Introduction to Statistics, Review I (Crain, Ronald C.; Havlicek, Larry

Introduction to Statistics, Review I (Crain, Ronald C.; Havlicek, Larry) ... Software designed to acquaint the user with some of the basic statistics ...
0 downloads 0 Views 3MB Size
Introduction t o Statistics RonaldC. Crainand Larry Havlicek. American Chemical Society. 1155 Sixteenth Street. N.W., Washington. DC 20036. Hardware: 48K Apple 11+ (or lle) Software: DOS 3.2.3.2.1,or 3.3;Applesofl Components: One Dlsk Level and Subject: Contmumg Educatm Cost: SlOO 00 OackuDdlskenes. $10 00 each. extra handbooks. $9:00each.

..

Summary Ratlngs Categwy Reviewer I Ease of use Average Subjkct matter Good content Pedagogic veloe Average Studentreaction Average

Rcvlewer 11 Excellent Average

~

AverageGocd Excellent

Review I The computer interactive course "Introduction to Statistic? by Ronald D. Crain and Larry Havlicek has two overall purposes described in the course handbook: "(1) to aequaint [the user] with some of the basic statistics that can he used for analyzing data from a research study and (2) to present the results of that study." There are three components in the course. First is a forty-one page study guide which "serves as course director," second is a primary text (Statistics for Experimenters, George E. P. Box, William G. Hunter, and J. Stuart Hunter, Wiley (1918)) which is an "information resource," and third is a series of seven computer modules which provide "open-ended interactions for examining the course contents, developing skills, and testing." The computer modules are written in EnBASIC augmented Applesoft BASIC to increase their "user-friendly" character. The modules are on a single diskette. In my opinion the course objectives should be achieved by the average user providing that the user is strongly motivated toward understanding the subject matter. Experimental statistics is confusing to most students (and thus to most chemists) regardless of the manner of presentation. This course does not reduce the confusion, but it bas the virtue of being under the full control of the user so that difficult points can be cleared up

hefore proceeding. Further, the interactive nature of the computer engages the user in a way not possible with ordinary texts, or even classroom presentations. For the most part the modules require the learner to "do something" and that is good pedagogy. The intended user is anv oerson workineu in chemistry (since the examples are all chemistry related) at or beyond the college level. Non-chemists can learn from the course as well, but they will need to extend the examples into their own subject areas. Even though the course is described as self-sufficient in the 1984ACS Course Catalog, it is my opinion (and that of my students who used the course) that the texthook is ~~~~~~~essential (current price is $35.95). All madules except the introduction have specific text reading assignments, and some terms and concepts are developed only in the text. At the very least, a standard introductory statistics text of some sort must be used for extended discussion of confusing ideas (with the student relying on the index or table of contents). No major technical difficulties have been cncountrreu in runninp the prugrnms. T h e study a u ~ d rhas explicit mitrucrionr fur starting up the Apple for new users. The prompts are generally clear and inputs are easily entered. The user is informed when slow operations are in progress. On a few oecasions the programs have hung, or given strange results, but the modules are short enough so that a great deal is not lost if the user has to start over. It would be helpful to have an "if all else fails" instruction in the directions for using the Apple, or in an early frame, to reassure the user that if things do go wrong it is frustrating but o.k. to start over. Upon starting the program the user is given a choice between overall instructions (which are well written and explicit) or an index of available modules. If the index is chosen, a title page appears, then several pages of credits and miscellaneous information, another title page, and finally the index. After the number for the chosen module is entered, another title page appears, followed by several frames of objectives, and then an index to the module. It is annoying to have to go through all of this material every time the program is turned on, but no way of avoiding it is urovided (a wav should be-a reader does not read the opening pages of a book every tune . -the hook is opened-the user should at

. I

eviewed In This Issue

Computer L e a r n i n g Package Review

Ronald C. Crain and Larry Havlicek, lntroduction to Statistics

least be in as much control with the eomputer). An excellent feature is the inclusion of a preparation assignment for the next module as part of the "Quit" routine for each module. The statement in the introduction to the index that "each module in this course will take 10to 40 min to complete" should be accompanied with an admonishment to spend several hours in careful reading of the text and study guide aspreparation and in parallel with the computer part. Nearly every module has what I judge to be flaws which can add up to confusion and frustration for the user. The introductory module asks for estimates of mean, made, and median, and will accept a range of answers hut gives no idea of the acceptable range other than to say "that is a little too far off' for unacceptable answers. Even though estimation is an important skill, students are frustrated when some "wrong" answers are "right," while others are truly "wrong."The module on data handling calculates basic statistics from input data. I t specifies a range from 1to 999 hut accepts numbers from 0 to 9999. The display of the normal curve is cut off on the right-hand end, making its symmetry somewhat unclear. When a value more than three standard deviations from the -~. mean is deleted by the program the deletion message overprints part of the data on the screen. Further, it is not madeclear that the deletion is really arbitrary (after all, some scientists maintain that you can neuer delete data, while others use other rejection tests). In the module where standard error is introduced, it is stated that standard error is discussed in a section of the text, but it is a d u ally discussed in the study guide. In the module on analysis of variance sigma is used as the symbol for the total sum of squares about the grand average, while the text uses So. The program reproduces data from the text but does not give explicit information on the example used, so the user has to s e m h for it. If the user goes hack to the text example section after using the "enter your own data" section, the data presented is the user's, not the text's, but the F ratios are from the text data. In the "enter your own data" section after some calculations have been made the following message appears: "Although you have come to a decision an the significance of (Continued on page A166) Reviewer Frank W. Darrow Charles E. Klopfenstein

A165

Carl D. Slater

A166

Books

John C. Bailer, Mary E. Castellion, Cyrus 0. Guss, J a c o b Kleinbert, TheraldMoeller, and Clyde Metz, Chemistry, Second Edition Textbook Announcements Monographs Titles of Interest

A169 A169 A169

Volume 62

Number 5

May 1985

A165

your data, you should have first checked to see if your data is evenly distributed. Let us look a t the data more closely." The data is then displayed, complete with normal curves and hasie statistics, but the user is not told exactlv what constitutes even distribution. In m u event.~I find unfair since the , the - criticism ~ ~ user hasonly been doing ns insrrurted. In the module on wrrelarion and h e a r regreision the least squares routine does not work correctly for small numbers (and small numbers are used in a study guide example). If the numbers are multiplied by a factor of ten, then the results are correct except, of course, for the scale factor. This was the anlv computationnl error ilrucred, but it is n xeriws une. In rhc Rnme m