I need your suggestion, I am in mid of my research and I collected data on income and expenditure on different commodities like meat, rice, pulses during holy month of Ramdan and again after Ramdan. Now I am searching for a suitable methodology to determine the effect of Ramdan Fasting on consumption patterns of the 87 households in my sample.
STATISTICAL ANALYSIS
This is a statistical calculation problem involving series of t-tests. The sample size is 87. There is no question raised about the adequacy of the sample size. Assume that there is no problem with sampling methodology or sample size, the following comments ensue:
COMPARISON STUDY
It is necessary to collect consumption data before, during, and after the fasting period. Each period may be marked, thus:
[1] Before Ramdan : a
[2] During Ramdan: b
[3] After Ramdan: c
The query is: whether consumption depends on fasting?
The data labeling is: consumption = dependent variable (Y) and fasting = (X).
There are three pairs to compare, namely:
Pair (ab): Before and during Ramdan
Pair (ac): Before and after Ramdan
Pair (bc): During and after Ramdan
The daily or weekly expenditure may be compared. Use simple d-bar analysis to determine whether there is a significant difference in expenditure in the paired time. The calculation is a straight forward statistical application. Find mean difference:
(1) d^ = (d1 + d2 =…+ dn)/n
Standard deviation follows:
(2) S = sqrt (sum (d-d^)^2/n)
Significance test:
(3) t(d) = d^ / (S/sqrt n)
If using 0.95 confidence interval, assuming infinity n, the t-standard score is 1.6. If your t(observed) is larger than 1.64, it means that there is a significance difference among expenditures in the paired period. If your t(observed) < 1.64, it means that there is no statistical difference among expenditures in the paired period.
The above test only tells whether there are differences among expenditures in the 3 periods. Next, it is necessary to test whether there is an association (relationship) between fasting and consumption pattern.
DATA TYPING & CORRELATION COEFFICIENT TO MEASURE ASSOCIATION
Prove that there is a relationship between fasting and consumption pattern. You might have 3 linear regression equation for the 3 periods:
Y1 = a + bX + c
Y2 = a + bX + c
Y3 = a + bX + c
In each case, prove that “b” does not equal to zero. Any case (period) that shows b = 0, it means that there is no relationship between fasting and consumption (spending pattern). Assume that b > 0 or b 1.64; otherwise, not significant.
SURGE BUYING MAY BE SPECULATED
There might be a surge buy right before the fasting period to stock up on food. Similar pattern may also be speculated right before or immediately after the end of the fasting period. The drastic change may be uncovered through examining the slope of the line in the designated period. It may be wise to re-construct two additional regression run covering said period and compare the slopes of these regression lines to those of the prior 3 lines.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology