APPENDIX V

STATISTICAL ANALYSIS

anal

To determine the statistical significance of the data obtained during this study, a one-way ysis of variance technique was employed. For the 0-12,000 mile test sequence, the hydrocarbon (HC) and the carbon monoxide (CO) data were analyzed for significant differences due to both the fuels and the oils used. For the test data from those vehicles for which the test sequence was extended to 24,000 miles, only the fuels effect on CO and HC emissions was analyzed.

Bowker, Albert H., and Lieberman, Gerald J., Engineering Statistics, Prentice-Hall, Inc., Englewood Cliffs, Nac

1

Jersey, 1965.

If FS F

there is no significant difference between the columns at the confidence level

there is a significant difference between the columns at the confidence level

The application of this technique to the HC data taken from vehicles using Indolene 30 and Indolene Clear fuel for the 3000-24,000 mile test sequence is carried out as an example. The data array is shown in Table AV-2.

The completed analysis of variance table is shown in Table AV-3.

The confidence level chosen was 99.9 percent. From a table of values for F.001, it is seen that F.001 1,30 = 13.3.

Since the Fration from the analysis of variance table is greater than F.001 1,30. there is a sig nificant difference between exhaust HC concentrations of vehicles using Indolene Clear and Indolene 30 gasoline at the 99.9 percent confidence level.

Vehicles 113 and 114 were run with Indolene 30 fuel from 3000 to 24,000 miles. From 24.000 to 30,000 miles, these two vehicles were run on Indolene Clear. A statistical analysis of these data was performed to determine whether there was a significant difference between the means of HC data obtained from 3000 to 24,000 miles and that obtained from 24,000 to 30,000 miles. The Student "t" statistic was chosen for this analysis. A summary of the data is contained in Table AV-4.

The calculated values for the means (X1) Standard Deviations (S1) and variances (S;2) are shown in Table AV-5.

The statistic"" is computed from this data by means of the following formula:

The value of "" computed from this formula is 3.26. This value of't "is compared to the value of a taken from a table of values for the "t"distribution. The symbol a refers to the confidence level chosen, and refers to the degrees of freedom (v = N, + N2 − 2). If the computed value of t is greater than av for the chosen confidence level, the difference in the means is significant. For a confidence level of 99.5 percent,

The difference in the means of the data obtained with Indolene 30 is significantly different from the mean of the data obtained with Indolene Clear at the 99.5 percent level.

Senator MUSKIE. Let us go on with our next statement. I think we have one more statement.

General Motors.

STATEMENT OF HARRY F. BARR, VICE PRESIDENT IN CHARGE OF ENGINEERING, GENERAL MOTORS CORP.

Mr. BARR. Good morning, Mr. Chairman and members of the subcommittee.

I am Harry F. Barr, vice president in charge of the engineering staff, General Motors Corp. It is my pleasure to speak to you on behalf of General Motors Corp. in response to the invitation extended to us by Senator Muskie. Some of our research and engineering people are with me today to assist in answering questions you may have after our presentation.

We hope that the information we are presenting today will estab lish to your satisfaction that we are determined to find solutions to the problem that concerns all of us. We hope, too, that your schedule will permit you, tomorrow, to see the air pollution research and engineering activities at our technical center and we are gratified you can do this. We are confident that a personal inspection would demonstrate to you how intensively we are working in this field and the progress we are achieving.

As Mr. Roche advised you in his letter of January 27, Mr. Chairman, you and your subcommittee can be assured of General Motors complete cooperation. And you can be equally sure, as Mr. Roche said, of our continued support and earnest desire to assist in every way we can in reaching a satisfactory solution to the automotive emission problem.

As has been the case in some complex scientific and medical problems, the solutions sometimes have been elusive and the search difficult. However, General Motors long ago accepted this challenge and for many years has been doing intensive research into the problem. We have shared the sense of urgency that you have expressed, Mr. Chairman, in our earnest search for solutions.

We feel that the record shows that General Motors has done and is doing significant work in the control of automotive emissions.

In the early 1950's the main effort was to determine the nature of the problem. As soon as new knowledge accumulated from various research activities, it was translated into experimental hardware or improved detection instrumentation and then exhaustively tested.

Tangible results obtained from this General Motors research have included: The positive crankcase ventilation control system, which individual car manufacturers introduced on their 1961 model cars in California; the GM smog chamber-the first and largest privatelyowned facility for laboratory simulation of actual smog formationwhich has been operational since 1962 and has contributed greatly to the understanding of photochemical reactions in the atmosphere; and the air injection reactor system, also known as AIR, which was introduced on California cars in 1966 to control exhaust emissions.

GM has shared its findings with others in industry, scientific and engineering societies and any governmental agencies concerned with