Statistics for Planners

• Context
• Statistics Learning Packages
• Evaluation

b) Its teaching had mainly resulted from:-

i) Arm-twisting of sufficiently numerate in-house staff.

ii) External tuition from professional statisticians ( They were resident on a campus six miles away, which necessitated a round trip having to be made each time that statistics teaching took place. This resulted in a severe disruption of the timetable).

c) Students selected for Planning & Housing degrees come to university with a wide range of numerate skills ( one could not take it for granted that all students could work out percentages!).

d) Discrepancies in student perception of the relevance of statistics to the general course.

e) Whilst the professional statisticians gave better (ie more error free) information, they were service teaching across several university departments, which made it difficult to specifically address Planning issues.

The teaching remit was fairly basic. Although the course was a preparatory grounding for a ‘research methods’ course in third year, staff comments summarised general expectations, ‘if only students could interpret tables’ or ‘if only they knew what p< .05 actually meant’. More formally, the course was expected to help students read reports and data intelligently and interpret them appropriately. While some students would progress to gather and interpret data of their own, most would ultimately be in management positions where understanding/interpreting figures was sufficient. The course structure included histograms, pie charts, means, modes, medians, standard deviations and inter quartile ranges, issues on sampling and standard error, hypothesis testing, t testing, contingency tables and chi square and correlations. Also included was a session on probability, with a small section on Bayes (personal prejudice) and a session on logic and decision making. The sessions on probability and logic were later omitted, but decision making was retained. The latter mainly concerned the psychological aspects of decision making. It was felt that this would be beneficial for Planning & Housing students who would find it both interesting and conceptually undemanding.

Having read several education reports on the deficiencies of the lecture as a means of communicating information, it was decided that the staging of statistics workshops would be appropriate. They were chosen as a medium to reinforce the concept of ‘learning through doing’ in response to a departmental belief that rising student numbers and increasing demands on staff, would require students to assume more responsibility for their own learning. The format was simple. A brief introduction was followed by more formal handouts and a set of exercises. The notes were a simplified rewrite of statistics books (Rowntrees ‘Statistics Without Tears’ being a core text). The exercises were based on part of a national household survey database to address the issue of relevance. The statistical package used was Minitab. Students were assessed on a formal examination (based around the handouts) and on a workbook which contained the exercises.

Initially, the workshops seemed to work well. There were however a number of problems that had been overlooked.

a) Students seemed less keen to take responsibility for their learning than had been anticipated. In some sense, students felt they were being short changed if they did not receive the standard lecture.

b) Handing out statistical notes and exercises in the same session tended to confuse some students rather than clarify issues.

c) Simplified lecture notes from a basic book were wrongly assumed to be a model of clarity!

d) The workshops increasingly focused on doing the exercises rather than appreciating the principles.

1) GraphIT! (University of Glasgow)

This forms part of the TILT TLTP project suite of learning packages. GraphIT! is an introduction to graphs and plots for basic statistics. The software is downloadable (both the demo and the full version) from the TILT Web site http://www.elec.gla.ac.uk/TILT/packages.html.

The complete pack, which incorporates GraphIT! software, additional datasets and the Editor and workbook, costs £25.00 for UK HE institutions and £50.00 for others. It is obtainable on floppy disk from TLTSN@gla.ac.uk.

Minimum Hardware Requirements

A Windows based 386 machine with VGA/SVGA screen and 8MB RAM. Approximately 5MB disk space and upwards is required, dependent on the use of additional datasets and the number of edits made through the Editor.

2) Introduction to Statistics (University of Nottingham)

This forms one of the TLTP institutional projects under the title ‘Cross Discipline Implementation of CAL into Mainstream Teaching’. The Introduction to Statistics material is aimed at students who have little statistical experience (i.e. nothing more than how to calculate simple arithmetic means). The software costs £15.00 + VAT. Further information may be obtained from calgroup@nottingham.ac.uk. Also see http://www.ccc.nottingham.ac.uk/cal/

Hardware Requirements

A 486 PC with 8MB RAM. 256MB of free hard disk space is also required.

3) Statistics for the Terrified (Stephen Morris, Jill Szuscikiewicz & Mark Preston)

It was thought that the software packages listed would solve several problems. As there was a wide range of numeracy skills amongst students, it was important that individuals would be able to work at their own pace. Insufficient computers for class needs (20 computers for a class of 60 students) made it essential that students be given the opportunity to use the self learning programs outside of scheduled teaching time.

The programs have now been used for one academic session. The structure of the class was changed to incorporate the packages. An introduction was followed by navigation through prescribed elements of the three learning programs. Due to class size, each student worked at a computer for one hour, and with 20 computers in the cluster, the class ran for three hours.

Although the course began well, after three weeks the situation began to get out of hand. Students were at such different points in the program that a summary at the end of the session or a brief introduction at the beginning of the next week, became difficult. In addition, students who missed a week were totally out of phase with others. Some students were simply lost. As a result, specific targets were then written on the board for each session, which students were asked to address. Each session was summarised with a ten minute talk. This seemed to be far more appreciated. In feedback, one comment was ‘Enjoyed your class’, ‘Preferred your talks though to working with the computer programs’. Others felt the statistics learning programs needed to be more integrated with course handouts. On reflection, the students were correct. The handouts were written well before the learning packages became available. Detailed responses are shown in Table 1. It should be emphasised that this is the student response to the way the programs had been used in the class context outlined.

The fact that staff members had been impressed by the statistical programs did not entirely coincide with student perceptions.

Programs which allowed students easy access to the point at which they had left off previously, were most appreciated. There was frustration with one program which although modularised, required students to work through all the module at any sitting.

2)Introduction to Statistics - This program comprises four units.

a)Average and standard deviation

b) Normal distribution

c) Sample and distribution

d) t distribution

The units include:-

• Simple data description (including standard error and confidence intervals)
• How to choose a test
• What does p< .05 actually mean
• Testing for difference between groups (parametric and nonparametric tests)
• Analysing repeated measures
• Analysing 2 x 2 classification tables
• Fitting lines to data
• Uncovering hidden influences

A test quiz is embedded in the program.

Apart from its use of medical examples, this was the best program viewed, with clarity of presentation and definition and progressive steps which seemed appropriate. Several students commented favourably and it was observed that there were less questions asked during scheduled classes.

We are now in the process of reviewing our teaching for next year and reflecting on student comment.

Two of the programs were thought to be so good (GraphIT! and Statistics for the Terrified), that the lecturer could almost become redundant. This is clearly not the case. Students do need a framework from staff, on which learning should be based. As term progressed, more and more remedial teaching in the old-fashioned lecture style was adopted. The programs nonetheless are excellent support to more conventional learning. It is difficult within an educational context to identify an appropriate validation criterion for learning - Is it exam performance? Is it final degree status? Is it use 10 years after graduation?

The class examination failure rate was low. However, there is no way of evaluating the effectiveness of the programs beyond student feedback. Nonetheless, their use is strongly recommended. Currently the programs are being used within a research methods course for PhD students.