91社区

Even with a group of excellent instructors, rankings will present strong instructors as being "below average." Rankings by default diminish the accomplishments of some instructors, even if all are excellent teachers.

An instructor who may be below the unit mean in a unit with a strong teaching culture may still be teaching well.听

To form a comprehensive and meaningful understanding of an individual's teaching, it is important to consider the entire pattern of results from a number of different courses and classes of students over a period of time.

Research suggests that data should be reviewed from courses every term for at least two years, totalling at least five courses.

In many academic units, there are different forms used for different types of courses; e.g., labs, seminars, undergraduate vs. graduate. As the means are calculated separately for each questionnaire in use in your unit, the comparator group is automatically generated.

When a course is a section of a multi-section offering, the other sections of the course are a good starting point.听As well, courses of the same level within the Faculty can help contextualize results.

Note that when comparisons are made, it is important to safeguard the confidentiality of results of the comparison group.

When interpreting the numerical results, consider information such as the distribution of responses by item as well as the variation in responses. To understand the range of opinion, interpret the mean in conjunction with the shape and frequency of responses along the scale.

Generally, differences that are less than .5 above or below the comparison mean (department, Faculty level, or class size) should be regarded as functionally equivalent.

(e.g., 鈥淥verall 听this instructor is an excellent teacher鈥�; 鈥淚 learned a great deal from this course鈥�). Responses to these questions are found to correlate most consistently with measures of actual student achievement. How an individual instructor compares to the departmental mean is actually less important than the overall rating. It is reasonable to strive to have all instructors and courses obtaining solid or strong results, but not everyone can be above average.

Generally, mean scores over 4 are considered strong, means from 3.5 to 4 represent solid results, and results below 3.5 should be of concern.听As well, it is advisable to follow-up on any result that is more than .5 below or above the comparison mean (department, Faculty by level or class size).

A standard deviation for a question greater than 1 indicates relatively high differences of opinion; in such cases, comments can be particularly useful to help understand the variation.听

If follow-up analyses are carried out on the data, do not look beyond 1 decimal place. Small differences that are statistically significant are common with large sample sizes. As a result, it is important to ask whether the difference is large enough to have practical implications.

For example, if two instructors in a department receive average ratings of 4.7 and 4.8 on the question "Overall, this instructor is an excellent teacher"; it would be difficult to argue that the difference of 0.1, although it might be statistically significant, is large enough to claim that the instructor with a rating of 4.8 is a better teacher.听

The Comments Analysis Worksheet is a tool to summarize comments to facilitate analysis.

• Focus on specific, descriptive items and look for patterns. Specific teaching behaviours (e.g., clarity of objectives) are easier to change than personal characteristics (e.g., enthusiasm).听
• Consider ratings in relation to written comments to see if the latter provide indicators and suggestions for improvement.
• If there is considerable variation in response on an item (e.g., some report assignments as appropriately challenging and others as too challenging), it may represent important differences in the nature of the students (e.g., senior versus first year, or an uneven distribution of background preparation for the course).

The number of students who submitted the evaluation.

The number of students who declined the evaluation.

The number of students enrolled in the course.

The sum of the 鈥淭otal Number of Completed Evaluations鈥� and the "Total Number of Declined Evaluations," divided by the 鈥淭otal Enrollment in the Course or Course Section," expressed as a percentage.

The 鈥淭otal Number of Completed Evaluations鈥� divided by the 鈥淭otal Enrollment in the Course or Course Section," expressed as a percentage.

The number of students who selected one of the options provided for the question.

The number of students who did not select an option for the question. Blank responses are NOT included in any calculations. Note: The number of 鈥淰alid responses鈥� plus the number of 鈥渂lank responses鈥� should equal the 鈥淭otal Number of Completed Evaluations鈥�.

The number of students who entered a comment for the question.

The sum of all 鈥渧alid responses鈥� divided by the number of 鈥渧alid responses鈥�. Note: The closer the mean is to 5, the closer it is to 鈥渟trongly agree/ excellent鈥�.

The standard deviation is a measure of the variation of the distribution of a data set. The standard deviation provides information about the distribution of responses, and underlines the danger of looking at the mean alone without considering the variance.

Formula: The standard deviation is calculated as the square root of the arithmetic mean of the squares of the deviation from the arithmetic mean. As this is a calculation on a sample rather than the population, the result is an estimated standard deviation and is expressed in the same units as the data.

Why look at the Standard Deviation?

For example, the following three cases illustrate how the same mean, 3, summarizes three data sets with completely different features. Clearly, the estimated standard deviation is needed in conjunction with the sample mean to better describe a data set.

This is an estimate of how representative the sample mean (those responding to the questionnaire) is of the population mean (the whole class).

Formula: The variability of the sample mean is, naturally, smaller than the variability in the individual observations. This is usually taken to be the estimated standard deviation of the observations, divided by the square root of the sample size. When sampling from a 鈥渟mall鈥� finite population this variability will be reduced. A finite population estimator of the standard deviation of the sample mean is given by:

This is the mean calculated treating all courses as if they were combined into one. It is the sum of all valid responses for a question in all courses in the department divided by the total number of such responses.

This is calculated as the 鈥渕ean of means鈥� and gives equal weight to mean evaluations from classes of different sizes. For this mean, the sum of the mean for each question per course is divided by the number of courses. It is calculated to avoid any weighting due to large courses in a department but perhaps gives undue weight to mean evaluations reported for classes with very small enrolments.

The decision about which mean to use depends on the course size and number of respondents. The Dept. Mean is more commonly used when there is little difference (+-.1) between the two means. This usually occurs when courses within a department have relatively uniform enrolments and there are similar response rates for all courses.

However, a range of course sizes or number of respondents within a department usually results in differences between the two means. Therefore, to compare like to like, large courses should be compared to the Dept Mean and small courses to the Dept Course Mean.

For example, Dept X has enrolments ranging from 6 to 400. For a given semester, the question, 鈥淥verall, this is an excellent course,鈥� gives a Dept Mean of 3.4 and a Dept Course Mean of 3.7.