Player Height has No Effect on RPG in 2k12 Association Simulation - return.specialist's Blog

return.specialist's Blog

Player Height has No Effect on RPG in 2k12 Association Simulation

Posted on July 3, 2012 at 01:41 AM.

(0)

Preface

Perhaps this has been addressed previously but I haven�t seen it mentioned while browsing through some archival threads. In my own ambivalent opinion, the experiment may have been superfluous but was also necessary. The Abstract provides sufficient summary of the study in whole, especially the underlined portions. Those interested in a more in depth description of what was done, may read the article in its entirety.

ABSTRACT

The present study was designed to examine the effects of height on rebounds per game average for a simulated season in NBA 2K 2012 Association mode. The effect of height on blocks per game average was assessed as well. Dwight Howard was the player of interest as he was the league leader in rebounds for the 2011-12 season; Serge Ibaka was of interest in regards to blocks. For this reason, a roster which had been engineered to realistically replicate statistics was employed and copied five times. One of the rosters was not edited; in the other five rosters, heights of Howard and Ibaka were adjusted to 1'10", 5'2", 6', 7'6", and 8'11". The 2011-12 season was simulated 5 times each for the players' actual height and for the five experimental heights' totaling 30 simulations. Season averages per game were recorded (as well as other data).

Average RPG for 5 simulated were:

6�11�: 14.2

6�: 13.92

5�2�: 14.22

1�10�: 13.88

7�6�: 14.20

8�11�: 13.88

Upon examinations of the averages, intuition would suggest that there is little, if any effect of height on rebounds (or blocks). However, an independent analysis of the average squared difference from the mean (variance) was used to determine whether any statistically significant differences existed between the means of the different heights (ANOVA). The statistics failed to reject the null hypothesis--meaning that the data gathered indicate that player height has little effect on the RPG or BPG stats in simulated seasons of Association mode. More explicitly, the data suggest that a 1'10" Dwight Howard at center would average a nearly identical number of rebounds as the actual 6�11� Howard. While the implications of the lack of height-effect may be vast, criticism will be forgone at present in favor of unbiased reporting.

INTRODUCTION

Complaints about the rebounding system for simulation as well as the in-game user-experience abound (comprehensive stat mechanics guide & wish list thread, page 83; respectively). This author has noted the latter on numerous occasions and found the most frustrating scenario to occur when a user-controlled player jumps for a rebound only to have another player retrieve the ball due to the user being locked in a predetermined animation wherein that player follows a preset path and does not make midair adjustments to play for the ball. Additionally, these animations limit user ability for multiple offensive rebounds by a single player, in a single possession. For an in depth discussion of complaints surrounding the offensive rebound system in particular see this thread. However it should be noted that when considering the scarcity of ≥30-rebound performances in the modern era�since 1973 there have been 28 instances of a player amassing 30+ rebounds in ~46,740 regular season NBA games; which is ~.00006% of games, or 1 for every ~1,670 games played�developers have managed to successfully prevent human domination of this dimension of the game and deliver a fairly realistic enactment, though it may come at the behest of some users.

Hypothesis

The research hypothesis is an attempt to establish that there will be an observed effect of height on average RPG or:

H1: 6�11� ≠ 6�0� ≠ 5�2� ≠ 1�10� ≠ 7�6� ≠ 8�11�

However, in order to empirically validate if an effect is present, the null hypothesis will test that there is no observed effect of height on average RPG or:

H0: 6�11� = 6�0� = 5�2� = 1�10� = 7�6� = 8�11�

METHODS

Roster

Logged onto 2kShare and downloaded iTofu�s stat-roster. This set was selected because league-wide player attribute-ratings were adjusted according to player statistics in the past seasons; thus this roster specifically geared toward accurate simulation and is well suited for the present study. Roster was saved to an empty flash memory stick six times with six different filenames: itofuDHcontrol, itofuDH600. itofuDH52, iofuDH110, itofuD76, and itofuDH811. Exited to main menu.

With the exception of �itofuDHcontrol�, the roster files were edited in REditor II, so that the corresponding number in the file name represented an imperial measure of height which in turn was converted to cm, i.e.:

600 = 6� = 182.88cm,

52 = 5�2� = 160.02cm (shortest height adjustable in game),

110 = 1�10� = 54.6cm (see shortest recorded person in medical history),

76 = 7�6� = 229cm (tallest height adjustable in game),

811 = 8�11� = 272cm (see tallest recorded person in medical history).

Dwight Howard�s height was adjusted in each file in accordance to the above measurements. Serge Ibaka�s height was also adjusted to the same measurements, in each file in order to observe the affect height may have on blocking in simulations. All other attributes and measurements remained adjusted.

SIMULATION

Inserted flash memory back into xBox360 and loaded the control roster. Began a new association. Switched the options for �Trade Deadline�, �CPU Trading�, and �Injuries� to the �off� position. Trade deadline was turned off simply for convenience as doing so would mean one less message while simulating the season. CPU Trading and Injuries were turned off to control for the confounding variables that trades and injuries could inject, ensuring that in each simulation of the 2011-12 season, at each height, Howard and Ibaka would be subject to identical teams, tendencies, and opponents.

Selected �Orlando Magic� as user-controlled team. With the exception of �Player Trading� all �In-Season Task Automation� options were switched to the �on� position for convenience sake. Player Trading automation remained off so trade requests could be denied, ensuring that rosters remained the same through the course of all simulations. Off-Season Tasks were not adjusted because the desired information would be obtained prior to the Playoffs, thus prior to the off-season as well. Association was saved on memory stick as AssDHcontroliT1�representing Association Dwight Howard Control-height, iTofu roster, 1st trial.

Elected to have the system Auto-Generate Rookies. Prior to simulating season, player-rating was recorded to note an if an effect of height evidenced.

Season was then simulated. If a trade offer appeared, �no trade messages� was selected. If a message appeared regarding award points, player of the month, etc., �no more notifications� was selected. If a message appeared about player resigning deadlines or D-League playoffs, �continue simulating� was selected.

Upon completion of simulation, regular season league leader statistics were loaded. These data points were recorded:

Dwight Howard: rebounds per game, minutes per game, defensive player of the year (y or n), game high (rebound)

Serge Ibaka: blocks per game, minutes per game

Exited to home screen and repeated the simulation process four more times, saving each association file with an ascending numeral at the end.

After the fifth trial of control-height, five trials were conducted with the players at each height, resulting in a total of 30 simulated seasons.

RESULTS

Rebounds

Table 1: Rebounds Per Game Averages

height	1	2	3	4	5	mean
6'11"	14.50	14.10	14.60	14.10	13.70	14.20
6'	14.30	13.90	13.40	13.90	14.10	13.92
5'2"	13.60	13.80	15.00	14.60	14.10	14.22
1'10"	13.70	13.80	14.00	13.80	14.10	13.88
7'6"	14.70	13.80	14.00	14.50	14.00	14.20
8'11"	13.90	14.00	13.80	13.90	13.80	13.88

The RPG averages for each individual season and 5-season averages for the different height iterations can be seen above in Table 1. Upon compiling these stats, the data was subject to a single factor ANOVA which can be viewed below in Table 2.

Table 2: ANOVA

Anova: Single Factor

SUMMARY
Groups	Count	Sum	Average	Variance
6'11"	5	71	14.2	0.13
6'	5	69.6	13.92	0.112
5'2"	5	71.1	14.22	0.332
1'10"	5	69.4	13.88	0.027
7'6"	5	71	14.2	0.145
8'11"	5	69.4	13.88	0.007


ANOVA
Source of Variation	SS	df	MS	F	P-value	F crit
Between Groups	0.743	5	0.1486	1.184064	0.346174	2.62065
Within Groups	3.012	24	0.1255

Total	3.755	29

The results did not garner sufficient statistical evidence to indicate that the average RPG was effected by height: f(5, 24) = 1.18, p = 0.35. Thus, the data fail to reject the null hypothesis.

Blocks

Additionally, in regards to blocks, results did not garner sufficient statistical evidence indicating that the average BPG was effected by height: f(5, 24) = 1.47, p = 0.27.

Serge Ibaka�s average BPG for each height: