Secondary Logo

Journal Logo

Validity of Self-Reported Physical Fitness and Body Mass Index in a Military Population

Martin, Robyn C.; Grier, Tyson; Canham-Chervak, Michelle; Anderson, Morgan K.; Bushman, Timothy T.; DeGroot, David W.; Jones, Bruce H.

Journal of Strength and Conditioning Research: January 2016 - Volume 30 - Issue 1 - p 26–32
doi: 10.1519/JSC.0000000000001026
Original Research
Free

Martin, RC, Grier, T, Canham-Chervak, M, Anderson, MK, Bushman, TT, DeGroot, DW, and Jones, BH. Validity of self-reported physical fitness and body mass index in a military population. J Strength Cond Res 30(1): 26–32, 2016—Many epidemiological studies rely on valid physical fitness data. The purpose of this investigation was to assess the validity of self-reported Army Physical Fitness Test (APFT) data and determine whether men and women recall APFT performance differently. U.S. Army soldiers (N = 1,047) completed a survey, including questions on height, weight, and most recent APFT performance. Height, weight, and APFT performance were also obtained from unit records. The mean ± SDs for unit and self-reported push-up repetitions were 63.5 ± 13.1 and 66.3 ± 14.0 for men and 37.7 ± 12.8 and 40.2 ± 12.8 for women, respectively. The mean ± SD for unit- and self-reported sit-up repetitions were 66.3 ± 11.4 and 68.1 ± 12.1 for men and 64.2 ± 13.6 and 66.5 ± 12.9 for women, respectively. The mean ± SD unit- and self-reported 2-mile run times were 15.2 ± 1.8 and 14.9 ± 1.6 minutes for men, and 18.0 ± 2.9 and 17.4 ± 1.9 minutes for women, respectively. Unit- and self-reported body mass indices (BMIs) (calculated by height and weight) were 26.4 ± 3.4 and 26.3 ± 3.6 for men and 24.6 ± 2.8 and 24.2 ± 3.3 for women. Correlations between unit- and self-reported scores for push-ups, sit-ups, 2-mile run, height, weight, and BMI were 0.82, 0.78, 0.85, 0.87, 0.97, and 0.88 for men and 0.86, 0.84, 0.87, 0.78, 0.98, and 0.78 for women, respectively. On average, men and women slightly overreported performance on the APFT and overestimated height, resulting in underestimated BMI. There was no difference in recall ability between men and women (p > 0.05). The very good to excellent correlations (r = 0.78–0.98) between unit- and self-reported scores indicate that self-reported data are valid for capturing physical fitness performance in this population.

Injury Prevention Program, Army Institute of Public Health, United States Army Public Health Command, Aberdeen Proving Ground, Aberdeen, Maryland

Address correspondence to Tyson Grier, tyson.l.grier.civ@mail.mil.

Back to Top | Article Outline

Introduction

In epidemiological investigations of fitness performance, current and valid physical fitness data are essential. These investigations may be conducted to assess injury risk in specific populations (12,14,15) or to establish fitness and performance standards (7,17). In military populations, fitness data are also necessary for investigations leading to the evaluation and monitoring of existing programs to decrease the incidence of injuries and increase the combat readiness of soldiers (4,11,18).

In the U.S. Army, physical fitness is routinely measured through completion of the Army Physical Fitness Test (APFT). Soldiers are required to take the APFT twice a year (1). The test consists of 3 events: a 2-minute maximal effort push-up event, a 2-minute maximal effort sit-up event, and a 2-mile run performed for time. Height and weight are also recorded at the time of the APFT (1). Unfortunately, accessing recent unit APFT records is not always possible for various reasons, such as troop movement or loss of data. Collection of physical fitness data by survey may be a more feasible method of obtaining data for large-scale epidemiological studies (3,13). If scientists wish to use self-reported physical fitness data from surveys to conduct epidemiological studies, it is necessary to ensure that self-reported data are a valid measure of fitness. The average age of soldiers in the U.S. Army is between 25 and 30 years and physiologically, they are similar to other young physically active populations (25). Self-reported physical fitness data from soldiers would be expected to be similar to self-reports by athletes and fitness program participants who compete in timed or measured events, such as 10 km runs, triathlons, weightlifting, or shot put (9,22,23).

Previous studies in civilian populations examining the validity of self-reported physical activity have shown mixed results. Depending on the type of question asked and the fitness measure assessed, findings have ranged from no or little association (correlation coefficient <0.25) to good associations (0.50–0.75) (8) between self-reported and actual physical activity (5,6,9,10,22,26). Although less frequently studied than self-reported physical activity, previous studies attempting to assess the validity of self-reported physical fitness have shown good to very good correlations (3,16,19,21). One previous assessment of APFT data showed good to excellent correlations (r = 0.71–0.85) between self-reported APFT data and data collected from unit records in basic training (16). However, overall, studies of self-reported physical fitness were limited by small sample sizes, lack of female subjects, or comparison of perceived physical fitness to clinical tests instead of field expedient tests (16,19,21). Studies performed on the validity of self-reported height and weight have shown very good to excellent correlations for subjects under the age of 60 but have not been assessed in the military population (20,23,24).

The purpose of this investigation was to determine the validity of self-reported height, weight, and APFT event performance among Army soldiers in an operational unit. We hypothesized that self-reported height, weight, and APFT scores would be valid indicators of physical fitness, regardless of the time since the soldiers' last APFT. We also hypothesized that men and women would recall height, weight, and APFT performance equally well.

Back to Top | Article Outline

Methods

Experimental Approach to the Problem

As part of a program evaluation, soldiers were asked to respond to a survey containing questions about their most recent APFT performance, height, and weight. Self-reported data were compared with cross-sectional data recorded by the unit during the subject's most recent APFT. The variables assessed were push-up repetitions, sit-up repetitions, 2-mile run time, height, weight, and body mass index (BMI) (as calculated by height and weight).

Back to Top | Article Outline

Subjects

The study population consisted of U.S. Army soldiers in a light infantry brigade who, as part of an ongoing injury prevention program evaluation, completed a voluntary survey containing information about performance on their most recent APFT. The light infantry brigade consisted of infantry, cavalry, field artillery, brigade support, and brigade special troops battalions. Unit rosters were requested and obtained through the brigade medical officer. Project plans were reviewed and approved as public health practice by the Army Public Health Command Public Health Review Board. All subjects signed an informed consent statement before participating in the investigation.

Back to Top | Article Outline

Procedures

A paper survey was administered to obtain information about the soldiers' most recent APFT and personal characteristics, including their current height and weight. The questions asked were as follows: (a) “What was the date of your last Army Physical Fitness Test (APFT) (to the best of your recollection)?” (b) “What were the raw scores on your last Army Physical Fitness Test (APFT)?” (push-up repetitions, sit-up repetitions, 2-mile run time in minutes and seconds, or 2-mile walk time in minutes and seconds) (c) “What is your height?” (inches) and (d) “What is your weight?” (pounds).

Unit APFT records obtained from the unit training (S3) office provided data on soldiers' most recent APFT performance (within the last 12 months), specifically the 2-minute push-up event (repetitions completed), the 2-minute sit-up event (repetitions completed), the 2-mile run event (minutes and seconds), and height (inches) and weight (pounds) at the time of the APFT. Two-mile run time was converted from minutes and seconds to minutes and fractions of a minute (seconds/60). Height was converted from inches to centimeters and weight was converted from pounds to kilograms. Body mass index was calculated as weight in kilograms divided by height in meters squared (kg/m2). Date of birth was also obtained from the Armed Forces Health Surveillance Center and subtracted from the survey date to calculate age.

Back to Top | Article Outline

Statistical Analyses

Validity of self-reported physical fitness was tested by comparing self-reported survey data to unit records and demographic data using Pearson product-moment correlation coefficients. Strength of association was assessed as follows: correlations from 0 to 0.25 indicate little or no relationship; from 0.25 to 0.5 indicate a fair degree of relationship; from 0.5 to 0.75 indicate a moderate to good relationship; and those greater than 0.75 indicate a very good to excellent relationship (8). Paired equivalence t-tests were used to determine whether unit-reported and self-reported data were equivalent. Equivalence intervals were set at ±5% of the unit-reported mean for each category. For men, equivalence intervals were defined as ±3.2 push-ups, ±3.3 sit-ups, ±0.76 minutes (in fractions of a minute for the 2-mile run), ±8.9 cm (height), ±4.1 kg (weight), and ±1.32 BMI units. For women, equivalence intervals were defined as ±1.9 push-ups, ±3.3 sit-ups, ±0.9 minutes (in fractions of a minute for the 2-mile run), ±8.2 cm (height), ±3.3 kg (weight), and ±1.2 BMI units.

The difference between self-reported and unit APFT scores and BMI was calculated to determine how much, on average, soldiers overestimated or underestimated their APFT performance, height, weight, and BMI compared with their unit records (self-reported performance minus unit record performance). Independent sample t-tests were used to test the hypothesis that there was no significant difference between how much men and women overreported or underreported their APFT performance, height, weight, and BMI. Statistical significance levels were based on a 2-tailed p value of less than 0.05.

To determine whether recall accuracy would decline over time, correlations between unit records and self-reported APFT performance, height, weight, and BMI were also calculated by recall time period. Time between unit record and self-report date was calculated as the number of days from the last APFT reported by the unit to the date the survey was completed. Four categories for time since last APFT were examined: less than 30 days, 31–60 days, 61–179 days, and 180–365 days. Mean values and SDs were calculated for age, height, weight, BMI, the time between unit record and self-report date, and APFT push-ups, sit-ups, and 2-mile run. All statistical analyses were performed using The Statistical Package for the Social Sciences (SPSS, IBM Corp., Armonk, NY), version 21.0 and Microsoft Excel 2010 (Microsoft, Redmond, Wash).

Back to Top | Article Outline

Results

Surveys were collected from 68.8% of 5,665 soldiers in the brigade and unit APFT scores were available for 49% of the soldiers in the brigade. Survey and unit APFT data were available for 953 men and 94 women (n = 1,047, 18.5% of soldiers in the brigade). Some APFT data were missing in each category because unit data were not available for all soldiers and some questions were left blank on the survey (approximately 80% of the soldiers answered all of the survey questions). Unit APFT data may have been missing because soldiers can be excused from one or all of the APFT events due to injury or illness on the day of the test. Approximately 6.6% of the survey and 6.1% of the unit-reported 2-mile run times were missing because the soldier was allowed to walk or bike the 2-mile run event due to injury or illness. There is no alternate test available for the push-up and sit-up portions of the APFT so it is not possible to determine whether the missing scores in those portions of the test were not recorded or whether the soldier was excused from the event due to injury, illness, or some other reason. Two female run times were dropped from the analysis due to obvious data entry errors.

Table 1 displays the age, height, weight, and BMI of the soldiers who completed the survey and had unit APFT scores. The average age of participants was 27.2 for men (range, 18–57) and 26.6 for women (range, 19–46), with an average BMI of 26.2 for men and 24.3 for women. The mean time between the APFT recorded in unit records and the self-report (by survey) was 86 days.

Table 1

Table 1

Table 2 displays the unit and self-reported APFT performance, height, weight, and BMI for men. On average, men reported slightly higher performance on all 3 APFT events, and accurately reported their weight but slightly overestimated their height. As a result, BMI was slightly underestimated. However, paired equivalence t-tests showed that unit- and self-reported data were equivalent in all categories (p ≤ 0.05). Correlations between self-reported and unit records ranged from 0.78 to 0.97, with weight showing the strongest correlation (r = 0.97).

Table 2

Table 2

Table 3 shows the unit- and self-reported APFT performance, height, weight, and BMI for women. On average, women also reported slightly higher performance on all 3 APFT events, slightly overestimated height, and underestimated weight, resulting in slight underestimates of BMI. Paired equivalence t-tests showed that unit- and self-reported data were equivalent for all categories (p ≤ 0.05) except in the push-up category among women (p = 0.53), where the self-reported mean push-up value was 6% greater than the mean unit-recorded push-up value. Correlations between self-reported and unit APFT performance, height, weight, and BMI for women, ranged from 0.78 to 0.98, with weight showing the strongest correlation (r = 0.98).

Table 3

Table 3

Table 4 shows the average difference between self-reported and unit APFT records for men and women. No differences in overreporting or underreporting of APFT performance, height, weight, and BMI were found between men and women. Therefore, men and women's recall of their APFT performance was similar.

Table 4

Table 4

Table 5 shows the correlations between unit records and self-reported APFT performance, height, weight, and BMI by the number of days between the soldier's last APFT and survey administration. Because there was little or no difference in how men and women self-reported APFT, height, weight, and BMI, the gender groups were combined for this analysis. Among soldiers who responded to the survey, recall was better for all APFT measures within 60 days of the last APFT. Correlation coefficients for BMI and APFT declined over time. However, even when more than 180 days had passed since the last APFT, correlations between unit records and self-reported APFT performance ranged from 0.70 to 0.85. Correlation coefficients for height and weight slightly increased over time, with the strongest correlations in the 180–365 days since last APFT time frame. This finding is unexpected and cannot be explained by this study.

Table 5

Table 5

Back to Top | Article Outline

Discussion

This investigation sought to determine whether self-reported height, weight, and APFT scores were valid substitutes for measured, recorded physical fitness test (PFT) data within a U.S. military population. Unlike previous investigations, which were limited by smaller sample sizes and few female participants (3,16,19), our study features a larger sample size and a female cohort in proportion to that found in the U.S. Army. In addition, this is the first study to report on the effect of time elapsed since the last APFT on recall ability.

Correlations between self-reported and unit-reported APFT performance, height, weight, and BMI were very good, indicating that U.S. Army soldiers are able to report these measures accurately enough for use in public health/epidemiologic investigations and evaluations. Both men and women slightly overestimated their performance on the push-up, sit-up, and 2-mile run events. Men and women also overestimated their height and women underestimated their weight, resulting in somewhat lower self-reported BMIs for men and women. Despite these slight overestimations of APFT performance and physical fitness, paired equivalence t-tests showed that unit- and self-reported data were equivalent (within 5%) in all categories (p ≤ 0.05) except for the push-up category among women (p = 0.53). There were also no significant differences in accuracy between men and women when reporting APFT performance, height, and weight. Although the strength of association between unit-recorded and self-reported data declined over time, correlations were still good to excellent (r = 0.70–0.99) after 180 days. This suggests that soldiers are able to accurately recall physical fitness performance, height, and weight even when surveys are administered 6 months after their last APFT.

The good to excellent correlations found through our analysis support other research suggesting that surveys can be an accurate tool for estimating physical fitness among the U.S. military population. A study by Riley et al. (21) asked U.S. Marine Corps recruits to answer questions about self-assessed physical fitness and activity, then compared those fitness and exercise parameters to their unit-recorded performance on the Marine Corps PFT-1. A correlation of 0.72 was found between the self-reported performance predictions and recorded number of pull-ups a soldier could perform. Our investigation found a slightly stronger relationship between self-reported and unit-recorded push-ups (0.82). The correlation found in our investigation is likely higher because the Army soldiers in our investigation were asked to recall how well they actually did, not predict how well they thought they could do on the PFT. Despite the difference in study methods, our findings generally support those of Riley et al. (21) and suggest that active duty soldiers in the U.S. military can accurately report their physical fitness.

In a study similar to ours, Jones et al. (16) assessed the validity of self-reported APFT scores among U.S. Army soldiers, but their sample size was much smaller (N = 88) than ours and there were only 5 female subjects. Because of the small sample size, that study was not able to assess differences between male and female subjects. The previous study was also performed among soldiers serving in the Army within a specific military occupational specialty (MOS), whereas volunteers in our investigation were from several different MOSs in an infantry brigade, strengthening the generalizability of our findings. Overall, our results are quite similar to the findings of Jones et al. (16), which showed good to very good correlations between self-reported and unit-recorded push-up repetitions (r = 0.83), sit-up repetitions (r = 0.71), and the 2-mile run times (r = 0.85).

Another investigation of U.S. Army soldiers stationed in Europe or deployed to Kosovo (N = 433) found a lower correlation (r = 0.6) between self-reported and unit-reported APFT scores than the current investigation (3). It is possible that the correlation found in that investigation differs from correlations in the current investigation because Adler and Thomas asked soldiers to report their total adjusted APFT score instead of the number of push-up and sit-up repetitions and 2-mile run time. Soldiers were also asked to report whether they passed or failed that APFT. After the APFT, soldiers can ask or will be shown their raw scores and verify that they have been correctly recorded. Their APFT score may or may not be calculated onsite, depending on the number of soldiers taking the test and graders present to help calculate the final APFT scores. If they did not receive their final APFT score onsite, they can calculate their final score using the APFT standard scoring charts as shown in the Army Physical Training Field Manual 7–22 (2). Typically, soldiers know the approximate number of repetitions and 2-mile run time they need to pass the APFT with their desired adjusted score, but if they pass with raw data that is close to the target goals they have set for themselves, they may not take the time to calculate their official score. Because soldiers may approximate their adjusted scores instead of looking up official scores, it is possible that self-reported APFT performance based on adjusted scores would not be as valid as APFT performance based on raw repetitions and 2-mile run time.

Soldiers are required to take the APFT every 6 months and performance on the test is reported in efficiency reports and is a factor in promotions. As a result, soldiers generally pay considerable attention to their performance on the APFT. This may allow soldiers to recall the number of sit-up and push-up repetitions and run time more accurately than might otherwise be expected of other athletes and fitness program participants. It is also important to consider that correlations may have been impacted by the fact that most soldiers have taken the APFT multiple times. Taking the test multiple times may increase the chance of accurately recalling performance, but it may also increase the chance that the performance values reported on the survey were not actually their most recent values, but values from a previous APFT, or represent how well they typically perform. It is also possible that some soldiers may have reported their adjusted APFT score (1–100 points possible for each event) instead of repetitions and run time on the survey. As with any survey, other soldiers may have simply made up a response to the survey question because they could not remember values from the last test. Many soldiers also did not report scores on every APFT event. This is a limitation, especially if those who did not respond were also those who were more likely to not accurately remember their score.

Another limitation of this investigation is the small sample size of women. However, females represented 9.9% of the sample, and in 2011 only 14.5% of Active Duty Army soldiers were female. With this in mind, our sample does reflect the demographics of the U.S. Army fairly well (25). Despite this, to perform a more accurate assessment of whether men and women self-report APFT scores the same way, it would be ideal to perform the investigation with a larger sample of women.

Back to Top | Article Outline

Practical Applications

To conduct large-scale epidemiological studies on the U.S. military population, it is important for researchers to have an easy to administer, valid means of obtaining physical fitness data. Despite the limitations listed above, our data support the conclusion that U.S. Army soldiers are able to provide acceptably accurate APFT performance, height, and weight data on a survey. Although a small number of self-reported scores varied widely from unit-recorded scores, strong correlations between the 2 suggest that surveys may be a valid and easy to administer means of collecting data for most large-scale epidemiological investigations requiring physical fitness data. Although this study was conducted among members of the U.S. military, these results may also be helpful for conducting epidemiological studies of other physically active populations, such as high school, college, professional, or recreational athletes who may be able to self-report their performance on physical fitness assessment tests.

Back to Top | Article Outline

Acknowledgments

This research was supported in part by an appointment to the Postgraduate Research Participation Program at the U.S. Army Public Health Command (USAPHC) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and USAPHC. This study is approved for public release, distribution is unlimited. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or reflecting the views of the U.S. Army, the Department of Defense, or the National Strength and Conditioning Association.

Back to Top | Article Outline

References

1. Army Regulation 350-1, Army Training and Leader Development. Washington, DC: US Department of the Army, 2009. pp. 1–33.
2. FM 7-22 Army Physical Readiness Training. Washington, DC: Department of the Army, 2012.
3. Adler AB, Thomas JL, Castro CA. Measuring up: Comparing self-reports with unit records for assessing soldier performance. Mil Psychol 17: 3, 2005.
4. Bullock SH, Jones BH, Gilchrist J, Marshall SW. Prevention of physical training-related injuries recommendations for the military and other active populations based on expedited systematic reviews. Am J Prev Med 38: S156–S181, 2010.
5. Chasan-Taber L, Bianca Erickson J, Nasca PC, Chasan-Taber S, Freedson PS. Validity and reproducibility of a physical activity questionnaire in women. Med Sci Sports Exerc 34: 987–992, 2002.
6. Chasan-Taber S, Rimm EB, Stampfer MJ, Spiegelman D, Colditz GA, Giovannucci E, Ascherio A, Willett WC. Reproducibility and validity of a self-administered physical activity questionnaire for male health professionals. Epidemiology 7: 81–86, 1996.
7. Constable S, Palmer B. The Process of Physical Fitness Standards Development. Wright-Patterson Air Force Base, OH: Human Systems Information Analysis Center, 2000.
8. Dawson B, Trapp RG. Basic & Clinical Biostatistics: McGraw-Hill, 2004.
9. Emaus A, Degerstrom J, Wilsgaard T, Hansen BH, Dieli-Conwright CM, Furberg A-S, Pettersen SA, Andersen LB, Eggen AE, Bernstein L, Thune I. Does a variation in self-reported physical activity reflect variation in objectively measured physical activity, resting heart rate, and physical fitness? Results from the Tromso study. Scand J Public Health 38: 105–118, 2010.
10. Friedenreich CM, Courneya KS, Neilson HK, Matthews CE, Willis G, Irwin M, Troiano R, Ballard-Barbash R. Reliability and validity of the past year total physical activity questionnaire. Am J Epidemiol 163: 959–970, 2006.
11. Grier T, Canham-Chervak M, McNulty V, Jones BH. Extreme conditioning programs and injury risk in a US Army Brigade Combat Team. US Army Med Dep J Oct-Dec: 36–47, 2013.
12. Grier TL, Morrison S, Knapik JJ, Canham-Chervak M, Jones BH. Risk factors for injuries in the U.S. Army Ordnance School. Mil Med 176: 1292–1299, 2011.
13. Johns G. How often were you absent? A review of the use of self-reported absence data. J Appl Psychol 79: 574, 1994.
14. Jones BH, Bovee MW, Harris JM III, Cowan DN. Intrinsic risk factors for exercise-related injuries among male and female army trainees. Am J Sports Med 21: 705–710, 1993.
15. Jones BH, Cowan DN, Tomlinson JP, Robinson JR, Polly DW, Frykman PN. Epidemiology of injuries associated with physical training among young men in the army. Med Sci Sports Exerc 25: 197–203, 1993.
16. Jones SB, Knapik JJ, Sharp MA, Darakjy S, Jones BH. The validity of self-reported physical fitness test scores. Mil Med 172: 115–120, 2007.
17. Knapik J, Bandaret L, Bahrke M, O'Connor J, Jones BH, Vogel J. Army Physical Fitness Test (APFT): Normative Data on 6022 Soldiers (Technical Report T94–7). Natick, MA: US Army Research Institute of Environmental Medicine, 1994.
18. Knapik JJ, Canham-Chervak M, Hoedebecke E, Hewitson WC, Hauret K, Held C, Sharp MA. The fitness training unit in U.S. Army basic combat training: Physical fitness, training outcomes, and injuries. Mil Med 166: 356–361, 2001.
19. Knapik JJ, Jones BH, Reynolds KL, Staab JS. Validity of self-assessed physical fitness. Am J Prev Med 8: 367–372, 1992.
20. Kuczmarski MF, Kuczmarski RJ, Najjar M. Effects of age on validity of self-reported height, weight, and body mass index: Findings from the Third National Health and Nutrition Examination Survey, 1988–1994. J Am Diet Assoc 101: 28–34, 2001.
21. Riley DJ, Wingard D, Morton D, Nichols JF, Ji M, Shaffer R, Macera CA. Use of self-assessed fitness and exercise parameters to predict objective fitness. Med Sci Sports Exerc 37: 827–831, 2005.
22. Siconolfi SF, Lasater TM, Snow RCK, Carleton RA. Self-reported physical activity compared with maximal oxygen uptake. Am J Epidemiol 122: 101–105, 1985.
23. Spencer EA, Appleby PN, Davey GK, Key TJ. Validity of self-reported height and weight in 4808 EPIC-Oxford participants. Public Health Nutr 5: 561–565, 2002.
24. Stewart AL. The reliability and validity of self-reported weight and height. J Chronic Dis 35: 295–309, 1982.
25. United States Department of Defense. 2011 Demographics Profile of the Military Community. Washington, DC: Office of the Deputy Under Secretary of Defense (Military Community and Family Policy), 2011.
26. Washburn RA, Goldfield SR, Smith KW, McKinlay JB. The validity of self-reported exercise-induced sweating as a measure of physical activity. Am J Epidemiol 132: 107–113, 1990.
Keywords:

army; self-report; APFT; exercise; survey; recall

Copyright © 2016 by the National Strength & Conditioning Association.