BAB 1

ACKNOWLEDGE

A. Background
In the case where the linked value is numerical, then the correlation analysis is
an option. However, if the two connected links are categorical, then the use of
correlation analysis can no longer be used because the numbers in a category are just
codes rather than true values so arithmetic operations are not valid for the case of
categorical data .Another reason why correlation analysis cannot be used in
categorical data because one of the categorical variables is a nominal that can’t be
sorted by category. Giving a different sequence will obviously give different
correlation values so that two people who calculate the value of a large correlation are
likely to give unequal results. For this reason, the chi square analysis will be used to
determine whether there are associations and differences between the categorical
variables.
Some statistical formulas are based on certain assumptions. The formula can
describe a phenomenon when these assumptions are met. Therefore , if we use the
formula then the expected data in accordance with the assumption of a research
formula. In this regard, this paper can be used as a reference to improve the
understanding of chi square or kai square and test the prerequisite of good and correct
analysis in a study.
Chi-Square is also called Kai Squares. Chi Square is one type of
nonparametric comparative test performed on two variables, where the data scale of
the two variables is nominal. (If from 2 variables, there is 1 variable with nominal
scale then chi square test with reference that must be used test at the lowest degree).
The chi-square test is the most widely used non parametric test. But please
note the requirements of this test is: the frequency of respondents or samples used
large, because there are several conditions in which chi square can be used are:
1. There is no cell with a reality frequency value or also called Actual Count
(F0) of 0 (Zero).
2. If the contingency table form is 2 X 2, then there should be no 1 cell which
has expected frequency or also called the expected count ("Fh") less than 5.
3. If the table form is more than 2 x 2, ie 2 x 3, then the number of cells with
expected frequency less than 5 can’t be more than 20%.
 BAB II
DICUSSION

The chi-square test is a test the hypothesis of the comparison between the
actual sample frequency (hereinafter referred to as the observation frequency,
denoted by fo) with the expected frequency based on a particular hypothesis in each
case or data (hereinafter referred to as the expected frequency, denoted by fe).

A. THE GOODNESS OF FIT TEST

The goodness of fit test, the null hypothesis is a provision of the expected pattern of
frequencies in certain categories (s). The pattern expected should be in accordance
with the assumption or assumption of the possibility of the same event and general
nature.

Note:
fo: observation frequency
fe: frequency of expectation
In the degree equivalence model, the degree of freedom (df) matches the
number of categories minus the number of estimators based on the sample and is
subtracted 1. The parameter estimator is an estimated parameter of value, since the
parameter value can not be precisely determined based on the available sample data.
If it is formulated to be:
df = k - m -1 with:
k: number of sample data categories
m: number of parameter values that are estimated
If the null hypothesis states that the observed frequencies are distributed equally to
the expected frequency, there is no parameter of the estimator. Thus the value m=0.
Example :
Sebuah distibutor alat penggilingan padi membagi pasar menjadi 4 wilayah (A, B, C,
dan D). Ada informasi bahwa pendistribusian alat penggilingan merata pada setiap
wilayah. Untuk membuktkan pernyataan tersebut diambil 40 arsip sebagai sampel.
Dari 40 arsip tersebut diperoleh informasi yang tertuang dapa tabel. Gunakan tingkat
signifikansi 5 persen untuk menguji hipotesis yang menyatakan bahwa distribusi alat
penggilingan di keempat wilayah merata (sama)!

Answer :
1. Hipotesis
Ho : distribusi alat penggilingan di keempat
wilayah merata (sama)
Ha : distribusi alat penggilingan di keempat wilayah tidak merata (tidak sama)
2. Nilai Kritis
Dalam kasus di atas tidak perlu ada parameter yang diestimasi. oleh karena
itu:
df = k – m – 1 = 4 – 0 – 1 = 3
x2(0,05;3) = 7,81
3. Nilai Hitung
Nilai uji statistik x2hitung diperoleh dengan cara sebagai berikut :

4. Simpulan
Karena nilai statistik x2hitung = 4,0 lebih kecil daripada nilai tabel x2(0,05;3) = 7,81
berarti kita tidak dapat menolak Ho menyatakan bahwa distribusi alat penggilingan di
keempat wilayah merata (sama)
 B. TEST CONTINGENCY TABLE
Contingency tables contain data obtained from simple random samples and
organized by rows and columns. Both rows and columns are each divided into criteria
or provisions. The data values in the contingency table are the observed frequencies
(fo).
With contingency table test we can test whether two variables (rows and
columns) are independent or not. This idea is based on the assumption that if the
categories are independent of each other, the value of the observed frequency is close
to the expected frequency value. The big differences will support us to reject the
hypothesis of independence.

Feij = (∑foi - ∑foj) / n

dengan derajat kebebasan
 df = (r – 1) (k – 1)
2
 Sedangkan rumus untuk memperoleh nilai x

Example :
Tabel berikut menunjukkan pengunjung pada salon TAMPAN pada tanggal
12 Oktober 2009 yang dikategorikan berdasarkan jenis kelamin dan umur. Ujilah
hipotesis bahwa jenis kelamin dan umur pengunjung adalah independen dengan
tingkat signifikansi α =0,01

Answer :
 Hipotesis
Ho : jenis kelamin dan umur pengunjung adalah independen
Ha : jenis kelamin dan umur pengunjung adalah tidak independen
 Nilai Kritis
Derajat kebebasan df:
df = (r – 1) (k – 1) = (2 – 1) (2 – 1) = 1
Nilai uji statistik x2(0,01;1) = 6,63
Kita menolak Ho jika x2hitung > 6,63
 Nilai Hitung
Berikut ini contoh perhitungan nilai frekuensi harapan.Berikut ini contoh perhitungan
nilai frekuensi harapan :

 E11 = (130 x 110 )/190 = 75,26

 E12 = (60 x 110)/190 = 34,74
 E21 = (130 x 80)/190 = 54,74
 E22 = (60 x 180) / 190 = 25,26

Nilai statistik x2 =

Kesimpulan :
Dengan tingkat signifikansi 1 persen Ho ditolak karena nilai statistik x2
sampel =23,28 lebih besar daripada x2(0,01;1) = 6,63. Ini berarti bahwa jenis kelamin
dan umur pengunjung tidak independen.
 BAB III
CONCLUSION

Chi square can be calculated using category data. The data is discrete or
nominal. The result of chi square is always positive because the result is always
squared. The result of chi squared count if smaller than chi squared table then the
result is no significant difference. But if the result is larger than chi square table then
there is a significant difference.
So it can be concluded the greater the result of chi squared count against chi
squared table hence the more significant difference. Test requirements analysis
include test of normality, homogeneity, linearity, heterokedasitas, multikolinieritas
and autokorelasi . The test is very important to do, because in the inferential statistics
for the conclusion of data analysis apply to the population, then the distribution of
data must meet some of these criteria.