Chapter II- Literature Review
3.5 Sampling Design
The sampling process of this study is based on the following steps:
1. Defining population
This study’s primary objects are online gamers, and the population of the study is then defined as the Taipei’s graduate students who have some experience in playing F2P MMORPGs. Since various studies provide very different findings on the age profile of the gamers, an estimation from the past studies will be used, and players ages 17~23 will be the core mass of players, while stretching the figure to approximately age 44 to cover about 90% of players (GameSpot Audience Profile Study, June 2008; IDC, 2008).
2. Confirming the sampling framework
The sampling framework uses members of Taipei’s various graduate students to participate in the questionnaire survey. The questionnaire surveys are sent out via email to the students currently participating in graduate studies and the link to the questionnaire survey (which is produced by an online survey maker- my3Q) is disclosed within the mails for the subjects to respond to the survey. Using this kind of sampling method means the samples are not completely randomly selected, which also means the representativeness of the sample might be biased. However, since the purpose of the research is to study the gamers’ behavior and how perceived attractiveness of game design element on part of the gamers will affect their game loyalty level, selecting the samples from those who have actually played online games is still appropriate (since they must have experienced it themselves to give more accurate opinions) and should have a above average level of representativeness.
3. Choosing the sampling method
Due to the size of the population, and also limited by time, human & financial resources, this study will use systematic sampling method. Since the questionnaire survey is done on partly on the internet and partly by face-to-face, part of the entered answers are automatically stored and filed, this automation process not only reduces the potential human errors, but can also reduce cost and time of the research. In addition, by performing a portion of the survey online removes the need for excessive printing
hard copies of the questionnaires, which not only saves cost, but saves the environment at the same time.
4. Deciding on sample size
The decision on the sampling size is based on the principles Roscoe (1975) proposed:
1). A reasonable sampling size should be between 30 ~ 500 samples to be appropriate.
2). When samples are divided into sub-sample groups, each sub-sample groups need to have at least 30 samples.
3). In performing a multivariate study, number of samples need to be 10 times or more the size of research variables to be appropriate.
This study uses systematic sampling method to focus the study group to those who have actually played a F2P MMORPG before, and then uses random sampling ratio technique to estimate the minimum sample size for the study. This method randomly selects n samples from the N parent population. The chance of occurrence of error which the variance between ratio of the sample with a particular characteristic and the parent ratio p cannot be over an acceptable error limit d must not be smaller than 1 - α.
And this can be represented by the equation as follows:
pˆ
pˆ p d
1
P ··········· (1)
Where: = sampling ratio, p = parent population ratio, d = level of acceptance of error, and 1 - α = degree of reliability.
pˆ
Then the size of the sample is big enough, the allocation of p can use normal distribution to come close to the actual value. Thus, using the above equation (1), we can get the minimum sample size, as shown in the following equation:
Where: n = sample size, N = parent population
When the parent population is very large, (2) can be simplified as:
)
Since p is unknown, the p(1- p) in the above equation acts as an increasing function, and since 0 therefore getting the maximum value of p(1- p) of ¼. Hence our equation will now look like this:
As described above, with the level of acceptance of error is d = 0.05 and with level of significance of α = 0.05 (95% level of confidence), our sample size n = 384.16.
Therefore, 385 will be the number of effective questionnaires retrieved for the samples to be representative.
5. Selecting sampling units
The questionnaire utilizes the online survey maker my3Q as the basis to produce the questionnaire survey. With a time frame of three weeks (from August 20th to September 10th of 2009 to distribute the surveys by means of e-mailing and link posting to various graduate students who have at least some experience in playing F2P MMORPGs in Taipei with help of classmates and friends. Also from September 15th to 30th of 2009, hard copies of the questionnaire survey were handed out to students.
6. Collecting Sample data
Since part of the survey questionnaire will be completed by the participants online via e-mails and links provided to the my3Q website, the sample data collection process is simplified since the input data are already stored on the website automatically as soon as they complete them
7. Assessing the sampling result:
The number of questionnaires retrieved is 406 copies. After the initial selection process, 12 copies are deemed to be invalid or incomplete, making the total number of valid questionnaires to 394 copies. The following provides an insight on the statistical distribution of the collected samples.
Gender
Frequency Percent Valid Percent
Cumulative Percent
Male 299 75.9 75.9 75.9
Female 95 24.1 24.1 100.0
Valid
Total 394 100.0 100.0
Age
Frequency Percent Valid Percent
Cumulative Percent
17~23 211 53.6 53.6 53.6
24~30 150 38.1 38.1 91.6
31~37 26 6.6 6.6 98.2
38~44 5 1.3 1.3 99.5
Above 45 2 .5 .5 100.0
Valid
Total 394 100.0 100.0
Monthly Income ($NT)
Frequency Percent Valid Percent
Cumulative Percent
Less than $10,000 283 71.8 71.8 71.8
$10,001~$20,000 96 24.4 24.4 96.2
$20,001~$30,000 13 3.3 3.3 99.5
$30,001~$40,000 2 .5 .5 100.0
Valid
Total 394 100.0 100.0