DOCUMENT RESUME TM 003 978 ED 097 350 Valentine, Lonnie D., Jr.; Cowan, Douglas K. AUTHOR Comparability Study of Armed Services Vocational TITLE Aptitude Battery Scores from Answer Sheet and Answer Card Administration. Final Report. Air Force Human Resources Lab., Lackland AFB, Tex. INSTITUTION Personnel Research Div. AFHRL-TR-73-55 REPORT NO Jan 74 PUB DATE 16p. NOTE MF-$0.75 HC-$1.50 PLUS POSTAGE EDRS PRICE *Answer Sheets; *Comparative Analysis; Mil.Aary DESCRIPTORS Personnel; Occupational Tests; *Response Mode; *scores; Testing; Test Results; Vocational Aptitude *Armed Services Vocational Aptitude Battery IDENTIFIERS ABSTRACT Armed Services Vocational Aptitude Battery (ASVAB) Form 2 was standardized for Digitek answer sheets. In September 1973, the Digitek scoring facility at Randolph AFB will be inadequate to handle anticipated scoring load. Consequently, ASVAB answer forms have been redesigned for processing via a Hewlett-Packard mark-sense reader from IBM card size forms which are more compressed than was the Digitek form. Score comparability from administrations on the two form answer forms was investigated. It was found that the answer change had negligible effect on average examinee performance overall the battery's tests. score distributions, or correlations among Significant Interaction between initial ability (as measured by the Armed Forces Qualification Test) and answer form used was found for two tests (Word Knowledge and Coding Speed); however the interaction did not follow a clearcut pattern. It was recommended that existing ASVAB conversion tables continue in use until completion of a full restandardization study. (Author) AFHRL-TR-73.65 R FORCE El COMPARABILITY STUDY OF ARMED SERVICES VOCATIONAL APTITUDE BATTERY SCORES FROM ANSWER SHEET AND ANSWER CARD H ADMINISTRATION U By M Lonnie D. Valentine, Jr. Douglas K. Cowan A PERSONNEL RESEARCH DIVISION Lackland Air Force Base, Texas 78236 January 1974 Approved for public release; distribution unlimited. u s DEPARTMENT or HEALTH EDUCATION & WELFARE NATIONAL INSTITUTE OP , EDUCATION uf tat f 4 , f1Ar. . ,' I t) t) t) I LP)... v F .F CA'',14P4 P't .4)", t Y r ,)c) N v.. ,:r; t; Put Nr., :.)( 4 r N4 9',. 4N° F ); I r r, .. r)k, p,), u pc, t-P 4, LABORATORY S AIR FORCE SYSTEMS COMMAND BROOKS AIR FORCE BASE,TEXAS 78235 BEST COPY 4161081E NOTICE When US Government drawings, specifications, or other data are used other than definitely purpose Government for related any a operation, the thereby Government procurement incurs no responsibility nor any obligation whatsoever, and the fact that the Government may have formulated, furnished, or in any way supplied the said drawings, specifications, or other data is not to be regarded by implication or otherwise, as in any manner licensing the holder or any other person or corporation, or conveying any rights or permission to manufacture, use, or sell any patented invention that may in any way be related thereto. This final report was submitted by Personnel Research Division, Air Force Human Resources Laboratory, Lack land Air Force Base, Texas 78236, under project 7719, with the Hq Air Force Human Resources Laboratory, Brooks Air Force Base, Texas 78235. This report has been reviewed and cleared for open publication and/or public release by the appropriate Office of Information (01) in accordance with AFR 190.17 and DoDD 5230.9. There is no objection to unlimited distribution of this report to the public at large, or by DDC to the National Technical Information Service (NTIS). This technical report has been reviewed and is approved. LELAND D. BROKAW, Chief Personnel Research Division HAROLD E. FISCHER, Colonel, USAF Commander Unclassified SECURITY :LASSIFICATION OF THIS PAGE (When Rine Frnered) READ INSTRUCTIONS REPORT DOCUMENTATION PAGE BEFORE OWL FTING FORM RECIPIENT''., cAT aLor, NUMBER 2. GOVT ACCESSION NO. REHOR ' NUMBER 3 1 AFFIRIAR-73.55 TYPE OF REPORT 6 PERIOD COVERED 5 TITLE !old Sublett,' 4 COMPARABILITY STUDY OF ARMED SERVICES VOCATIONAL Final AVM DE BATTERY SCORES FROM ANSWER SHEET AND ANSWI.R CARD ADMINISTRATION 6. PERFORMING ORG. REPORT NUMBER CONTRACT OR GRANT NUMBER(5) AUTHOfq / 6 7. Lonnie D.'Valentine, Jr. Douglas K. Co,,'in 10. PROGRAM ELEMENT. PROJECT. TASK RGANIZATION NAME AND ADDRESS PES FnrIMIN 9 AREA 6 WORK UNIT NUMBERS Personnel Research Division 77191002 Air Force Human Resources Laboratory Lackland Air Force Base, Texas 78236 12. REPORT DATE 11. CONTROLLING OFFICE NAME AND ADDRESS January 1974 Fig Air Force Human Resources Laboratory Brooks Air Force Base, Texas 78235 NUMBER OF PAGES 13 14 SECURITY CLASS. (of this report) 14. MONITORING AGENCY NAME Pt ADORESS(if different from Controlling Office) 15 Unclassified DECLASSIFICATION DOWNGRADING 15a SCHEDULE STATEMENT /of this Report) 16 DISTRIBUTION Approved for public release; distribution unlimited. DISTRIBUTION STATEMENT (of the almtract entered in Block 20, if different from Report) 17. Ie. SUPPLEMENTARY NOTES 0 19. KEY 140 IDS (Continue on reverse side if nece.,saty and Identify by block number) Armed E entices Vocational Aptitude Battery answer t mnat perform tnce comparability high school testing . A (Continue on teversu side If ne essary and identify by block number) ABSTRACT 20. Armed Services Vocational Aptitude Battery (ASVAB) Form 2 was standardized for Digitek answer sheets. In Septetnuer 1973, the Digitek scoring facility at Randolph APB will be inadequate to handle anticipated scoring load. Consequently, ASVAB answer forms have been redesigned for processing via a HewlettPackard marksatise reader from IBM card size forms which are more compressed than was the Digitek form. Score comparability from administrations on the two answer forms was investigated. It was found that the answer form change had negligble effect on average examinee performance overall score distributions, or correlations among the battery's tests. Significant interaction between initial ability (as measured by the AFQT) and answer form used was found for two tests Word Knowled e and Codin 1 Steed . hoWever the " DO " EDITION OF 1 Nov 66 IS OBSOLETE Unclassified 1473 1 JAN 7 1 beta entered) SECURITY CLASSIFICATION OF THIS PAGE (When Unclassified S5GURITY CLASSIFICATION OF THIS PAGE(When Data Enter itd) Item 20 Continued. interaction did not follow a clearcut pattern. It was recommended that existing ASVAI3 conversion tables continue in use until completion of a full restandardization study. Unclassified SECURITY CLASSIFICATION OF THIS PAGE(Wheft Data Entered) PREFACE This report describes effect of answer form change (from a sheet to IBM size cards) on performance on the Armed Services Vocational Aptitude Battery. Work was accomplished under Project 7719, Ai; Force Personnel System Development on Retention, Promotion, and Control, Quality Evaluation, Selection, Assignment, Utilization; Task 771910, Armed Forces Operational Selection Tests. TABLE OF CONTENTS Page 5 Introduction I. 5 Procedure H. 5 Results III. 6 Conclusions and Recommendations IV. 14 Refemces '444.04, LIST OF TABLES Page Table 7 Sample Distribution by AFQT Decile 1 7 Distribution of Educational Level 2 Number of Cases Reporting Completion of High School Mathematics 3 7 and Science Courses 7 AFQT Mental Category Distribution of Samples 4 8 Distribution of AQE Scores 5 9 Intercorrelations Among Twenty Variables 6 10 ASVAB Means and Standard Deviations for Samples 2XC and 2X 7 Analysis of Variance Summary 11 8 12 ASVAB Means and Standard Deviations with AFQT Deci les 9 Raw Score Distributions on Coding Speed 13 10 14 Raw Score Distributions for ASVAB Tests 2 Through 9 11 COMPARABILITY STUDY OF ARMED SERVICES VOCATIONAL. APTITUDE BATTERY SCORES FROM ANSWER SHEET AND ANSWER CARL) ADMINISTRATION Test (AFQT) on Armed Forces Qualification I. INTRODUCTION scores contained in their records. Matching was as many percentile completed for The Armed Services Aptitude scores Vocational as available from within the total number of airmen Battery (SVAB), described by Vito la and Alley Distribution of AFQT deciles for the tested. (1968) an ..1 Bayroff and Fuchs (1970), is currently for testing of high school seniors in samples is given in Table 1. operation; 1 voluntarily participating high schools throughout Testing was accomplished under normal testing the United States. The current form of ASVAB room conditions. ASVAB-2XC and 2X were use of a DIGITEK answer sheet, requires the administered alternately during morning and after- scored on a DIGITEK DM100 optical scanner noon testing sessions; each form was administered located at Randolph Air Force Base, Texas. half-day to an equal number of times during each preclude time-of-day effects. Scoring was accom- With only one form of the ASVAB available, Randolph AFB on the DIGITEK plished at test security requirements dictated that alternate DM-100 for Form 2X and on the Hewlitt-Pachrd forms of the test be produced. Ceinsequently, mark sense reader for Form 2XC. ASVAB-2 and -3 were developed and were normed using a geographically representative sample tested Two samples were processed separately. Means, specified Armed Forces Entrance and at standard deviations, and Pearson Product Moment Examining Stations (AFEES). Beginning in for matrices were computed int ercorrelation September 1973, ASVAB-2 will be used for high twenty variables (these included years of educa- school testing, with ASVAB-3 reserved as the joint tion, AFQT, four composites from the Airman services qualifying entrance test. Qualifying Examination, data on completion of specific high school courses, and all nine tests of Form-2 of the ASVAB was standardized on the the ASVAB). In addition, each of the two samples DIGITEK answer sheets, but with the beginning of was divided into eight subsamples based on AFQT high school testing during the fall of 1973, the decile, and means and standard deviations of the DIGITEK DM-100 scoring capability at Randolph ASVAB tests were computed for each resulting AFB will he inadequate to handle the anticipated subsample. Raw score distributions on the nine load. ASVAB tests were prepared for each of the two A revised answer form, printed on a three part samples. IBM card. was designed to replace the answer For each ASVAB test, a two-way classification sheet, will, a resultant reduction in the amount of analysis of variance was computed to assess effects space allowed examinees for answer marking. The of format and of the format x AFQT decile inter- revised form was designed for scoring via the action on scores. In this regard, it is noted that a Hewlett-Packard model WAOO /D mark sense for AFQT decile would be effect significant reader also at Randolph AFB. expected in all cases as a funrtion of correlation Test norms, as between AFQT and the test. A prat iinary study, to compare the two test if in- appropriate previously established, are response 1:cording methods, was conducted at consequential effects are found both for format Lackland ifB. Texas, to determine whether there interaction of AFQT decile with for the and was test ne rin slippage due to answer form change. format; a significant F ratio on either of these effects, however, would suggest normative adjust- ments as a consequence of format change. II. PROCIDURI. Two matched samples. each consisting of 248 III. RESULTS male USAF basic airmen assigned to Lackland AFB for processing and training, were admin- Although the answer sample (2X) sheet istered Fo..in 2 of the ASVAB. One sample was contains fewer cases with education beyond high administered Form 2XC (answer card format) school than does the answer card sample (2XC), while the ether was tested on ASVAB Form 2X the average educational levels are comparable. (answer slit et format). The samples were matched 5 BEST COPY AVAILABLE BEST COPY AVAILABLE Answer sheet sample airmen (2X) indicated significant at the .01 level for word knowledge and completion for more high school math and science at the .05 level for coding speed. Interestingly; the courses than did answer card sample cases (2XC) interaction largest effect would have been even though sample 2X(' had more airmen expected on coding speed since is a closely it completing 12 or more years of education (Tables timed test of simple clerical speed and accuracy: 2 and 31. consequently, a change in size of the response form had greater potential for affecting The two samples were matched on AFQT score; performance of "high intelligence" and "low each case in sample 2X was matched with a sample intelligence" subjects differentially. An inspection 2XC case with the same AFQT score. AFQT scores of the decileby-decile group means for the two appear to be fairly evenly distributed across the format groups (Table 9) reveals that in AFQT deciles 20 through 90; there were no scores below decile 20.29, mean coding speed performance of 20. Actual score range was 21 through 98. Enlist- the answer sheet group is 6.16 raw score points is not authorized for mental category V ment higher than that of the answer card group, and ( AFQT score 0 - 9) applicants, and recent policy in decile 50.59, a difference of about the that has also excluded those scoring below 21 on .same magnitude occurs, but in favor of the answer AFQT. The percentage of airmen in the samples' card group. Other decile group means for coding AFQT categories I through IV appear to be quite speed are approximately equal. In the case of the similar to the distribution for recent first term word knowledge test, format group means differ in USAF airman accessions. (Unpublished data. See favor of the card format group by a little more Table 4 for AFQT category distributions.) than three raw score points in the AFQT decile Further evidence of comparability of the two 50.59 and by a little over two raw score'points in samples was found in the distribution of AQE favor of the answer sheet format in the AFQT percentile While statistically scores. significant decile 60-69. Other decile group means are approx differences were not found between the two imately equal. groups on any of the four AQE aptitude These two significant interaction effects would composites, there were small mean differences in suggest possible differential effect of test response favor of the answer sheet sample; the largest of format change on test performance as a function these were on the Administrative (2.27) and of level of AFQT performance. In both format Electronics (1.77) Ms. Distributions, means, and samples, correlation between AFQT and coding standard deviations of the AQE composites are speed is quite modest (low .30s), and correlation shown in Table 5. between AFQT and word knowledge in both Minor differences.of no significance were noted samples is in the low .50s. AFQT has served as the among the intercorrelatiuns for the two samples. standard against which ASVAB norms are This serves as further evidence that format change calibrated, but with modest relatively these did not materially affect the battery (Table 6). is possible that part of the within correlations, it AFQT decile difference washes out in conversions Table 7 shows, for both samples, means and utilizing the full samples (as a function of AFQT for each ASVAB test. deviations standard In distri!itional decile addition, no overlap). this However, significant ASVAB test mean cannot be eltarly determined front these data differences were found; the largest difference (.62) because of the truncation of the samples on AFQT is on space perception. Thus, it can be seen that the bottom 20 percent of (approximately the the format change makes no difference in average normative reference population is nut represented performance. here). Outcomes of the nine two-way classification Raw score distribution for both groups on all analysis of variance problems are summarized in nine ASVAB tests are presented in Tables 10 and Table 8. As expected, highly significant F ratios It the ASVAB test 11. noteworthy that is were obtained cases for AFQT deck, all in distributions of the two format samples differ very reflecting correlation between the test and AFQT. little from each other. In no instance was there a statistically significant answer format main effect; this finding is the quite small overall mean with consistent IV. CONCLUSIONS AND RI.COMMI.NDATIONS differences between the two sanyies (largest difference was .62 on space percept ion). Signifi- The results of this.study suggest compaable cant format x ANT decile effects were found for ASVAB score results using either the answer card two of the nine interaction was This tests. 6 or the answer sheet. Presently estab- Because of AFQT truncation in the data for I )1G1.1 lished A:NABForm 2 norms may he satisfactory this study, determination of whether final a tier answ, card administration. Although a signifi conversions for the ASVAB tests are affected by cant format by Al:QT decile interaction was found answer form format change must await completion on coding speed and word knowledge. overall of a full standardization study. Data for such a means. s...indard deviations, and intercorrelations study have been collectedand are being analyzed. for the two format conditions are nearly the same. is reconitnended that present conversions for It This suggests that, despite the significant inter- ASVAB Form 2 continue in use until completion action or; these two tests. conversion tables may of this standardization study. be affected very little if at all. Table 1. Sample Distribution Table 3. Number of Cases Reporting by AFQT Decile Completion of High School Mathematics and Science Courses Sample* Total AFQT Cecile 2X Percent 2XC Sample* 2X 2 XC Course 70 90.99 35 35 14.11 28 11.29 28 80,89. 56 154 Algebra 162 60 30 30 70-79 12.10 Geometry 95 106 60.69 27 54 27 10.89 Trigonometry 40 29 66 33 33 50-59 13.31 63 Chemistry 76 4049 66 33 33 13.31 Physics 34 42 30.39 74 37 37 14.91 25 25 50 20.29 10.08 'chi square tests revealed no significant differences 496 248 248 100.00 Total between the two groups. The two samples are evenly distributed across the range of AFQT scores. No score lower than 21 nor higher The two than 98 was reported. samples arc pre. Table 4. AFQT Mental Category cisley matched on AFQT scores. Distribution of Samples Distribution of Educational Table 2. Sample* Level Mental 2X Category Total 2XC Percent* Sample* Ed ucat Ion Total (years) 2 KC 2 X (93.100) 46 23 9.27 23 II 8 3 5 9 (65.92) 80 9 4 80 32.26 160 13 42 10 23 III 19 28 37 65 (31.64) 11 46.77 116 232 116 147 292 12 145 IV 20 41 13 21 (10.30) 29 29 58 11.69 14 10 25 15 Total 496 248 248 99.99 6 15 3 3 6 16 7 1 *Percentage of Basic Airmen in each mental category 248 496 Tot al 248 is similar to the percentage in each category into enlisted USAF during recent years (July 69 through September 11.96 11.76 11.86 Mean 71). S.D. 1.25 1.16 1.21 'A t t . of difference between the sample means revealed no agnificant difference (t = 1.83). 7
Description: