from school every day and throw themselves in front of the TV. 10. object pronouns as subjects 4.0 The girl talked about how her and her friends would go out and steal things. 11. run-on sentences 4.0 The topics are good they cover a lot of controversial issues. 12. definitely (definately) 3.8 You can take care of your body by lifting weights, running, and most definately by not taking drugs. 13. tense shift 3.7 This imaginary stone is large and round, much like me, and had always wondered about the sea. 14. it's/its 3.7 It's once flawless complexion is now a muddy brown color. 15. lose/loose 3.6 The mare was loosing a lot of blood and struggling badly. 16. wrong word 3.6 The article also cites that children with dyslexia usually begin to speak late and cannot read aloud. 17. dangling modifier 3.5 By being told what's incorrect in their compositions, the teacher may give a wrong direction to the writers. 18. comma splice 3.5 I'm in college now, that is perhaps the greatest change I have ever made. 19. affect/effect 3.5 My high school French club had a tremendous affect on my plans for my future. 20. non-parallel order 3.5 My goals include maintaining a G.P.A. in the 3.0 range and to try to be happy with my life. 21. vague pronoun reference 3.5 I have weaknesses in the area of finding the right words for what I want to say. It ends up confusing and hard for readers to understand. 22. a lot (alot) 3.3 I feel I need to build alot of endurance to make it through college. 23. receive (recieve) 3.3 In Missouri, coaches recieve stipends ranging from $500 to $2000 per year. 24. separate (seperate) 3.3 In order to recycle your cans, bottles, and paper, keep them in seperate bags. 25. incorrect relative pronoun 3.3 Yesterday, the person that always rides with me to school was not ready on time. 26. apostrophe used as a possessive 3.2 Television is needed in todays society to enable people to learn about issues of fiction and non-fiction. 27. incorrect commas with parenthetical or nonrestrictive element 3.2 Mike Rose's, Lives on the Boundary, finds a middle ground but still gets its message across. 28. OK sentence #5 3.0 The two drivers involved in the accident, who have both been convicted of drunken driving, should lose their licenses. (correctly punctuated restrictive clause) 29. no comma in compound sentence 2.8 My sources suggested using tented overlays on text but I modified the technique for use with the overhead projector. 30. OK sentence #2 2.8 Seventy years ago, Americans speculated in Florida real estate as though it were a risk-free investment. (correct use of subjunctive tense) 31. OK sentence #3 2.8 The job demands that the employee be in top physical condition. (correct use of subjunctive mood) 32. OK sentence #4 2.7 An important nineteen-century sociologist was Karl Marx, who believed that his role as a social thinker was to change the world. (correctly punctuated dependent clause) 33. no comma in introductory element 2.7 Instead of just taking out words the student might choose a different approach to the paper. 34. OK sentence #1 2.3 There are many people who fear success because they fear they do not deserve it. (correct--or at least acceptable--use of "there" as an expletive) 35. OK sentence #6 2.1 With pens poised in anticipation, the students waited for the test to be distributed. (correctly punctuated introductory phrase)
Average Group Number in Group 4.06 Business faculty 28 3.93 31+ years of teaching 10 3.9 under 30 years of age 8 3.9 Full Professors 48 3.88 women 42 3.87 11 - 20 years of teaching 42 3.83 see a lot of writing 29 3.82 Hard Science faculty 16 3.78 40 - 49 years of age 38 3.77 Assistant Professors 34 3.71 30 - 39 years of age 24 3.71 3 - 10 years of teaching 35 3.7 50 - 59 years of age 51 3.67 see little writing 80 3.63 21 - 30 years of teaching 41 3.63 see some writing 34 3.61 men 101 3.6 Instructors 27 3.6 Liberal Arts faculty 68 3.48 over 60 years of age 22 3.45 Associate Professors 27 3.35 Vocational Technology faculty 31 3.22 0 - 2 years of teaching 14 3.12 not having faculty rank 6
% Overlooks % Misdiagnoses Group Average n* none low some many none low some many X Gender women 39 20.5 48.5 25.6 5.1 0 28.2 30.8 4.1 3.88 men 93 5.4 52.7 26.9 15.1 0 16.1 30.1 53.9 3.61 Age under 30 8 12.5 12.5 37.5 37.5 0 25.0 37.5 37.5 3.90 aged 30 - 39 22 4.5 50.0 36.4 9.1 0 27.1 18.2 54.5 3.71 aged 40 - 49 33 12.1 45.5 30.3 12.1 0 15.2 36.4 48.5 3.78 aged 50 - 59 48 10.4 58.3 22.9 8.3 0 18.7 29.2 52.1 3.70 over 60 21 9.5 57.1 19.0 14.3 0 19.0 33.3 47.6 3.48 Years of Teaching 0 -2 yrs. 14 7.1 28.6 28.6 35.7 0 14.3 28.6 57.1 3.22 3 - 10 yrs. 32 6.3 53.1 31.3 9.4 0 15.6 40.6 43.8 3.71 11 - 20 yrs. 39 17.9 43.6 25.6 12.8 0 25.6 23.1 51.3 3.87 21 - 30 yrs. 37 8.1 56.8 29.7 5.4 0 24.3 24.3 51.4 3.63 31 + yrs. 9 0 88.9 11.1 0 0 0 44.4 55.6 3.93 Rank Instructor 26 11.1 53.8 19.2 15.4 0 19.2 38.5 42.3 3.60 Assistant 33 12.1 45.5 30.3 12.1 0 27.3 21.2 51.5 3.77 Associate 23 8.7 39.1 43.5 8.7 0 26.1 26.1 47.8 3.45 Full Prof 45 8.9 62.2 20.0 8.9 0 13.3 33.3 53.3 3.90 Non-faculty 4 0 50.0 0 50.0 0 0 50.0 50.0 3.12 Amt. of Writing Assigned little writing 72 9.7 48.6 26.4 15.3 0 11.1 30.6 58.3 3.67 some writing 32 3.1 56.3 28.1 12.5 0 15.6 28.1 56.3 3.63 much writing 28 17.9 53.6 25.0 3.6 0 37.5 37.5 25.0 3.83 Academic Discipline Business 25 16.0 44.0 16.0 24.0 0 8.0 40.0 52.0 4.06 Liberal Arts 59 10.2 52.5 27.1 10.2 0 30.5 28.8 40.7 3.60 Hrd Sciences 12 0 75.0 0 25.0 0 25.0 25.0 50.0 3.82 Voc Tech 29 3.4 37.9 55.2 3.4 0 3.4 24.1 72.4 3.35 ______________________ *The n's are smaller for this data than for other data sets because some respondents sent back blank pages, or responded to sentences as containing an error without indicating what the error was. Those data forms were not used for this analysis.
